Resolving the Collatz Conjecture: A Rigorous Proof through Inverse Discrete Dynamical Systems and Algebraic Inverse Trees

1. Introduction

Discrete dynamical systems have been a fundamental object of study in mathematics for centuries, with applications spanning fields as diverse as physics, biology, computer science, and social sciences. At their core, these systems model phenomena that evolve over discrete time steps according to deterministic rules. Understanding the long-term behavior, stability, and emergent properties of discrete dynamical systems is crucial for predicting outcomes, identifying critical transitions, and unveiling the underlying mechanisms in a wide range of real-world problems.

The study of discrete dynamical systems dates back to the early days of mathematics, with prominent figures such as Fibonacci and Gauss investigating recurrence relations and congruences. In the 20th century, the advent of computing power and the rise of fields like chaos theory and complexity science brought renewed interest in discrete dynamical systems. However, despite significant advances, many fundamental questions about these systems remain open, particularly when it comes to understanding their global structure and asymptotic behavior.

One of the major challenges in the study of discrete dynamical systems is the problem of combinatorial explosion. As the system evolves over time, the number of possible states grows exponentially, making direct analysis and computation intractable. Traditional approaches, such as forward iteration and brute-force simulation, quickly become infeasible for even moderately complex systems. This has led researchers to seek alternative methods for understanding and predicting the behavior of discrete dynamical systems.

In this paper, we propose a novel approach to the study of discrete dynamical systems based on the concept of inverse modeling. Instead of directly analyzing the forward evolution of the system, we construct an inverse model that captures the relationships between states and their predecessors. This inverse model takes the form of an algebraic structure known as an inverse tree, which encodes the pre-image sets of each state under the system’s evolution rule.

The main objectives of this work are twofold. First, we aim to develop a rigorous mathematical framework for inverse modeling of discrete dynamical systems, establishing the theoretical foundations and key properties of inverse trees. Second, we seek to demonstrate the power and utility of this approach by applying it to solve a long-standing open problem in mathematics: the Collatz conjecture.

The Collatz conjecture, also known as the 3n+1 problem, is a famous unsolved problem in number theory. It states that for any positive integer n, the sequence obtained by iterating the function $f\left(n\right)=n/2$ if n is even, and $f\left(n\right)=3n+1$ if n is odd, will eventually reach the number 1, regardless of the starting value. Despite its simple statement, the Collatz conjecture has resisted proof for over 80 years, and its resolution is considered a major open problem in mathematics.

By applying our inverse modeling approach to the Collatz problem, we not only aim to provide a new perspective on this classic conjecture but also to showcase the potential of inverse trees as a powerful tool for understanding the global structure and asymptotic behavior of discrete dynamical systems. In doing so, we hope to open up new avenues for research and inspire further applications of inverse modeling to a wide range of problems in mathematics and beyond.

Note 1. 

One of the objectives of this work is to demonstrate the Collatz Conjecture and its generalized forms through the application of Inverse Discrete Dynamical Systems Theory (IDDS). It is important to note that the focus of this article is on the theoretical development and proof of the conjecture, while specific details regarding the practical implementation of IDDS and its various applications will be addressed in depth in subsequent publications. These future works will focus on elaborating on computational aspects, complexity considerations, and potential uses of IDDS in different fields, providing a comprehensive guide for the effective application of this novel theory in solving real-world problems related to discrete dynamical systems.

2. Definitions and Preliminary Concepts

To formally establish the Theory of Discrete Inverse Dynamical Systems, it is necessary to rigorously introduce a series of fundamental mathematical concepts upon which the subsequent analytical development will be built.

Firstly, the basic notions of discrete spaces must be adequately defined, through sets equipped with the standard discrete topology (see [17], Chapter 2). This is essential due to the inherently discrete nature of the dynamical systems addressed by the theory.

Definition 2.1. 

Metric Space: Let X be a non-empty set. A function $d:X\times X\to \mathbb{R}$ is called a metric on X if it satisfies:

• $d(x,y)\ge 0$, $\forall x,y\in X$ (Non-negativity)
• $d(x,y)=0$ if and only if $x=y$, $\forall x,y\in X$ (Discernibility)
• $d(x,y)=d(y,x)$, $\forall x,y\in X$ (Symmetry)
• $d(x,z)\le d(x,y)+d(y,z)$, $\forall x,y,z\in X$ (Triangle Inequality)

Then, the ordered pair $(X,d)$ is called a metric space .

Definition 2.2. 

Discrete System: Let $(X,d)$ be a metric space. We say that $(X,d)$ is a discrete system if:

• X is countable (finite or countably infinite)
• d is a discrete metric, i.e., the triangle inequality holds with equality:

  $$\forall x,y,z\in X,d(x,z)=d(x,y)+d(y,z)$$

Definition 2.3. 

Continuous System: Let $(X,d)$ be a metric space. We say that $(X,d)$ is a continuous system if:

• X is uncountable (uncountably infinite)
• d is a continuous metric, i.e., the triangle inequality is strict:

  $$\forall x,y\in X,\exists z\in X such that d(x,z)<d(x,y)+d(y,z)$$

Definition 2.4. 

(Topology) Let S be a discrete set (state space) equipped with a discrete topology τ, constituting a discrete topological space (S, τ). Formally:

$\exists \tau$: (S, τ) is a discrete topological space.

Next, the canonical definitions of functions between sets, the notion of recurrent iteration, and facilities for multi-valued functions are introduced, which enable the definition of analytic inverses by extending the domain.

Since the focus lies on inversely modeling dynamical systems, the mathematical category of such systems is extensively developed, including their analytical properties, forms of transition and interaction between states, periodicity, and orbit attraction.

Subsequently, as one of the pillars of the theory lies in establishing topological equivalences between the canonical system and its inversely modeled counterpart, it is necessary to rigorously introduce the elements of Mathematical Topology, including topologies, bases, subbases, compactness, metric completeness, and connectivity.

Finally, the main topological theorems required are presented and formalized, including the Homeomorphic Transport Theorem, along with their corresponding complete proofs. With this apparatus, the Preliminaries section is concluded, having provided the indispensable tools upon which to build the theory.

Definition 2.5 

(Topology). Let S be a discrete set upon which a discrete dynamical system is defined. A topology τ on S consists of a family of subsets of S, called open sets, which satisfy:

$\varnothing ,S\in \tau$ Every union of open sets is open. Every finite intersection of open sets is open. Then the ordered pair $(S,\tau )$ constitutes a discrete topological space.

Definition 2.6 

(Topological Compatibility). Let $(S,\tau )$ be a discrete topological space and $A,B\subseteq S$. We say that τ satisfies the compatibility property if:

$\forall A,B\left[\right(A\in \tau \wedge B\in \tau )\to (A\cap B)\in \tau ]$

That is, the intersection of two open sets is open.

Definition 2.7 

(Compactness). Let $(S,\tau )$ be a discrete topological space. We say that S is compact if:

$\forall U_{\alpha }\alpha \in A[(U_{\alpha }\in \tau \wedge {\bigcup }_{\alpha \in A}{U}_{\alpha }=S)\to \exists A^{\prime }\subseteq A,|A^{\prime }|<{\aleph }_0\wedge {\bigcup }_{\alpha \in A^{\prime }}{U}_{\alpha }=S]$

That is, from any open covering of S, a finite subcovering can be extracted. Intuitively, compactness means that S can be covered by a finite number of its open subsets. The definition states that given any possible infinite open cover $\left\{U_{\alpha }\right\}$ of S, we can always extract a finite sub-collection of sets from $\left\{U_{\alpha }\right\}$ that also covers S.

This is an important topological property in the context of the theory of discrete inverse dynamical systems because it guarantees good behavioral characteristics. Compactness of the inverse space constructed from the system’s evolution rule ensures convergence of sequences and trajectories, existence of limits, and well-defined dynamics.

Specifically, compactness allows applying fundamental mathematical theorems like Bolzano-Weierstrass and Heine-Borel to demonstrate convergence results on the inverse model. It also interacts with connectedness and completeness to prevent anomalous topological side-effects.

Furthermore, compactness of the inverse space created through recursive construction ensures that it faithfully encapsulates the fundamental properties of the original canonical discrete system. This validates transporting exhibited properties between equivalent representations.

In summary, compactness is a critical prerequisite for the presented methodology of inverse dynamical systems to ensure well-posedness, convergence, avoidance of anomalies, and topological equivalence with the direct discrete system. Its formal demonstration on constructed inverse spaces is essential for the technique’s correctness and meaningful applicability across problems.

Definition 2.8 

(Connectedness). Let $(S,\tau )$ be a discrete topological space. We say that S is connected if:

$\neg \exists A,B\subseteq S[A\ne \varnothing \wedge B\ne \varnothing \wedge A\cap B=\varnothing \wedge A\cup B=S\wedge A,B$ closed]

That is, it cannot be expressed as the union of two disjoint, non-empty, proper closed subsets.

Definition 2.9 

(Topological Equivalence). Let $(X,\tau )$ and $(Y,\sigma )$ be discrete topological spaces. A topological equivalence between $(X,\tau )$ and $(Y,\sigma )$ is a bijective and bicontinuous homeomorphic correspondence $f:(X,\tau )\to (Y,\sigma )$ that preserves the cardinal topological properties between both discrete spaces.

Definition 2.10 

(State Space). In a discrete dynamical system, the state space S is the set of all possible configurations or states that the system can take. Each element $s\in S$ represents a unique state of the system at a given moment. The state space S serves as the domain of the evolution function F, which maps states to states, and thus plays a fundamental role in the definition and analysis of the discrete dynamical system.

Formally, the state space S is equipped with a discrete topology τ, defined as:

$$\tau =\{U\subseteq S:U=\varnothing \mathrm{or}\forall s\in U,\{s\}\in \tau \}$$

In other words, τ is the collection of all subsets of S, including the empty set and all singleton sets. The pair $(S,\tau )$ forms a discrete topological space, where every subset of S is both open and closed.

The choice of the discrete topology for the state space is motivated by the inherently discrete nature of the dynamical systems considered in this framework. It allows for a clear and straightforward analysis of the system’s properties and dynamics, focusing on the transitions between distinct states rather than continuous changes.

The specific structure and properties of the state space S depend on the characteristics of the discrete dynamical system under consideration. For example:

• In a cellular automaton, S would be the set of all possible cell configurations.
• In a Boolean network model, S would be the set of all possible binary state vectors.
• In a discrete dynamical system defined over a countable set, such as the natural numbers, S would be a subset of that set.

Definition 2.11 

(Discrete Dynamical System). A discrete dynamical system is an ordered pair $(S,F)$ such that:

• S is a discrete set (state space) equipped with a discrete topology τ, constituting a discrete topological space $(S,\tau )$. Formally:

  $$\exists \tau :(S,\tau )\mathrm{is}\mathrm{a}\mathit{discrete}\mathit{topological}\mathit{space}$$
• $F:S\to S$ is a function (evolution rule) that maps states in S to S, recursively and deterministically over S. Formally:

  -

  F preserves the discreteness of elements in S:

  $$\forall x,y\in S:x\ne y\Rightarrow F\left(x\right)\ne F\left(y\right)$$

  -

  F is deterministic over S:

  $$\forall x\in S,\exists !F^n\left(x\right),\forall n\in \mathbb{N}$$

  -

  F is recursive: successive iteration $F^n\left(x\right)$.

  -

  F preserves the topology τ of S:

  $$F^{-1}\left(V\right)\mathit{is}\mathit{open}\Rightarrow F\left(U\right)\subseteq V,\mathit{for}\mathit{open}\mathit{sets}U,V\subseteq S$$

Where $F^n\left(x\right)$ denotes the n-th iterate of F applied to the state $x\in S$.

Examples of discrete dynamical systems include:

• Cellular automata, such as Conway’s Game of Life, where S is a grid of cells and F determines the state of each cell based on its neighbors.
• Iterative maps, like the Logistic Map, where S is a subset of real numbers and $F\left(x\right)=rx(1-x)$ for some parameter r.

Example of a simple SIR model:

$$\begin{array}{cc}\hfill S(t+1)& =S\left(t\right)-\beta S\left(t\right)I\left(t\right)\hfill \end{array}$$

(1)

$$\begin{array}{cc}\hfill I(t+1)& =I\left(t\right)+\beta S\left(t\right)I\left(t\right)-\gamma I\left(t\right)\hfill \end{array}$$

(2)

$$\begin{array}{cc}\hfill R(t+1)& =R\left(t\right)+\gamma I\left(t\right)\hfill \end{array}$$

(3)

Definition 2.12 

(Orbit in DIDS). Let $F:S\to S$ be a discrete dynamical system defined on a state space S, where F represents the evolution rule mapping the state space to itself. For any initial state $x_0\in S$, the orbit of $x_0$ under F is the sequence ${\left\{x_n\right\}}_{n=0}^{\infty }$ defined recursively by $x_{n+1}=F\left(x_n\right)$ for $n\ge 0$. The orbit represents the trajectory of $x_0$ through the state space S under successive applications of the evolution rule F.

Definition 2.13. 

Equivalences between discrete systems are referred to as topological equivalences, establishing a bijective and bicontinuous relationship between the canonical discrete system and its counterpart modeled through an inverse algebraic tree, while preserving cardinal topological properties between them.

Let $(S,\tau )$ be a discrete topological space. A homeomorphic correspondence is a bijective and bicontinuous function $f:(S,\tau )\to (S^{\prime },{\tau }^{\prime })$ that establishes a topological equivalence between discrete spaces.

Definition 2.14. 

Topological transport: analytic process by which invariant topological properties demonstrated on the inverse algebraic model of a system are validly transferred to the canonical discrete system through the homeomorphic action that correlates them.

Definition 2.15 

(Discrete Topology). Let S be a set. A discrete topology τ on S is defined as:

$$\begin{array}{c}\hfill \tau =\{U\subseteq S:U=\varnothing \vee (\forall x\in U,\left\{x\right\}\in \tau \left)\right\}\end{array}$$

In other words, τ is the set of all subsets U of S such that U is the empty set or for each element x in U, the singleton set $\left\{x\right\}$ belongs to τ.

Furthermore, τ satisfies the following axioms:

• $\varnothing ,S\in \tau$
• $\forall \mathcal{F}\subseteq \tau :\bigcup \mathcal{F}\in \tau$ (Closure under arbitrary unions)
• $\forall \mathcal{F}\subseteq \tau ,\left|\mathcal{F}\right|<\infty :\bigcap \mathcal{F}\in \tau$ (Closure under finite intersections)

Then, $(S,\tau )$ constitutes a discrete topological space.

Definition 2.16 

(Power Set). Given a set S, the power set of S, denoted as $\mathcal{P}\left(S\right)$, is defined as:

$$\begin{array}{c}\hfill \mathcal{P}\left(S\right)=\{U:U\subseteq S\}\end{array}$$

In other words, $\mathcal{P}\left(S\right)$ is the set of all subsets of S, including the empty set ∅ and S itself.

Formally, we can express this using first-order logic as:

$$\begin{array}{c}\hfill \forall U(U\in \mathcal{P}(S)\leftrightarrow U\subseteq S)\end{array}$$

which means that for every set U, U belongs to the power set $\mathcal{P}\left(S\right)$ if and only if U is a subset of S.

Definition 2.17 

(Discrete Space). Let S be a set equipped with a discrete topology τ. Then the ordered pair $(S,\tau )$ constitutes a discrete space.

Definition 2.18 

(Discrete Function). Let $f:S\to S^{\prime }$ be a function between discrete spaces. We say that f is a discrete function if it preserves the discreteness of elements in its image when $S^{\prime }$ is a discrete space. That is, for all $x,y\in S$ such that $x\ne y$, it holds that $f\left(x\right)\ne f\left(y\right)$.

Definition 2.19 

(Categories of DDS). Let $\left(X\right)$ be a discrete topological space and $(F:X\to X)$ an evolution rule in $\left(X\right)$. We define the following categories of discrete dynamical systems (DDS):

• According to the cardinality of $\left(X\right)$:

  -

  Finite: $\left(\right|X|<{\aleph }_0)$

  -

  Countable: $\left(\right|X|={\aleph }_0)$

  -

  Continuous: $\left(\right|X|=2^{{\aleph }_0})$

• According to the recursiveness of $\left(F\right)$:

  -

  Recursive: $(\exists F^{-1}:F^{-1}\left(F\left(x\right)\right)=x)$

  -

  Non-recursive: Does not satisfy the above

• According to sensitivity to initial conditions:

  -

  Non-sensitive: $(\exists \delta >0:d(x,y)<\delta \Rightarrow d(F^n\left(x\right),F^n\left(y\right))\le M)$

  -

  Sensitive: Does not satisfy the above

• According to the degree of combinatorial explosiveness:

  -

  Limited: $\left(\right|F^{-n}\left(x\right)|=O\left(p\left(n\right)\right))$

  -

  Unbounded: $\left(\right|F^{-n}\left(x\right)|\gg p\left(n\right);\forall p\left(n\right))$

where $\left(p\right(n\left)\right)$ is a polynomial.

Theorem 2.1 

(Conditions for Topo-Invariant Transport). Let $(X,F)$ be a DDS and P a topo-invariant property. If:

(1)

F is recursive over X

(2)

The combinatorial explosiveness of F is bounded

(3)

P is demonstrated in the inverse algebraic model of $(X,F)$

Then P is invariably preserved in $(X,F)$ by topological transport.

Proof. 

Let $(X,F)$ be a discrete dynamical system and P a topologically invariant property. Suppose the following conditions hold:

(1)

$\forall x\in X,\exists !F^{-1}\left(x\right)\wedge F^{-1}\left(F\left(x\right)\right)=x$ (Recursivity of F)

(2)

$\exists p\left(n\right)\in \mathbb{N}\left[x\right]:\forall x\in X,|F^{-n}\left(x\right)|=\mathcal{O}\left(p\left(n\right)\right)$ (Bounded Combinatorial Explosiveness)

(3)

$P\left(T\right)$, where T is the inverse algebraic model of $(X,F)$ (Proof of P in the inverse model)

We want to prove that $P\left(X\right)$, i.e., that the property P holds in the original system $(X,F)$.

Let $h:T\to X$ be the homeomorphism that correlates the nodes of the algebraic inverse tree T with the states of the canonical system X. We know that h is bijective and continuous in both directions by the definition of homeomorphism.

Since $P\left(T\right)$ by hypothesis and P is a topologically invariant property under homeomorphisms, we have:

$$\begin{array}{cc}\hfill P\left(T\right)& \Rightarrow P\left(h\right(T\left)\right)\left(\mathrm{By}\mathrm{invariance}\mathrm{of}P\mathrm{under}\mathrm{homeomorphisms}\right)\hfill \\ \hfill & \Rightarrow P\left(X\right)\left(\mathrm{Sin}\mathrm{ce}h\left(T\right)=X\mathrm{by}\mathrm{the}\mathrm{bijectivity}\mathrm{of}h\right)\hfill \end{array}$$

Therefore, we have demonstrated that the topological property P exhibited in the inverse model T is transferred invariably to the original system $(X,F)$ through the homeomorphism h, under the conditions of recursivity of F and bounded combinatorial explosiveness.    □

Theorem 2.2. 

Let $(S,\tau ,F)$ be a discrete dynamical system. Then, given an initial condition $x\in S$ and a sequence $F^{\left(k\right)}\left(x\right)$ obtained by iterating the evolution rule F starting from x, it holds that:

$$\forall x\in S,\forall k\in \mathbb{N},\exists !F^{\left(k\right)}\left(x\right)$$

In other words, starting from any initial state x, F always generates a unique trajectory $F^{\left(k\right)}\left(x\right)$ under iteration.

Proof. 

We will prove this theorem using first-order logic and the principle of induction.

Base case: For $k=1$, we have:

$$\forall x\in S,\exists !F^{\left(1\right)}\left(x\right)\equiv \forall x\in S,\exists !F\left(x\right)$$

This is true by the definition of a discrete dynamical system, as F is a function from S to itself.

Inductive step: Assume that the statement holds for some $k\in \mathbb{N}$, i.e.:

$$\forall x\in S,\exists !F^{\left(k\right)}\left(x\right)$$

We want to prove that it also holds for $k+1$:

$$\forall x\in S,\exists !F^{(k+1)}\left(x\right)$$

Let $x\in S$ be arbitrary. By the inductive hypothesis, there exists a unique $F^{\left(k\right)}\left(x\right)$. Let’s call this unique state y, so $y=F^{\left(k\right)}\left(x\right)$.

Now, since $y\in S$ and F is a function from S to itself, there exists a unique $F\left(y\right)$. But $F\left(y\right)=F\left(F^{\left(k\right)}\left(x\right)\right)=F^{(k+1)}\left(x\right)$.

Therefore, for any $x\in S$, there exists a unique $F^{(k+1)}\left(x\right)$, which is what we wanted to prove.

Conclusion: By the principle of induction, we have shown that:

$$\forall x\in S,\forall k\in \mathbb{N},\exists !F^{\left(k\right)}\left(x\right)$$

   □

Definition 2.20 

(Power Set). Given a set S, the power set of S, denoted as $P\left(S\right)$, is the collection of all subsets of S, including the empty set ∅ and S itself. Formally:

$$P\left(S\right)=\{A:A\subseteq S\}$$

This definition establishes the power set $P\left(S\right)$ as the family of all possible subsets of S. In other words, each element of $P\left(S\right)$ is itself a subset of S. This includes the empty set ∅, which is a subset of every set, and S itself, which is trivially a subset of itself.

Some key points about the power set:

• If S is a finite set with $\left|S\right|=n$ elements, then $P\left(S\right)$ will contain $2^n$ elements. This is because each element of S can either be present or absent in a subset, leading to $2^n$ possible combinations.
• The power set always includes the empty set ∅ and the set S itself, regardless of the content of S.
• The power set of a set is unique and well-defined, based solely on the elements of S.

Definition 2.21. 

Analytic Inverse Function Let $(S,F)$ be a discrete dynamical system, where $F:S\to S$ is the evolution function defined on the discrete space S. The analytic inverse $G:S\to P\left(S\right)$ of F is defined as the function that recursively undoes the steps of F.

Formally, G satisfies:

(1) 

Domain(G) = Range(F)

(2) 

Range(G) = Domain(F)

(3) 

G analytically undoes F: $\forall x\in S:x\in G\left(F\right(x\left)\right)$

Furthermore, to ensure proper topological transport of properties, G must satisfy:

• Injectivity: $\forall x,y\in S,G\left(x\right)=G\left(y\right)\Rightarrow x=y$
• Surjectivity: $\forall z\in S,\exists x\in S:G\left(x\right)=z$
• Exhaustiveness: Recursion through G reaches all states in S.

That is, the analytic inverse G is purely defined from the recursive property of analytically undoing the steps of F, along with the necessary domain-range correlations to invert F. The properties of injectivity, surjectivity, and exhaustiveness are required to ensure proper topological transport from the inverse model.

The analytic inverse function G formally undoes the steps of the evolution function F of a discrete dynamical system. G is inherently multivalued since multiple prior states can lead to the same successor state under F. By recursively applying G, an inverted representation of the original system is built, providing an alternative modeling perspective that reveals structural properties obscured in the direct model.

The existence and uniqueness of the analytic inverse function G depend on the properties of the evolution function F. If F is bijective, then G is guaranteed to exist and be unique.

Property 1 

(Recursive Inverse Function). Let $(S,F)$ be a discrete dynamical system, where $F:S\to S$ is the evolution function. Let $G:S\to P\left(S\right)$ be the analytical inverse function of F, recursively undoing its steps. Then:

Proof 

Let $x\in S$ be an arbitrary state. By definition of G as the analytic inverse function, we have:

$$G\left(F\right(x\left)\right)=x,\forall x\in S$$

Applying F on both sides:

$$F\left(G\right(F\left(x\right)\left)\right)=F\left(x\right)$$

Since F is injective:

$$G\left(F\right(x\left)\right)=x$$

Therefore, G recursively undoes the steps of F. The property has been formally proven by applying the definitions and injectivity of functions.    □

2.1. Combinatorial Complexity and Inverse Model Constructibility

Definition 2.22 

(Moderate Combinatorial Explosion). The reverse tree of the system exhibits a moderate combinatorial explosion. Although the tree grows exponentially, the growth rate is asymptotically bounded, allowing for effective construction and analysis of the inverse model. Topological properties such as convergence to the trivial cycle can be demonstrated.

Let $(S,F)$ be a discrete dynamical system with an evolution function $F:S\to S$ defined on the discrete space S. Let $G:S\to \mathcal{P}\left(S\right)$ be the inverse analytic function of F that recursively undoes its steps, generating the inverse algebraic tree $T=(V,E)$.

We say that $(S,F)$ exhibits a moderate combinatorial explosion if the following conditions are met:

(1) 

Growth rate bound: There exists a function $f:\mathbb{N}\to \mathbb{N}$ such that for any initial state $s\in S$, the number of reachable states after n recursive applications of G is bounded by $f\left(n\right)$, i.e., $|G^n\left(s\right)|\le f\left(n\right)$ for all $n\in \mathbb{N}$, and f is asymptotically less than an exponential function, i.e., $f\left(n\right)=o\left(k^n\right)$ for all $k>1$.

(2) 

Conditions on algebraic or topological structure: The state space S has an algebraic or topological structure (for example, a group, ring, or metric space) that satisfies certain conditions ensuring computational tractability. These conditions may include:

• The composition operation in S is computable in polynomial time.
• S has a finite or efficiently computable representation.
• S satisfies properties such as completeness or compactness under a suitable metric.

(3) 

Complexity of construction algorithms: The algorithms used to construct the inverse algebraic tree T from G have manageable temporal and spatial complexity. Formally:

• The time required to compute $G\left(s\right)$ for any state $s\in S$ is polynomial in the size of the representation of s.
• The depth of the tree T (i.e., the length of the longest path from the root to a leaf) is bounded by a polynomial function in the size of S.
• The maximum degree of any node in T (i.e., the maximum number of children of a node) is bounded by a constant.

If these conditions are met, we say that $(S,F)$ exhibits a moderate combinatorial explosion, implying that the construction and analysis of the inverse algebraic model are computationally tractable.

3. Axiomatic Foundations of DIDS

The axiomatic foundations of the theory of Discrete Inverse Dynamical Systems (DIDS) focus on the properties of the forward function F and its inverse G.

Definition 3.1. 

A discrete dynamical system $(S,F)$ is a DIDS if and only if $F:S\to S$ is a deterministic and surjective function.

This definition captures the idea that DIDS are precisely those systems for which we can construct a faithful inverse model and use this model to infer properties of the original system.

Theorem 3.1. 

If $(S,F)$ is a DIDS, then there exists an inverse function $G:S\to \mathcal{P}\left(S\right)$ that is injective, surjective, and exhaustive.

Proof. 

Let $F:S\to S$ be a deterministic and surjective function. We define $G:S\to \mathcal{P}\left(S\right)$ as follows:

$$G\left(s\right)=\{t\in S:F(t)=s\}$$

We will show that G is injective, surjective, and exhaustive.

(1)

G is injective: If $G\left(a\right)=G\left(b\right)$, then for each $s\in G\left(a\right)$, there exists a $t\in a$ such that $F\left(t\right)=s$, and for each $s\in G\left(b\right)$, there exists a $t\in b$ such that $F\left(t\right)=s$. Since F is deterministic, this t is unique. Since $G\left(a\right)=G\left(b\right)$, these t must be the same for a and b. Therefore, $a=b$.

(2)

G is surjective: For each $B\in \mathcal{P}\left(S\right)$, let $A=\{t\in S:F(t)\in B\}$. Since F is surjective, for each $s\in B$, there exists a $t\in A$ such that $F\left(t\right)=s$. Therefore, $G\left(A\right)=B$.

(3)

G is exhaustive: Since F is surjective, for each $s\in S$, there exists a $t\in S$ such that $F\left(t\right)=s$. Therefore, $s\in G\left(t\right)$. Since this is true for all $s\in S$, the union of $G\left(t\right)$ for all $t\in S$ is equal to S.

Therefore, G is injective, surjective, and exhaustive.    □

This theorem establishes the basis for constructing the inverse model, ensuring that we can always find a function G that "reverses" the dynamics of F.

Theorem 3.2. 

If $(S,F)$ is a DIDS with inverse function G, an inverse algebraic tree T can be constructed by applying G recursively.

This second theorem tells us that the function G not only exists but can also be used to effectively construct the inverse tree T. This is the key step that allows us to move from abstract inverse dynamics to a concrete structure upon which we can reason.

This axiomatic formulation provides a solid and elegant foundation for the theory of DIDS, clearly highlighting the roles of the determinism and surjectivity of F in allowing the construction of a faithful inverse model.

4. Inverse Modeling of Systems

Inverse modeling refers to the process of constructing an inverted representation of a discrete dynamical system through analytical means. Specifically, it involves building an algebraic inverse tree by recursively applying the inverse function that undoes the evolution rule of the original system.

Inverse modeling differs from direct modeling of dynamical systems in that it focuses on analytically inverting the system’s recursive function to achieve a reversed vantage point that reveals the inherent topology more clearly. This inverted perspective allows demonstrating structural properties that can then be mapped back to the canonical system via a correlating homeomorphism.

Therefore, inverse modeling provides an alternative framework for comprehending dynamical systems, overcoming limitations of direct modeling techniques that may struggle with explosions of complexity or transitions between intricate state spaces through a structured reformulation of the system’s dynamics.

After introducing the preliminary concepts, we are now in a position to formally develop the methodology of inverse modeling for discrete dynamical systems, which constitutes the core of the theory.

Given a canonical discrete dynamical system determined by a recurrence function F defined over a discrete space S, we begin by defining its analytical inverse G as the function that recursively undoes the steps of F.

Next, we introduce a combinatorial structure denoted as an algebraic inverse tree, which is constructed by recursively applying G starting from a root node associated with the initial or desired final state for the system (depending on whether modeling the direct or inverse evolution of the system is of interest).

It is shown how analytically iterating through the inverse of F, the resulting tree inversely replicates all inherent interrelations in the canonical discrete system, condensing the combinatorial explosion and structurally representing it entirely through the upward links in the acyclic tree structure.

Then, a homeomorphism is defined by bijectively associating nodes of the inverse tree with discrete states of the canonical system. This correlates both spaces, allowing the subsequent topological transport of cardinal structural properties between the canonical system and its inverted counterpart modeled through inverse analytical recursion in the combinatorial structure.

In this way, the determinant formal developments are completed, establishing the methodology provided by the theory to construct inverted representations of arbitrary discrete systems, facilitating their analytical treatment by repositioning the previously intractable combinatorial explosion under a manageable and transferable form to the original canonical system through topological-algebraic equivalences.

Definition 4.1 

(Discrete Topological Space). Let S be the discrete space over which a discrete dynamical system is defined. The discrete topology on S is defined as:

$\tau =\{\varnothing ,\left\{x_1\right\},\left\{x_2\right\},\cdots \}$

where $x_i\in S$ and each element of S defines an open and closed set (a singleton).

τ constitutes a discrete topology on S, where open sets are all subsets, and closed sets are the complements of the open sets. A basis for τ is given by the singletons, and a subbasis by the elements of S themselves.

Then $(S,\tau )$ is said to be the relevant discrete topological space for the system.

Definition 4.2 

(Discrete Function). Let $f:S\to S^{\prime }$ be a function between discrete spaces. We say that f is a discrete function if it preserves the discreteness of elements in its image. That is, $\forall x,y\in S$ such that $x\ne y$, it holds that $f\left(x\right)\ne f\left(y\right)$.

Definition 4.3 

(Discrete Dynamical System). Let S be a discrete set (state space) equipped with a discrete topology τ, forming a discrete topological space $(S,\tau )$. Let $F:S\to S$ be a function (evolution rule) that maps states in S to S, recursively and deterministically over S.

Formally, a Discrete Dynamical System (DDS) is an ordered pair $(S,F)$ such that:

• S is a discrete set with discrete topology τ, making $(S,\tau )$ a discrete topological space.
• $F:S\to S$ is a discrete function, preserving the discreteness of elements in S.
• F is deterministic over S: $\forall x\in S,\exists !F^n\left(x\right),\forall n\in \mathbb{N}$
• F is recursive: successive iteration $F^n\left(x\right)$.
• F preserves the topology τ of S: $F^{-1}\left(V\right)$ is open $\Rightarrow F\left(U\right)\subseteq V$, with $U,V\subseteq S$ open sets.

Where $F^n\left(x\right)$ denotes the n-th iteration of F applied to the state $x\in S$.

Definition 4.4 

(Inverse Function). Let $(S,F)$ be a DIDS, with $F:S\to S$ the deterministic and surjective evolution function defined over the discrete space S. The inverse function $G:S\to \mathcal{P}\left(S\right)$ of F is defined as:

$$G\left(s\right)=\{t\in S:F(t)=s\}$$

That is, for each $s\in S$, $G\left(s\right)$ is the set of all elements in S that map to s under F.

Furthermore, G satisfies the following properties:

• Injectivity: $\forall a,b\in S,G\left(a\right)=G\left(b\right)\Rightarrow a=b$
• Surjectivity: $\forall B\in \mathcal{P}\left(S\right),\exists A\in S:G\left(A\right)=B$
• Exhaustiveness: ${\bigcup }_{s\in S}G\left(s\right)=S$

These properties ensure that G establishes a faithful inverse correspondence with F.

Definition 4.5 

(Algebraic Inverse Tree). Let $(S,\tau ,F)$ be a DDS with analytic inverse G. The algebraic inverse tree (AIT) $T=(V,E)$ is constructed recursively:

• V is the set of nodes
• $E\subseteq V\times V$ is the set of edges
• $r\in V$ is the root node
• $\forall (u,v)\in E:v\in G\left(u\right)$

Definition 4.6 

(Metric on Algebraic Inverse Tree). Let $T=(V,E)$ be an Algebraic Inverse Tree (AIT). We define the metric $d:V\times V\to \mathbb{R}$ as follows:

$$\begin{array}{c}\hfill \forall a,b\in V:d(a,b)=\left\{\begin{array}{cc}0\hfill & \mathit{if}a=b\hfill \\ min\{n\ge 1:\exists (v_0,v_1,\dots ,v_n)\in V^{n+1},\hfill \\ (v_i,v_{i+1})\in E;\forall i\in \{0,\dots ,n-1\},\hfill \\ v_0=a,v_n=b\}\hfill & \mathit{if}a\ne b\hfill \end{array}\right.\end{array}$$

In other words, $d(a,b)$ is the length of the shortest path from a to b in T.

It is important to note that the spaces considered in this article, particularly the Algebraic Inverse Trees (AITs), are not only topological spaces but also metric spaces. The metric structure, induced by the path length metric d defined above, is compatible with the topological structure and provides additional information about the distance between nodes in the AIT.

The use of distances in this context does not conflict with the topological perspective and does not affect the validity of the analysis. In fact, the metric structure enhances our understanding of the relationships between states in the discrete dynamical system and their corresponding nodes in the inverse model.

Moreover, many of the key results in this article, such as the completeness and compactness of the AIT (Lemmas 5.1 and 5.7), rely on the metric structure. The metric also plays a crucial role in the definition and analysis of convergence properties, such as the convergence of infinite paths to the root node (Lemma 5.9).

Therefore, the use of distances in the context of AITs is justified and does not introduce any inconsistencies or issues in the analysis. The topological and metric perspectives complement each other, providing a richer and more comprehensive framework for studying discrete dynamical systems and their inverse algebraic models.

Theorem 4.1 

(Properties of AITs). Let $T=(V,E)$ be an Algebraic Inverse Tree (AIT) constructed from a Discrete Dynamical System $(S,\tau ,F)$ with the analytic inverse function G. Then:

(1) 

T has no non-trivial cycles, except for the trivial cycle containing the point of contact $pc$.

(2) 

All paths in T converge to the root node r, except for paths ending at $pc$.

Proof. 

We prove each property separately:

Property 1: Absence of Non-Trivial Cycles (Except for the Trivial Cycle Containing $pc$)

Step 1: Define the notion of a non-trivial cycle.

$$\begin{array}{cc}\hfill \forall v_1,\dots ,v_k\in V:\mathrm{NTC}(v_1,\dots ,v_k)\iff & (k\ge 3)\wedge (v_1=v_k)\wedge \hspace{0.17em}\hfill \\ \hfill & (\forall i\in \{1,\dots ,k-1\}:(v_i,v_{i+1})\in E)\hfill \end{array}$$

Step 2: Prove that any non-trivial cycle must include the point of contact $pc$.

$$\begin{array}{c}\hfill \forall v_1,\dots ,v_k\in V:(\mathrm{NTC}(v_1,\dots ,v_k)\wedge pc\notin \{v_1,\dots ,v_k\})\Rightarrow \perp \end{array}$$

Proof: Assume, for contradiction, that there exists a non-trivial cycle $v_1,\dots ,v_k$ such that $pc\notin \{v_1,\dots ,v_k\}$. By the definition of G being multivalued injective for all points except $pc$, we have:

$$\begin{array}{c}\hfill \forall i,j\in \{1,\dots ,k\},i\ne j:G\left(v_i\right)\cap G\left(v_j\right)=\varnothing \end{array}$$

However, since $v_1=v_k$ and $v_2\in G\left(v_1\right)$, we have $v_2\in G\left(v_k\right)$, contradicting the multivalued injectivity of G for points other than $pc$. Therefore, any non-trivial cycle must include $pc$.

Step 3: Prove that the trivial cycle containing $pc$ is the only possible non-trivial cycle in T.

$$\begin{array}{c}\hfill \forall v_1,\dots ,v_k\in V:(\mathrm{NTC}(v_1,\dots ,v_k)\wedge pc\in \{v_1,\dots ,v_k\})\Rightarrow \hspace{0.17em}\\ \hfill (\exists i\in \{1,\dots ,k\}:v_i=pc\wedge \forall j\in \{1,\dots ,k\},j\ne i:v_j\in G\left(pc\right))\end{array}$$

Proof: Let $v_1,\dots ,v_k$ be a non-trivial cycle that includes $pc$. By Step 2, we know that $pc\in \{v_1,\dots ,v_k\}$. Without loss of generality, let $v_1=pc$. Since G is multivalued injective for all points except $pc$, each $v_i$ for $i\in \{2,\dots ,k\}$ must be in $G\left(pc\right)$. Therefore, the only possible non-trivial cycle containing $pc$ is the trivial cycle $(pc,v_2,\dots ,v_k)$, where $v_2,\dots ,v_k\in G\left(pc\right)$.

Property 2: Convergence of Paths to Root Node (Except for Paths Ending at $pc$)

Step 1: Define the convergence of a path to a node.

$$\begin{array}{c}\hfill \forall P\subseteq V,\forall v\in V:\mathrm{Converges}(P,v)\iff \hspace{0.17em}\\ \hfill (\forall \epsilon >0,\exists N\in \mathbb{N},\forall n>N,\forall w\in P:d(w,v)<\epsilon )\end{array}$$

where d is the graph distance in T.

Step 2: Prove that every node in T has a unique path to the root node r or to the point of contact $pc$.

$$\begin{array}{c}\hfill \forall v\in V,\exists !P\subseteq V:(v\in P)\wedge \left(\mathrm{Converges}\right(P,r)\vee \mathrm{Converges}(P,pc\left)\right)\end{array}$$

Proof: By the recursive construction of T using the injective function G, each node has a unique parent, except for the root node r and the point of contact $pc$. Therefore, for any node $v\in V$, there exists a unique path from v to either r or $pc$, which is obtained by following the parent nodes until reaching r or $pc$.

Step 3: Prove that if a path P does not contain $pc$, then it converges to the root node r.

$$\begin{array}{c}\hfill \forall P\subseteq V:(pc\notin P\Rightarrow \mathrm{Converges}(P,r\left)\right)\end{array}$$

Proof: Let P be a path in T that does not contain $pc$. By Step 2, there exists a unique path from any node in P to either r or $pc$. Since $pc\notin P$, the path must converge to r.

Step 4: Prove that if a path P contains $pc$, then it converges to $pc$.

$$\begin{array}{c}\hfill \forall P\subseteq V:(pc\in P\Rightarrow \mathrm{Converges}(P,pc\left)\right)\end{array}$$

Proof: Let P be a path in T that contains $pc$. By Step 2, there exists a unique path from any node in P to either r or $pc$. Since $pc\in P$, the path must converge to $pc$.

Therefore, we have shown that T has the stated properties, considering the impact of the failure of multivalued injectivity at the point of contact $pc$.    □

Theorem 4.2 

(Uniqueness of Paths). Let $T=(V,E)$ be an Algebraic Inverse Tree (AIT) constructed from a Discrete Dynamical System (DDS) $(S,F)$ with the analytic inverse function G. For any two nodes $u,v\in V$, there exists a unique path from u to v in T.

Proof. 

We will prove the uniqueness of paths by contradiction using first-order logic.

Step 1: Define the existence of a path between two nodes in T.

$$\begin{array}{c}\hfill \forall u,v\in V:\exists P\subseteq E:\mathrm{Path}(P,u,v)\iff \\ \hfill (P=\{(w_1,w_2),(w_2,w_3),\dots ,(w_{n-1},w_n)\}\\ \hfill \wedge w_1=u\wedge w_n=v\wedge \forall i\in \{1,\dots ,n-1\}:(w_i,w_{i+1})\in E)\end{array}$$

Step 2: Assume, for contradiction, that there exist two distinct paths between nodes u and v in T.

$$\begin{array}{c}\hfill \exists u,v\in V,\exists P_1,P_2\subseteq E:\\ \hfill (\mathrm{Path}(P_1,u,v)\wedge \mathrm{Path}(P_2,u,v)\wedge P_1\ne P_2)\end{array}$$

Step 3: Let w be the first node at which the paths $P_1$ and $P_2$ differ.

$$\begin{array}{c}\hfill \exists w\in V,\exists i,j\in \mathbb{N}:\\ \hfill (w\in P_1\wedge w\in P_2\wedge P_1\left[i\right]=w\wedge P_2\left[j\right]=w\\ \hfill \wedge \forall k<min(i,j):P_1\left[k\right]=P_2\left[k\right]\\ \hfill \wedge P_1[i+1]\ne P_2[j+1])\end{array}$$

Step 4: By the construction of T using the injective function G, each node has a unique parent. Therefore, w cannot have two distinct children in T.

$$\begin{array}{c}\hfill \forall w\in V,\forall x,y\in V:\\ \hfill \left(\right(w,x)\in E\wedge (w,y)\in E\to x=y)\end{array}$$

Step 5: The existence of two distinct paths $P_1$ and $P_2$ contradicts the unique parent property of T. Therefore, the assumption in Step 2 must be false.

Step 6: We conclude that for any two nodes $u,v\in V$, there exists a unique path from u to v in T.

$$\begin{array}{c}\hfill \forall u,v\in V,\exists !P\subseteq E:\mathrm{Path}(P,u,v)\end{array}$$

Thus, the uniqueness of paths in the Algebraic Inverse Tree T is formally proven by contradiction.    □

Theorem 4.3 

(Uniqueness of Non-Trivial Cycles in DIDS). Let $G:S\to \mathcal{P}\left(S\right)$ be the inverse function of a generic DIDS $(S,F)$, where S is the state space and $F:S\to S$ is the evolution function. Let $pc\in S$ be the point of contact, i.e., the only point where G fails to be multivalued injective. Then:

(1) 

Any non-trivial cycle in the inverse algebraic tree of $(S,F)$ must include the point of contact $pc$.

(2) 

Any non-trivial cycle including $pc$ must have a specific structure:

$$\begin{array}{c}\hfill \exists k\in \mathbb{N},\exists y_1,\dots ,y_k\in S:(x_1=pc)\wedge \hspace{0.17em}\\ \hfill (\forall i\in \{1,\dots ,k\}:x_{i+1}=y_i)\wedge (y_1\in G\left(pc\right))\wedge \hspace{0.17em}\\ \hfill (\forall i\in \{1,\dots ,k-1\}:y_{i+1}\in G\left(y_i\right))\wedge (pc\in G\left(y_k\right))\end{array}$$

where k is a constant specific to the system.

(3) 

There exists a unique non-trivial cycle including $pc$, which is an attractor cycle:

$$\begin{array}{c}\hfill \exists !x_1,\dots ,x_k\in S:\mathit{NTC}(x_1,\dots ,x_k)\wedge pc\in \{x_1,\dots ,x_k\}\end{array}$$

Proof. 

Let $G:S\to \mathcal{P}\left(S\right)$ be the inverse function of a generic DIDS $(S,F)$, where S is the state space and $F:S\to S$ is the evolution function. Let $pc\in S$ be the point of contact, i.e., the only point where G fails to be multivalued injective.

Step 1: Define the notion of a non-trivial cycle.

$$\begin{array}{cc}\hfill \forall x_1,\dots ,x_n\in S:\mathrm{NTC}(x_1,\dots ,x_n)\iff & (n\ge 3)\wedge (x_1=x_n)\wedge \hspace{0.17em}\hfill \\ \hfill & (\forall i\in \{1,\dots ,n-1\}:x_{i+1}\in G\left(x_i\right))\hfill \end{array}$$

Step 2: Prove that any non-trivial cycle must include the point of contact $pc$.

$$\begin{array}{c}\hfill \forall x_1,\dots ,x_n\in S:\mathrm{NTC}(x_1,\dots ,x_n)\Rightarrow pc\in \{x_1,\dots ,x_n\}\end{array}$$

Proof: Assume, for contradiction, that there exists a non-trivial cycle $x_1,\dots ,x_n$ such that $pc\notin \{x_1,\dots ,x_n\}$. Then, by the definition of G being multivalued injective for all points except $pc$, we have:

$$\begin{array}{c}\hfill \forall i,j\in \{1,\dots ,n\},i\ne j:G\left(x_i\right)\cap G\left(x_j\right)=\varnothing \end{array}$$

However, since $x_1=x_n$ and $x_2\in G\left(x_1\right)$, we have $x_2\in G\left(x_n\right)$, contradicting the multivalued injectivity of G for points other than $pc$. Therefore, any non-trivial cycle must include $pc$.

Step 3: Prove that any non-trivial cycle including $pc$ must have a specific structure.

$$\begin{array}{c}\hfill \forall x_1,\dots ,x_n\in S:(\mathrm{NTC}(x_1,\dots ,x_n)\wedge pc\in \{x_1,\dots ,x_n\})\Rightarrow \hspace{0.17em}\\ \hfill (\exists k\in \mathbb{N},\exists y_1,\dots ,y_k\in S:(n=k+1)\wedge (x_1=pc)\wedge \hspace{0.17em}\\ \hfill (\forall i\in \{1,\dots ,k\}:x_{i+1}=y_i)\wedge (y_1\in G\left(pc\right))\wedge \hspace{0.17em}\\ \hfill (\forall i\in \{1,\dots ,k-1\}:y_{i+1}\in G\left(y_i\right))\wedge (pc\in G\left(y_k\right)))\end{array}$$

where k is a constant specific to the system.

Proof: Let $x_1,\dots ,x_n\in S$ be a non-trivial cycle that includes $pc$. Without loss of generality, let $x_1=pc$. Since G is multivalued injective for all points except $pc$, each $x_i$ for $i\in \{2,\dots ,n-1\}$ must have a unique predecessor in the cycle. Let $y_1,\dots ,y_{n-2}$ be these unique predecessors, i.e., $x_{i+1}=y_i$ for $i\in \{1,\dots ,n-2\}$.

Now, since $x_n=pc$, we have $y_{n-2}\in G\left(pc\right)$. Furthermore, since G is exhaustive, there exists a finite sequence of applications of G that leads from $y_{n-2}$ back to $pc$. Let k be the length of this sequence, and let $y_{n-1},\dots ,y_k$ be the elements of this sequence, i.e., $y_{i+1}\in G\left(y_i\right)$ for $i\in \{n-2,\dots ,k-1\}$ and $pc\in G\left(y_k\right)$.

Then, the non-trivial cycle $x_1,\dots ,x_n$ has the structure:

$$\begin{array}{cc}\hfill x_1& =pc\hfill \\ \hfill x_2& =y_1\hfill \\ \hfill & ⋮\hfill \\ \hfill x_{k+1}& =y_k\hfill \\ \hfill x_{k+2}& =pc\hfill \end{array}$$

which satisfies the claimed properties with $n=k+1$.

Step 4: Prove that any non-trivial cycle including $pc$ is unique and must be an attractor cycle.

$$\begin{array}{c}\hfill \exists !x_1,\dots ,x_k\in S:\mathrm{NTC}(x_1,\dots ,x_k)\wedge pc\in \{x_1,\dots ,x_k\}\end{array}$$

Proof: Suppose, for contradiction, that there exist two distinct non-trivial cycles $x_1,\dots ,x_k$ and $x_1^{\prime },\dots ,x_{k^{\prime }}^{\prime }$ that include $pc$. Without loss of generality, let $x_1=x_1^{\prime }=pc$.

By Step 3, both cycles must have the structure:

$$\begin{array}{cc}\hfill x_1& =pc\hfill \\ \hfill x_2& =y_1\hfill \\ \hfill & ⋮\hfill \\ \hfill x_k& =y_{k-1}\hfill \\ \hfill x_1& =pc\in G\left(y_{k-1}\right)\hfill \end{array}$$

and

$$\begin{array}{cc}\hfill x_1^{\prime }& =pc\hfill \\ \hfill x_2^{\prime }& =y_1^{\prime }\hfill \\ \hfill & ⋮\hfill \\ \hfill x_{k^{\prime }}^{\prime }& =y_{k^{\prime }-1}^{\prime }\hfill \\ \hfill x_1^{\prime }& =pc\in G\left(y_{k^{\prime }-1}^{\prime }\right)\hfill \end{array}$$

where $y_1,\dots ,y_{k-1}$ and $y_1^{\prime },\dots ,y_{k^{\prime }-1}^{\prime }$ are the unique predecessors of $x_2,\dots ,x_k$ and $x_2^{\prime },\dots ,x_{k^{\prime }}^{\prime }$, respectively.

Now, since G is a function, we have $y_1=y_1^{\prime }$. By induction, this implies $y_i=y_i^{\prime }$ for all $i\in \{1,\dots ,min(k-1,k^{\prime }-1)\}$. If $k<k^{\prime }$, then $pc\in G\left(y_{k-1}\right)=G\left(y_{k-1}^{\prime }\right)$, contradicting the fact that $y_{k-1}^{\prime }$ has a unique successor in the cycle $x_1^{\prime },\dots ,x_{k^{\prime }}^{\prime }$. Similarly, if $k^{\prime }<k$, we obtain a contradiction. Therefore, $k=k^{\prime }$, and the two cycles are identical.

To show that the unique non-trivial cycle including $pc$ is an attractor cycle, let $x\in S$ be any point such that there exists a finite sequence of applications of G leading from x to $pc$. By the exhaustiveness of G, such a point always exists. Then, by the same argument as in Step 3, the sequence of applications of G starting from x must eventually enter the unique non-trivial cycle including $pc$, which implies that this cycle is an attractor.

Therefore, we have shown that the only possible non-trivial cycle in the inverse algebraic tree of a generic DIDS is a unique attractor cycle that includes the point of contact $pc$, and there cannot be any other non-trivial cycles in the system.    □

Theorem 4.4 

(Convergence of Distinct Trajectories). Let $(S,F)$ be a discrete dynamical system and $T=(V,E)$ be the associated inverse algebraic tree generated by the inverse analytic function $G:S\to \mathcal{P}\left(S\right)$. For any two distinct trajectories $P_1,P_2\subset V$ in the same tree T, both trajectories converge to a common node $u\in V$, which is ultimately the root node of T.

Proof. 

Let $(S,F)$ be a discrete dynamical system and $T=(V,E)$ be the associated inverse algebraic tree generated by the inverse analytic function $G:S\to \mathcal{P}\left(S\right)$. Consider two distinct trajectories $P_1,P_2\subset V$ in the same tree T.

Step 1: Define the notion of a trajectory in T.

$$\begin{array}{c}\hfill \forall P\subseteq V:\mathrm{Trajectory}\left(P\right)\iff (\forall u,w\in P:(u,w)\in E\vee (w,u)\in E)\end{array}$$

Step 2: Define the convergence of a trajectory to a node.

$$\begin{array}{c}\hfill \forall P\subseteq V,\forall u\in V:\mathrm{Converges}(P,u)\iff \\ \hfill (\forall \epsilon >0,\exists N\in \mathbb{N},\forall n\ge N,\forall w\in P:d(w,u)<\epsilon )\end{array}$$

where d is the graph distance in T.

Step 3: Prove that every node in T has a unique path to the root node.

$$\begin{array}{c}\hfill \forall v\in V,\exists !P\subseteq V:\left(\mathrm{Trajectory}\right(P)\wedge v\in P\wedge \exists r\in V:\\ \hfill \left(\mathrm{Root}\right(r)\wedge r\in P\wedge \forall u\in P\setminus \{r\}:(u,r)\notin E))\end{array}$$

Proof: By the recursive construction of T using the injective function G, each node has a unique parent. Therefore, for any node $v\in V$, there exists a unique path from v to the root node r, which is obtained by following the parent nodes until reaching r.

Step 4: Prove that if $P_1$ and $P_2$ are in the same tree T, they must share a common node.

$$\begin{array}{c}\hfill \mathrm{Trajectory}\left(P_1\right)\wedge \mathrm{Trajectory}\left(P_2\right)\wedge P_1,P_2\subset V\\ \hfill \Rightarrow \exists v\in V:(v\in P_1\wedge v\in P_2)\end{array}$$

Proof: Assume, for contradiction, that $P_1$ and $P_2$ do not share any common node. Then, there exists a node $w\in P_1$ such that $w\notin P_2$. By Step 3, there is a unique path from w to the root node r. This path must intersect $P_2$ at some node v, as both paths end at r. Therefore, $v\in P_1$ and $v\in P_2$, contradicting the assumption that $P_1$ and $P_2$ do not share any common node.

Step 5: Let v be a common node of $P_1$ and $P_2$, and let $P_v$ be the unique path from v to the root node r. Prove that $P_1$ and $P_2$ converge to r.

$$\begin{array}{c}\hfill \mathrm{Trajectory}\left(P_1\right)\wedge \mathrm{Trajectory}\left(P_2\right)\wedge v\in P_1\cap P_2\\ \hfill \Rightarrow \exists P_v\subseteq V:(v\in P_v\wedge P_v\subseteq P_1\wedge P_v\subseteq P_2)\\ \hfill \Rightarrow \mathrm{Converges}(P_1,r)\wedge \mathrm{Converges}(P_2,r)\end{array}$$

Proof: By Step 4, there exists a common node $v\in P_1\cap P_2$. By Step 3, there is a unique path $P_v$ from v to the root node r. Since $v\in P_1$ and $v\in P_2$, and $P_v$ is the unique path from v to r, we have $P_v\subseteq P_1$ and $P_v\subseteq P_2$. Therefore, both $P_1$ and $P_2$ converge to the root node r via the common subpath $P_v$.

Therefore, if $P_1$ and $P_2$ are in the same inverse algebraic tree T, they necessarily converge to a common node, which is ultimately the root node r of T, completing the proof.    □

Remark 1 

(Observation on the Theorem of Convergence of Trajectories). The convergence of distinct trajectories to a common node is supported by the theorem of uniqueness of non-trivial cycles in DIDS, which ensures that there are no additional cycles that could trap trajectories and prevent their convergence towards the root node.

Remark 2 

(Observation on the Theorem of Universal Convergence of Trajectories). The universal convergence of trajectories towards the root node is supported by the theorem of uniqueness of non-trivial cycles in DIDS, which establishes the existence of a unique non-trivial cycle that includes the point of contact $pc$ and demonstrates its attracting nature, ensuring that all trajectories eventually converge towards the root node.

Corollary 4.1. 

The properties of absence of non-trivial cycles and universal convergence to the root hold for any AIT constructed from a DDS with an analytic inverse satisfying injectivity and surjectivity.

Proof. 

Let $T=(V,E)$ be an AIT constructed from a DDS $(S,\tau ,F)$ with an analytic inverse G that satisfies injectivity and surjectivity.

To show that T has no non-trivial cycles, suppose for contradiction that there exists a non-trivial cycle $C=v_1,\dots ,v_k$ with $k\ge 3$. By the injectivity of G, each node has a unique parent. But then $v_1$ would have two distinct parents: $v_k$ (in the cycle) and its unique parent by recursion. This leads to a contradiction, so no such cycle exists.

To show that all paths in T converge to the root node r, let $P=(v_1,v_2,\dots )$ be an arbitrary infinite path in T. By the surjectivity of G, each node has a child. By injectivity, the sequence of depths $d\left(v_i\right)$ is strictly decreasing. As natural numbers are well-ordered, there exists an n such that $d\left(v_n\right)=0$, i.e., $v_n=r$. By the uniqueness of paths, P converges to r.

Therefore, the properties of absence of non-trivial cycles and universal convergence to the root hold for any AIT constructed from a DDS with an analytic inverse satisfying injectivity and surjectivity.    □

4.1. Algebraic Inverse Tree Construction

The construction of the algebraic inverse tree $T=(V,E)$ is done by recursively applying the analytical inverse function $G:S\to P\left(S\right)$, which undoes the steps of the evolution rule F of the canonical discrete dynamical system $(S,\tau )$. This process generates a hierarchical structure where each node $v\in V$ represents a state in S, and each edge $(u,v)\in E$ indicates that v is a predecessor of u under the inverse dynamics determined by G.

Given this construction, we can naturally define a function $f:T\to S$ that associates each node $v\in V$ with its corresponding state $s\in S$. Formally:

$$f\left(v\right)=s\iff v\mathrm{represents}\mathrm{state}s\mathrm{in}T$$

Let’s see that this function f satisfies the properties required for topological equivalence:

(1)

f is bijective: By construction, each node $v\in V$ represents a unique state $s\in S$, and each state $s\in S$ is represented by at least one node $v\in V$ (due to the exhaustiveness of G). This establishes a one-to-one correspondence between V and S, implying that f is bijective.

(2)

f and $f^{-1}$ are continuous: To show the continuity of f and $f^{-1}$, we must verify that the inverse images of open sets are open in the respective topologies.

• Continuity of f: Let $U\in \tau$ be an open set in $(S,\tau )$. We need to prove that $f^{-1}\left(U\right)$ is open in $(T,\rho )$. By definition of the discrete topology $\tau$, each state $s\in S$ is an open set. Thus, $f^{-1}\left(U\right)=\{v\in V:f\left(v\right)\in U\}$ is a union of individual nodes in T, which are open in the natural topology $\rho$. Therefore, $f^{-1}\left(U\right)$ is open in $(T,\rho )$.
• Continuity of $f^{-1}$: Let $W\in \rho$ be an open set in $(T,\rho )$. We need to prove that $f\left(W\right)$ is open in $(S,\tau )$. Since $\rho$ is the natural topology on T, each node $v\in V$ and each set of nodes form an open set. Hence, $f\left(W\right)=\{s\in S:f^{-1}\left(s\right)\in W\}$ is a union of individual states in S, which are open in the discrete topology $\tau$. Therefore, $f\left(W\right)$ is open in $(S,\tau )$.

Thus, we have demonstrated that the function f induced by the construction of the algebraic inverse tree T from the function G satisfies the properties of bijectivity and bicontinuity, establishing a topological equivalence between $(S,\tau )$ and $(T,\rho )$.

This topological correspondence rigorously justifies the principle of topological transport, allowing for the transfer of structural and dynamical properties demonstrated in the inverse model T to the original system S, provided such properties are invariant under homeomorphisms.

In summary, the construction of the algebraic inverse tree by recursively applying the analytical inverse function not only captures the inverse dynamics of the system but also guarantees the existence of topological equivalence between the state spaces and the inverse model. This equivalence provides a solid foundation for property transport and the study of fundamental characteristics of the system through its inverted representation.

4.2. Steps of the Inverse Modeling Process

Definitions:

• Dynamic_System = (E, R) where:
  E is the discrete set of states
  R is the evolution function
• Inverse_Function = (R${\hspace{0.17em}}^{-1}$, A) where:
  R${\hspace{0.17em}}^{-1}$ is the inverse function of R
  A is the resulting Inverse_Tree
• Inverse_Tree = (N, V) where:
  N is the set of nodes
  V are the upward links between nodes

Construction:

(1)

Given Dynamic_System, determine R${\hspace{0.17em}}^{-1}$ by applying the definition of Inverse_Function.

(2)

Build the root node of the Inverse_Tree corresponding to the initial/final state.

(3)

Apply R${\hspace{0.17em}}^{-1}$ recursively on nodes to generate upward_links.

(4)

Repeat step 3 until exhausting states in E, completing V.

(5)

Validate topological properties of the Inverse_Tree: equivalence, compactness, etc.

(6)

Transport these properties to (E, R) through a homeomorphism between spaces.

5. Structural Analysis

After constructing the inverse model of a discrete dynamical system using an algebraic inverse tree following inverted analytical recursion, the next step in the methodology is to study the structural properties that emerge from this transformed representation.

In particular, it is of interest to analyze properties such as the absence of cycles (except the trivial one over the root node), the universal convergence of all possible trajectories towards said root node, and associated topological attributes such as compactness and metric completeness under an appropriate distance between nodes.

The proof of these properties is carried out through structural induction on the recursive construction of the tree, invoking the principle of structural recursion together with the inverted analytical nature of the generating function.

Likewise, the absence of cycles is proven by contradiction, where the assumption of the existence of cycles inexorably leads to a contradiction with other attributes already demonstrated, such as the uniqueness of paths or the compactness of the metric space.

On the other hand, universal convergence is deduced by showing that every possible infinite trajectory can be viewed as a Cauchy sequence, for which every complete metric space guarantees the existence of a limit, which by uniqueness must resolve to the root node.

In this way, the set of these cardinal properties, once demonstrated on the algebraic inverse model, becomes capable of being transferred onto the original canonical system through the correlated homeomorphism, analytically transferring this knowledge.

Definition 5.1 

(Path in a Tree). Let $T=(V,E)$ be a directed tree. A path in T is a finite or infinite sequence of nodes $P=\langle v_1,v_2,\dots \rangle$ such that $(v_i,v_{i+1})\in E,\forall i$.

Definition 5.2 

(Cycle). A cycle is a closed path $C=\langle v_1,\dots ,v_k\rangle$ where $v_1=v_k$ and $(v_i,v_{i+1})\in E,\forall 1\le i<k$. We say that C is non-trivial if $k\ge 3$.

Definition 5.3. 

Let $(X,d)$ be a metric space. A sequence ${\left(x_n\right)}_{n\in \mathbb{N}}$ in X is called a Cauchy sequence if:

$$\forall \epsilon >0,\exists N\in \mathbb{N}:\forall n,m\ge N,d(x_n,x_m)<\epsilon$$

Definition 5.4. 

A metric space $(X,d)$ is said to be complete if every Cauchy sequence ${\left(x_n\right)}_{n\in \mathbb{N}}$ in X converges to some point $x\in X$. In other words:

$$\forall \left(x_n\right)\subseteq X,\left(x_n\right)\mathrm{is}\mathit{Cauchy}\Rightarrow \exists x\in X:\underset{n\to \infty }{lim}{x}_n=x$$

Lemma 5.1

(Metric Completeness). Let $(T,d)$ be an algebraic inverse tree with the path length metric d. Then $(T,d)$ is a complete metric space.

Proof. 

Let $(T,d)$ be the inverse algebraic tree equipped with the metric d. We aim to prove that $(T,d)$ is complete, meaning every Cauchy sequence in T converges to a point in T.

Consider a Cauchy sequence $\left(v_n\right)$ in T. Formally, this means:

$$\begin{array}{c}\hfill \forall \epsilon >0,\exists N\in \mathbb{N},\forall m,n\ge N:d(v_m,v_n)<\epsilon \end{array}$$

Since T is recursively constructed from the complete metric space $(X,d_X)$, and the inverse function G is exhaustive, for each $v_n$, there exists a unique path $P_n$ from $v_n$ to a root node $r_n$ in T, corresponding to a Cauchy sequence $\left(x_n\right)$ in X.

Because $(X,d_X)$ is complete, $\left(x_n\right)$ converges to a point x in X. Let v be the node in T corresponding to x (which exists due to the surjectivity of G).

We now demonstrate that $\left(v_n\right)$ converges to v in T:

$$\begin{array}{cc}\hfill \forall \epsilon >0,\exists N\in \mathbb{N},\forall n\ge N:d(v_n,v)& =d_X(x_n,x)(\mathrm{by}\mathrm{the}\mathrm{definition}\mathrm{of}d)\hfill \\ \hfill & <\epsilon (\mathrm{since}\left(x_n\right)\to x\mathrm{in}X)\hfill \end{array}$$

Consequently, $\left(v_n\right)$ converges to v in T, affirming that $(T,d)$ is complete. □

Definition 5.5. 

Let $(X,d_X)$ be a complete metric space and let $T=(V,E)$ be an inverse algebraic tree constructed from a discrete dynamical system $(X,f)$, where $f:X\to X$ is a continuous function.

Definition 5.6. 

The metric $d_T:V\times V\to \mathbb{R}$ on the inverse algebraic tree T is defined as follows:

$$d_T(u,v)=\left\{\begin{array}{cc}0\hfill & \mathit{if}u=v\hfill \\ d_X(x_u,x_v)\hfill & \mathit{if}u\ne v\hfill \end{array}\right.$$

where $x_u,x_v\in X$ are the states corresponding to the nodes $u,v\in V$, respectively.

Lemma 5.2. 

$(V,d_T)$ is a metric space.

Proof. 

The proof follows directly from the properties of the metric $d_X$ on the complete metric space $(X,d_X)$. For any $u,v,w\in V$:

(1)

Non-negativity: $d_T(u,v)=d_X(x_u,x_v)\ge 0$ since $d_X$ is a metric.

(2)

Identity of indiscernibles: $d_T(u,v)=0$ if and only if $x_u=x_v$, which implies $u=v$ since each node in T corresponds to a unique state in X.

(3)

Symmetry: $d_T(u,v)=d_X(x_u,x_v)=d_X(x_v,x_u)=d_T(v,u)$.

(4)

Triangle inequality: $d_T(u,w)=d_X(x_u,x_w)\le d_X(x_u,x_v)+d_X(x_v,x_w)=d_T(u,v)+d_T(v,w)$.

Therefore, $(V,d_T)$ is a metric space. □

Theorem 5.3 

(Relative Metric Completeness). The inverse algebraic tree $(T,d_T)$ is relatively complete if the metric space $(X,d_X)$ is complete.

Proof. 

Let $\left(s_n\right)$ be a Cauchy sequence in $(S,d_S)$. We want to prove that $\left(G\left(s_n\right)\right)$ is a Cauchy sequence in $(T,d_T)$.

First, we formalize the definition of a Cauchy sequence:

$$\begin{array}{c}\hfill \forall \epsilon >0,\exists N\in \mathbb{N},\forall m,n\ge N:d_S(s_m,s_n)<\epsilon \end{array}$$

Since G is the analytic inverse of F, we have:

$$\begin{array}{c}\hfill \forall s\in S,\forall t\in T:t\in G\left(s\right)\iff F\left(t\right)=s\end{array}$$

Now, let $\epsilon >0$ be given. By the Cauchy property of $\left(s_n\right)$, we know that:

$$\begin{array}{c}\hfill \exists N\in \mathbb{N},\forall m,n\ge N:d_S(s_m,s_n)<\frac{\epsilon }{L}\end{array}$$

where L is the Lipschitz constant of F.

Let $m,n\ge N$. For any $t_m\in G\left(s_m\right)$ and $t_n\in G\left(s_n\right)$, we have:

$$\begin{array}{cc}\hfill d_T(t_m,t_n)& =d_S(F\left(t_m\right),F\left(t_n\right))(\mathrm{by}\mathrm{definition}\mathrm{of}{d}_T)\hfill \\ \hfill & =d_S(s_m,s_n)(\sin \mathrm{ce}{t}_m\in G\left(s_m\right)\mathrm{and}{t}_n\in G\left(s_n\right))\hfill \\ \hfill & <\frac{\epsilon }{L}(\mathrm{by}\mathrm{Cauchy}\mathrm{property}\mathrm{of}\left(s_n\right))\hfill \\ \hfill & \le \epsilon (\sin \mathrm{ce}F\mathrm{is}L-\mathrm{Lipschitz})\hfill \end{array}$$

Therefore, we have shown that:

$$\begin{array}{c}\hfill \forall \epsilon >0,\exists N\in \mathbb{N},\forall m,n\ge N,\forall t_m\in G\left(s_m\right),\forall t_n\in G\left(s_n\right):d_T(t_m,t_n)<\epsilon \end{array}$$

which means that $\left(G\left(s_n\right)\right)$ is a Cauchy sequence in $(T,d_T)$. □

Definition 5.7 

(Algebraic Inverse Tree). Let $(S,F)$ be a discrete dynamical system with analytic inverse G. An algebraic inverse tree is a tuple $(V,E,r,f)$ constructed recursively from G, satisfying:

• V is the set of nodes.
• $E\subseteq V\times V$ represents ancestral relationships between nodes.
• $r\in V$ is the root node.
• $f:V\to S$ is a bijective function correlating nodes with states.
• $\forall (u,v)\in E:v\in G\left(f\right(u\left)\right)$.

Additionally:

• T is compact and complete under a metric.
• T combinatorially condenses all interrelations of $(S,F)$.
• T is recursively constructed from G.
• Absence of non-trivial cycles.
• Universal convergence of paths towards r.

• Flexible Selection of Root Node

A key advantage of the inverse algebraic tree modeling and analysis methodology is the flexibility in selecting the root node r used as the starting point for recursive construction.

Formally, given the discrete state space S of a dynamical system, the root node r can be chosen as any state $s\in S$ that is desired to be used as the final condition or target optimal value for analysis.

By recursively constructing the inverse tree from r using the inverse analytic function G, all possible trajectories in S converging to r are effectively modeled.

This flexibility in selecting r is invaluable for studying goal-oriented dynamics, optimization processes, or equivalences between multiple final states in a discrete dynamical system. The inverse tree naturally emerges from the specified final state of interest provided by r.

Definition 5.8. 

Let (S, F) be the canonical discrete dynamical system (DIDS), with $S=\{s_1,s_2,\dots ,s_n\}$ the discrete state space. Let $T=(V,E)$ be the associated inverse algebraic tree, with $V=\{v_1,v_2,\dots ,v_m\}$ the set of nodes.

The bijective homeomorphic correlation function $f:V\to S$ is defined as:

$$f\left(v_i\right)=\left\{\begin{array}{cc}{s}_i,\hfill & \mathit{if}i\le min(n,m)\hfill \\ s_j,\hfill & \mathit{if}i>n\mathit{and}f\mathit{is}\mathit{injective}\mathit{in}\{v_{n+1},\dots ,v_m\}\hfill \end{array}\right.$$

This function explicitly establishes an identity correlation between each node $v_i$ of the inverse tree T and the corresponding state $s_i$ in the discrete canonical system S, for all $i\le min(n,m)$. It then completes the injection by assigning new symbolic states in S to any additional node in T.

Definition 5.9 

(Inverse Forest). Let $(S,F)$ be a discrete dynamic system with n possible final states $r_1,\dots ,r_n\subseteq S$. The inverse forest F is defined as the collection of n disjoint inverse trees $F=\{T_{r_1},\dots ,T_{r_n}\}$, where each tree $T_{r_i}$ is constructed by recursively applying the inverse function G rooted at the final state $r_i$.

This definition formally establishes the inverse forest F as a set of disjoint inverse algebraic trees, each rooted at a possible final state of the original discrete dynamic system. Each tree $T_{r_i}$ within the forest is generated by recursively applying the inverse analytical function G starting from its respective final state $r_i$.

Definition 5.10 

(Total State Space). Let $F=\{T_{r_1},\dots ,T_{r_n}\}$ be the inverse forest of a discrete dynamic system $(S,F)$ with n possible final states $r_1,\dots ,r_n$. We define the total state space $\widehat{S}$ as the union of nodes contained in each inverse tree:

$$\widehat{S}=\bigcup _{i=1}^{n}V\left(T_{r_i}\right)$$

where $V\left(T_{r_i}\right)$ denotes the set of nodes of tree $T_{r_i}$.

This definition introduces the total state space $\widehat{S}$ as the union of all nodes belonging to each inverse tree in the forest F. Intuitively, $\widehat{S}$ represents the complete set of reachable states in the original discrete dynamic system, as captured through the structure of the inverse model.

Theorem 5.4. 

Let $T_{r_i},T_{r_j}\in \mathcal{F}$ be two distinct inverse trees rooted at the final states $r_i$ and $r_j$ respectively. Then $T_{r_i}\cap T_{r_j}=\varnothing$.

Proof. 

We reason by contradiction. Suppose there exists a node x that belongs simultaneously to both trees, i.e.:

$$x\in T_{r_i}\mathrm{and}x\in T_{r_j}$$

By the recursive construction of the trees applying G, we have:

$$G^p\left(x\right)=r_i\mathrm{and}{G}^q\left(x\right)=r_j$$

for some orders $p,q\in \mathbb{N}$.

But as G is injective, if $G^p\left(x\right)=a$ and $G^q\left(x\right)=b$, it must necessarily hold that $a=b$. In particular, for the final states $r_i$ and $r_j$.

Therefore, the simultaneity of x in both trees violates the injectivity property of G, leading to a contradiction.

Thus, by contradiction, it is concluded that:

$$T_{r_i}\cap T_{r_j}=\varnothing$$

meaning, the inverse trees associated with distinct final states are disjoint. □

Definition 5.11 

(Total State Space). Let $\mathcal{F}=\{T_{r_1},\dots ,T_{r_n}\}$ be the inverse forest of a DIDS with n possible final states $\{r_1,\dots ,r_n\}$. We define the total state space $\widehat{S}$ as the union of the nodes contained in each inverse tree:

$$\widehat{S}=\bigcup _{i=1}^{n}V\left(T_{r_i}\right)$$

where $V\left(T_{r_i}\right)$ denotes the set of nodes of the tree $T_{r_i}$.

Theorem 5.5 

(Completeness of the State Space). Let $(S,F)$ be a DIDS and $\mathcal{F}$ its inverse forest. Then the total state space $\widehat{S}$ contains all the reachable states in the original discrete system. That is:

$$S\subseteq \widehat{S}$$

Proof. 

Let $(S,F)$ be a DIDS and $\mathcal{F}=\{T_{r_1},\dots ,T_{r_n}\}$ its inverse forest with n trees rooted at the final states $\{r_1,\dots ,r_n\}\subseteq S$.

By the exhaustiveness property of the inverse function G, we have that $\forall x\in S,\exists k\in \mathbb{N}:G^k\left(x\right)=r_i$, for some final state $r_i$.

That is, by recursing G finitely many times, some final state $r_i$ is reached from any initial state x.

Due to the recursive construction of each tree $T_{r_i}\in \mathcal{F}$ applying G, any state $x\in S$ leading to $r_i\in S$ under the iteration of F is contained as a node in $T_{r_i}$.

Formally:

$$x\in S,G^k\left(x\right)=r_i\Rightarrow x\in V\left(T_{r_i}\right)$$

Taking the union over all trees:

$$\bigcup _{i=1}^{n}V\left(T_{r_i}\right)\supseteq S$$

Thus, it’s demonstrated that the total state space $\widehat{S}$ contains S, completing the proof. □

Theorem 5.6. 

Let $(S,F)$ be a Discrete Dynamical System, where S is a countable state space and $F:S\to S$ is the deterministic and surjective evolution function. Let $G:S\to P\left(S\right)$ be the analytic inverse of F, which is multivalued injective, surjective, and exhaustive. Let $\mathcal{F}=\{T_1,\dots ,T_k\}$ be the Inverse Algebraic Forest generated by G, where each $T_i$ is a tree.

Then, $\mathcal{F}$ is unique and each $T_i\in \mathcal{F}$ is a single connected component.

Proof. 

First, we prove that each $T_i$ is connected.

Suppose, for contradiction, that there exist two nodes $v_1,v_2\in V_i$ such that there is no sequence of edges connecting $v_1$ and $v_2$. This implies that $v_1$ and $v_2$ belong to two separate connected components, say $T_{i1}$ and $T_{i2}$, respectively.

Step 1: Exhaustiveness of G (Generalized to countable S) By the exhaustiveness property of G, for each node $v\in V_i$, there exists a finite sequence of applications of G that leads to a root node $r_i$. Formally:

$$\forall v\in V_i,\exists n\in \mathbb{N},\exists r_i\in V_i:(Root\left(r_i\right)\wedge v\in G^n\left(r_i\right))$$

where $Root\left(r_i\right)$ denotes that $r_i$ is a root node, and $G^n$ represents the n-fold composition of G with itself.

Let $r_{i1}$ and $r_{i2}$ be the root nodes of $T_{i1}$ and $T_{i2}$, respectively.

Step 2: Determinism and Surjectivity of F (Generalized to countable S) By the determinism of F, each node in $T_i$ has a unique child. By the surjectivity of F, each node in $T_i$, except for the root nodes, has a unique parent. Formally:

$$\forall v\in V_i\setminus \{r_{i1},r_{i2}\},\exists !u\in V_i:(u,v)\in E_i$$

Step 3: Contradiction We have shown that the existence of separate components $T_{i1}$ and $T_{i2}$ leads to a contradiction when F is deterministic and surjective, and G is exhaustive, even for a countable state space S.

Therefore, each $T_i$ must be a single connected component.

Now, we prove the uniqueness of $\mathcal{F}$ using the Path Uniqueness Theorem.

Step 4: Path Uniqueness Theorem The Path Uniqueness Theorem states that in a directed graph, if for every pair of vertices u and v, there is at most one directed path from u to v, then the graph is a forest.

In the context of our Inverse Algebraic Forest $\mathcal{F}$, this means that if for every pair of nodes $v_1,v_2\in V_i$ in each tree $T_i$, there is at most one sequence of edges from $v_1$ to $v_2$, then $\mathcal{F}$ is unique.

Step 5: Uniqueness of Paths in each $T_i$ Let $v_1,v_2\in V_i$ be any two nodes in $T_i$. Suppose there are two distinct sequences of edges from $v_1$ to $v_2$, denoted by $P_1$ and $P_2$.

Let u be the last common node of $P_1$ and $P_2$ before they diverge. Let $u_1$ and $u_2$ be the next nodes after u in $P_1$ and $P_2$, respectively.

By the determinism of F, u can have only one child. Therefore, $u_1=u_2$, contradicting the assumption that $P_1$ and $P_2$ are distinct paths.

Thus, there can be at most one path between any two nodes in each $T_i$.

Step 6: Application of Path Uniqueness Theorem By Step 5, each $T_i$ satisfies the condition of the Path Uniqueness Theorem. Therefore, $\mathcal{F}$ is unique.

Conclusion: We have shown that the Inverse Algebraic Forest $\mathcal{F}$ generated by G is unique and each tree $T_i\in \mathcal{F}$ is a single connected component, even when the state space S is countable. □

Corollary 5.1. 

Given a Discrete Inverse Dynamical System (DIDS) with a state space S (either finite or countably infinite) and an analytic inverse function $G:S\to P\left(S\right)$ that is injective, multivalued, surjective, and exhaustive, the system has a unique attractor set.

Proof. 

By the theorem, the inverse model of the system can be represented by a unique inverse algebraic forest $\mathcal{F}$. Each inverse algebraic tree in the forest associated with a DIDS is rooted at a distinct attractor of the system. Since the forest $\mathcal{F}$ is unique and consists of disjoint trees, the attractor set of the system is also unique. □

Theorem 5.7. 

Let $(S,F)$ be a Discrete Dynamical System, where S is a countable state space and $F:S\to S$ is the deterministic and surjective evolution function. Let $G:S\to P\left(S\right)$ be the analytic inverse of F, which is multivalued injective, surjective, and exhaustive, except at the point of contact. Then, the Inverse Algebraic Forest $\mathcal{F}$ generated by G does not guarantee the existence of a single tree.

Proof. 

Step 1: Define the exception to multivalued injectivity at the point of contact.

$$\exists pc\in S,\exists x\in S\setminus \left\{pc\right\}:G\left(x\right)\cap G\left(pc\right)\ne \varnothing$$

where $pc$ is the point of contact.

Step 2: Assume, for contradiction, that the Inverse Algebraic Forest $\mathcal{F}$ always consists of a single tree.

$$\forall (S,F),\forall G:\left|\mathcal{F}\right|=1$$

where $\left|\mathcal{F}\right|$ denotes the number of trees in the forest $\mathcal{F}$.

Step 3: Construct a counterexample. Consider the following Discrete Dynamical System $(S,F)$:

$$\begin{array}{cc}\hfill S& =\{0,1,2,3\}\hfill \\ \hfill F\left(0\right)& =0\hfill \\ \hfill F\left(1\right)& =0\hfill \\ \hfill F\left(2\right)& =1\hfill \\ \hfill F\left(3\right)& =2\hfill \end{array}$$

Let the point of contact be $pc=0$. Then, the analytic inverse function G is given by:

$$\begin{array}{cc}\hfill G\left(0\right)& =\{0,1\}\hfill \\ \hfill G\left(1\right)& =\left\{2\right\}\hfill \\ \hfill G\left(2\right)& =\left\{3\right\}\hfill \\ \hfill G\left(3\right)& =\varnothing \hfill \end{array}$$

Note that $G\left(0\right)\cap G\left(1\right)=\left\{1\right\}\ne \varnothing$, violating multivalued injectivity at the point of contact $pc=0$.

Step 4: Construct the Inverse Algebraic Forest $\mathcal{F}$. Following the construction process outlined in the previous proof, we obtain two trees:

$$\begin{array}{cc}\hfill T_1& =\left(\right\{0,1,2,3\},\{(1,0),(2,1),(3,2)\left\}\right)\hfill \\ \hfill T_2& =\left(\right\{3\},\varnothing )\hfill \end{array}$$

Therefore, $\left|\mathcal{F}\right|=2$, contradicting the assumption in Step 2.

Step 5: Conclude that the analytic inverse function G does not guarantee the existence of a single tree in the Inverse Algebraic Forest.

$$\exists (S,F),\exists G:\left|\mathcal{F}\right|>1$$

The exception to multivalued injectivity at the point of contact allows for the possibility of multiple trees in the Inverse Algebraic Forest generated by the analytic inverse function G. □

Definition 5.12 

(Cardinal Properties of AIT). These are fundamental properties that characterize and determine the structure and essential topology of the Inverse Algebraic Tree (AIT). They include:

1. 

Absence of anomalous cycles: There are no closed cycles of length $\ge 3$ in the AIT, since each node has a unique predecessor.

2. 

Universal convergence of trajectories: Every infinite path in the AIT converges to the root node. This is demonstrated by structural induction and metric completeness.

3. 

Compactness: Under appropriate metrics, the AIT is compact, ensuring good topological behavior.

4. 

Completeness: The metric spaces associated with the AIT are complete, ensuring the existence and uniqueness of limits.

5. 

Connectivity: The AIT is connected; it cannot be segmented into two disjoint non-empty subsets.

Definition 5.13 

(Non-Cardinal Properties of AIT). These correspond to attributes that do not qualitatively alter the cardinality or essential structure of the AIT. They include:

1. 

Labeling: The names or labels assigned to the nodes.

2. 

Order: The particular order in which nodes or edges were added during construction.

3. 

Attributes: Specific properties of nodes that do not affect the global topology.

Lemma 5.8 

(Compactness). Every finite algebraic inverse tree $(T,d)$ is compact under the natural topology.

Proof. 

Let $(T,d)$ be a finite algebraic inverse tree. We prove its compactness:

(1)

T is totally bounded: Since T is finite, it is bounded. Therefore, there exists $M>0$ such that $T\subseteq B_d(v,M)$ for some $v\in T$. Explicitly, the open balls $B_{\epsilon }\left(v_k\right)$ with radii $\epsilon >0$ centered at nodes $v_k\in T$ cover T due to its finite size.

(2)

T is complete: Every finite set is complete under the metric d. Specifically, any closed and bounded subset $K\subseteq T$ is contained within a closed ball of radius R that only contains a few points (as T is finite), making K a finite set and thus compact.

(3)

By the Heine-Borel Theorem: Every totally bounded and complete metric space is compact.

Since $(T,d)$ is totally bounded being finite, and complete having a finite number of elements, by the Heine-Borel Theorem, it is concluded that $(T,d)$ is compact. □

Definition 5.14. 

Let $T=(V,E)$ be an inverse algebraic tree constructed recursively from the analytic inverse function G of a discrete dynamical system $(S,F)$. We say that T satisfies K-bounded growth if there exists $K\in \mathbb{N}$ such that:

$$\forall v\in V:\left|Children\right(v\left)\right|\le K$$

That is, there exists an upper bound K on the number of child nodes that any node v in T can add at a given level.

Theorem 5.9 

(Relative Compactness). Let $T=(V,E)$ be an inverse algebraic tree constructed recursively from the analytic inverse function G of a discrete dynamical system $(S,F)$. Suppose that there exists a function $f:\mathbb{N}\to \mathbb{N}$ such that:

(1) 

f is non-decreasing, i.e., $\forall n,m\in \mathbb{N},n\le m\Rightarrow f\left(n\right)\le f\left(m\right)$.

(2) 

f is unbounded, i.e., $\forall M\in \mathbb{N},\exists n\in \mathbb{N}\mathrm{such}\mathrm{that}f\left(n\right)>M$.

(3) 

f grows slower than any exponential function, i.e., $\forall a>1,{lim}_{n\to \infty }\frac{f\left(n\right)}{a^n}=0$.

(4) 

For any node $v\in V$, the number of descendants of v at distance n is bounded by $f\left(n\right)$, i.e.,

$$\forall v\in V,\forall n\in \mathbb{N},\left|\right\{u\in V:d(v,u)=n\left\}\right|\le f\left(n\right)$$

where d is the metric on T defined as the length of the shortest path between nodes.

Then T satisfies relative compactness under the metric d.

Proof. 

Let $T=(V,E)$ be the inverse algebraic tree constructed recursively from the analytic inverse function G of a discrete dynamical system $(S,F)$.

Definitions:

• Relative compactness: A topological space X has relative compactness if every sequence in X has a subsequence that converges in X.
• Bolzano-Weierstrass theorem: Every bounded sequence of real numbers has a convergent subsequence.

We will prove that T has relative compactness:

(1)

Let $\left(v_n\right)$ be an arbitrary sequence in V.

(2)

Define $f:V\to \mathbb{R}$ such that $f\left(v\right)$ is the maximum number of nodes in the subtree rooted at v.

(3)

Since by hypothesis there can be no more than K children per node, we have $f\left(v\right)\le M<\infty$ for all $v\in V$. Hence, f is bounded.

(4)

Therefore, $\left(f\left(v_n\right)\right)$ is a bounded sequence in $\mathbb{R}$. By the Bolzano-Weierstrass theorem, it has a subsequence $\left(f\left(v_{n_j}\right)\right)$ that converges to some $L\in \mathbb{R}$.

(5)

Moreover, there exists a subsequence $\left(v_{n_j}\right)$ of $\left(v_n\right)$ such that $f\left(v_{n_j}\right)\to L$.

(6)

Since $d(v_{n_j},v_0)$ is monotonically increasing or decreasing, and bounded (being in $\mathbb{N}$), it converges by the Monotone Convergence Theorem.

(7)

Therefore, $\left(v_{n_j}\right)$ converges in T since T is complete.

(8)

We have shown that every sequence in T has a convergent subsequence. Thus, T has relative compactness.

□

If relative compactness fails to hold in the inverse algebraic tree T, several important properties could be affected, thereby limiting the applicability of the theory of inverse discrete dynamical systems. Here are some properties that might be compromised:

• Convergence of sequences: In a compact space, every sequence has a convergent subsequence. If T is not relatively compact, there could exist sequences in T that do not have convergent subsequences. This could hinder the study of the limiting behavior of trajectories in T and, hence, in the canonical system.
• Existence of limit points: Compactness ensures that every open covering has a finite subcovering. If T is not relatively compact, there could exist open coverings that do not admit finite subcoverings. Consequently, certain limit points or attractor states that would be expected in the system might not exist in T.
• Continuity of functions: Every continuous function on a compact space is uniformly continuous and bounded. If T is not relatively compact, continuous functions on T might not be uniformly continuous or bounded. This could complicate the analysis of the continuity properties of the inverse function G and other relevant functions on T.
• Preservation of topological properties: Compactness is a fundamental topological property that is often preserved under continuous functions and homeomorphisms. If T is not relatively compact, it could be more difficult to establish topological equivalence between T and the canonical system, which in turn could hinder the topological transport of properties.
• Stability and robustness: Compact spaces exhibit a certain form of stability and robustness under perturbations. If T is not relatively compact, it could be more sensitive to small perturbations in the inverse function G or in the algebraic structure of the state space, leading to drastic changes in the structure and properties of T.

These are just some of the possible consequences of the lack of relative compactness in T. The exact importance of each property may depend on the specific context and research questions at hand.

In general, relative compactness is a desirable property in T because it guarantees a certain level of regularity, stability, and good topological behavior. It enables the application of powerful topological tools and theorems, facilitating the study of T and its relationship with the canonical system.

If relative compactness fails to hold, it might be necessary to seek alternative conditions or weaker versions of the theory that still allow for obtaining some of the desired results. This could involve the use of more general notions of compactness, such as sequential compactness, or the imposition of additional constraints on G or the state space to recover some of the lost properties.

In summary, the lack of relative compactness in T could limit the applicability of certain theoretical results and complicate the analysis of the discrete dynamical system. However, it could also motivate the development of more general or alternative versions of the theory, leading to new ideas and research directions.

Lemma 5.10. 

Every inverse algebraic tree $T=(V,E)$ satisfying K-bounded growth for some $K\in \mathbb{N}$ has relative compactness under the metric d.

Proof. 

Let T be an inverse algebraic tree with K-bounded growth. By hypothesis, $\exists K\in \mathbb{N}$ such that $\forall v\in V:\left|Children\right(v\left)\right|\le K$.

Defining $f:V\to \mathbb{R}$ such that $f\left(v\right)$ is the maximum number of nodes in the subtree rooted at v, since by hypothesis there can be at most K children per node, we have:

$$f\left(v\right)\le M<\infty ,\forall v\in V$$

Hence, f is bounded. Therefore, by the Bolzano-Weierstrass theorem, which states that every bounded sequence in $\mathbb{R}$ has a convergent subsequence, it follows that:

• T is totally bounded as it has f bounded.
• By the Heine-Borel Theorem, T is relatively compact.

Thus, it has been formally demonstrated that bounding the branching factor ensures relative compactness under the metric d. □

Theorem 5.11 

(Absence of Anomalous Cycles). Let $(S,F)$ be a discrete dynamical system and $T=(V,E)$ the algebraic inverse tree recursively constructed from the analytical inverse G. Then T does not contain any non-trivial anomalous cycle. That is:

$$\nexists \gamma =\langle v_1,\dots ,v_k\rangle ,k\ge 3:(v_k=v_1\wedge \forall i\in \{1,\dots ,k-1\}:(v_i,v_{i+1})\in E)$$

Proof. 

Let $(S,F)$ be a discrete dynamical system and $T=(V,E)$ be the inverse algebraic tree constructed recursively from the analytic inverse function G. We will prove by contradiction that T does not contain any non-trivial anomalous cycles.

Step 1: Assume, for contradiction, that there exists a non-trivial anomalous cycle $\gamma$ in T.

$$\exists \gamma =\langle v_1,\dots ,v_k\rangle ,k\ge 3:(v_k=v_1\wedge \forall i\in \{1,\dots ,k-1\}:(v_i,v_{i+1})\in E)$$

Step 2: By the recursive construction of T through the injective function G, each node in T has a unique parent.

$$\forall v\in V,\exists !u\in V:(u,v)\in E$$

Step 3: Consider two consecutive nodes $v_i$ and $v_{i+1}$ in the cycle $\gamma$. By Step 2, $v_{i+1}$ has a unique parent in T, which must be $v_i$ according to the cycle’s definition.

Step 4: However, by Step 2, $v_{i+1}$ also has a unique parent in T outside the cycle, as the tree extends infinitely upwards from each node.

Step 5: This leads to a contradiction, as $v_{i+1}$ cannot have two distinct parents in T due to the injectivity of G. More formally:

$$\exists v_j\in V\setminus \{v_1,\dots ,v_k\}:(v_j,v_{i+1})\in E\wedge (v_i,v_{i+1})\in E$$

$$\Rightarrow v_j=v_i(\mathrm{by}\mathrm{injectivity}\mathrm{of}G)$$

$$\Rightarrow v_j\in \{v_1,\dots ,v_k\}\left(\mathrm{contradiction}\right)$$

Step 6: Therefore, the assumption in Step 1 must be false, and there cannot exist a non-trivial anomalous cycle $\gamma$ in T.

$$\nexists \gamma =\langle v_1,\dots ,v_k\rangle ,k\ge 3:(v_k=v_1\wedge \forall i\in \{1,\dots ,k-1\}:(v_i,v_{i+1})\in E)$$

Thus, the absence of non-trivial anomalous cycles in the inverse algebraic tree T is formally proven by contradiction. □

Theorem 5.12 

(Universal Convergence in AIT). Let $T=(V,E)$ be an Algebraic Inverse Tree constructed from a Discrete Dynamical System $(S,F)$ with the analytic inverse function G. Then, for every infinite path $P=(v_1,v_2,\dots )$ in T, P converges to either the root node r or the point of contact $pc$.

Proof. 

Step 1: Define the convergence of a path.

$$\begin{array}{c}\hfill \forall P\in \mathcal{P}\left(T\right),\forall v\in V:\mathrm{Converges}(P,v)\iff \hspace{0.17em}\\ \hfill \forall \epsilon >0,\exists N\in \mathbb{N},\forall n>N:d(v_n,v)<\epsilon \end{array}$$

where $\mathcal{P}\left(T\right)$ is the set of all paths in T, and d is the graph distance in T.

Step 2: Prove that every node has a unique path to either the root or the point of contact.

$$\begin{array}{c}\hfill \forall v\in V,\exists !P\in \mathcal{P}\left(T\right):\hspace{0.17em}\\ \hfill (P=(v,\dots ,r)\vee P=(v,\dots ,pc\left)\right)\wedge \hspace{0.17em}\\ \hfill \forall i\in \{1,\dots ,|P|-1\}:\left(P\right[i],P[i+1\left]\right)\in E\end{array}$$

where $P\left[i\right]$ denotes the i-th node in the path P.

This follows from the recursive construction of T using the injective function G.

Step 3: Prove that every infinite path is a Cauchy sequence.

$$\begin{array}{c}\hfill \forall P=(v_1,v_2,\dots )\in \mathcal{P}\left(T\right):\hspace{0.17em}\\ \hfill \forall \epsilon >0,\exists N\in \mathbb{N},\forall m,n>N:d(v_m,v_n)<\epsilon \end{array}$$

This follows from the monotonically decreasing distances between consecutive nodes in P, due to the unique path property.

Step 4: Prove that T is complete.

$$\begin{array}{c}\hfill \forall {\left(v_n\right)}_{n\in \mathbb{N}}\subseteq V:\hspace{0.17em}\\ \hfill (\forall \epsilon >0,\exists N\in \mathbb{N},\forall m,n>N:d(v_m,v_n)<\epsilon )\hspace{0.17em}\\ \hfill \Rightarrow \exists v\in V:\underset{n\to \infty }{lim}{v}_n=v\end{array}$$

This follows from the finiteness of paths between any node and the root or the point of contact, and the completeness of $\mathbb{R}$ with the usual metric.

Step 5: Conclude that every infinite path converges to either the root or the point of contact.

$$\begin{array}{c}\hfill \forall P=(v_1,v_2,\dots )\in \mathcal{P}\left(T\right):\mathrm{Converges}(P,r)\vee \mathrm{Converges}(P,pc)\end{array}$$

This follows from Steps 3 and 4, as every infinite path is a Cauchy sequence in the complete space T, and thus converges to a unique limit, which must be either the root node r or the point of contact $pc$ by the unique path property.

Therefore, we have proven that every infinite path in the Algebraic Inverse Tree T converges to either the root node r or the point of contact $pc$, considering the impact of the failure of multivalued injectivity at $pc$. □

Theorem 5.13 

(Unique AIT Generation). Let $(S,F)$ be a discrete dynamical system and $G:S\to P\left(S\right)$ its analytic inverse. It is proven that:

If G satisfies:

Injectivity Surjectivity Exhaustiveness Then, the inverse algebraic tree $T=(V,E)$ constructed recursively applying G is unique and satisfies:

Absence of anomalous cycles: $\nexists \gamma$ non-trivial cycle in T Universal convergence of trajectories: $\forall P\in T,\underset{n\to \infty }{lim}P=r$ where r is the root.

Proof. 

Let $(S,F)$ be a discrete dynamical system and $G:S\to S$ its analytic inverse. It is proven that:

• $\forall x,y\in S,G\left(x\right)=G\left(y\right)\Rightarrow x=y$
• $\forall z\in S,\exists x\in S,G\left(x\right)=z$
• $\forall x\in S,\exists n\in \mathbb{N},G^n\left(x\right)=r$

Where r denotes the root node of the inverse algebraic tree $T=(V,E)$ constructed by iterations of G.

Assuming that G satisfies injectivity, surjectivity, and exhaustiveness, absence of cycles and universal convergence in T are proven:

• Absence of anomalous cycles: Suppose $\exists \gamma =(v_1,...,v_k)$, a non-trivial cycle in T. By the injectivity hypothesis, $\forall u,v\in V,G\left(u\right)=G\left(v\right)\Rightarrow u=v$. Taking consecutive nodes $v_i,v_{i+1}$, a contradiction is obtained $\Rightarrow \nexists \gamma$ non-trivial cycle.
• Universal convergence: $\forall x\in S$, by exhaustiveness of G, $\exists n\in \mathbb{N}$ such that $G^n\left(x\right)=r$. That is, $\forall P\in T,{lim}_{n\to \infty }P=r$.

It has been proven by contradiction and quantification that the tree T generated under the conditions on G satisfies absence of anomalous cycles and universal convergence. □

6. Properties of the Inverse Function G in a DIDS

Given that a discrete dynamical system $(S,F)$ is a DIDS if and only if $F:S\to S$ is a deterministic and surjective function, we can derive several important properties of the inverse function $G:S\to \mathcal{P}\left(S\right)$ defined as:

$$G\left(s\right)=\{t\in S:F(t)=s\}$$

Theorem 6.1. 

If $(S,F)$ is a DIDS, then the inverse function $G:S\to \mathcal{P}\left(S\right)$ satisfies the following properties:

(1) 

Injectivity: $\forall a,b\in S,G\left(a\right)=G\left(b\right)\Rightarrow a=b$

(2) 

Surjectivity: $\forall B\in \mathcal{P}\left(S\right),\exists A\in S:G\left(A\right)=B$

(3) 

Exhaustiveness: ${\bigcup }_{s\in S}G\left(s\right)=S$

Proof. 

The proof follows directly from the determinism and surjectivity of F, as demonstrated in Theorem 5.1. □

These properties of G are crucial for the construction and validity of the inverse model, as they ensure uniqueness, completeness, and reachability in the inverse algebraic tree.

6.1. Injectivity of G

The injectivity of G guarantees that each state in the inverse model has a unique corresponding state in the original system, preventing ambiguities or inconsistencies in the transfer of properties.

6.2. Surjectivity of G

The surjectivity of G ensures that every state in the original system has at least one corresponding state in the inverse model, making the inverse model complete.

6.3. Exhaustiveness of G

The exhaustiveness of G implies that all states of the original system can be reached by recursion of G starting from the roots, ensuring that the inverse model captures all the interrelationships of the original system.

7. Constructibility of the Inverse Model

Theorem 7.1 

(Conditions for Inverse Model Constructibility). Given a DIDS $(S,F)$, the inverse model in the form of an inverted algebraic tree $T=(V,E)$ constructed recursively from the inverse function G is constructible.

Proof. 

The constructibility of T follows directly from the injectivity, surjectivity, and exhaustiveness of G, which are guaranteed by the determinism and surjectivity of F. □

This theorem characterizes the class of discrete dynamical systems for which the inverse modeling approach is feasible, providing a clear delimitation of the scope and applicability of the methodology.

8. Uniqueness of the Inverse Model

The injectivity, surjectivity, and exhaustiveness of the inverse function G also ensure the uniqueness of the inverse model, even when dealing with a forest of inverse trees.

Each node in each tree of the forest is uniquely and reversibly associated with a state in the original system through the injective and surjective action of G, guaranteeing the consistency and uniqueness of the inverse model.

9. Decidable Inference and Property Transfer

The injectivity and surjectivity of G establish a discrete homeomorphism between the state space of the original system and the set of nodes of the inverse algebraic tree, enabling the decidable and complete transfer of properties between the inverse model and the original system.

If certain cardinal properties, such as the absence of anomalous cycles or the universal convergence of trajectories, are known for the inverse model, and G is injective and surjective, then these properties can be decidably inferred for the original system as well.

Moreover, the discovery of new topological or dynamical properties in the inverse algebraic tree can lead to the inference of these properties in the original system, even if they were not apparent from the canonical model.

10. Convergence in DIDS

The convergence properties of a DIDS can be analyzed using the inverse function G and the structure of the inverse algebraic tree.

10.1. Finite Case

Theorem 10.1. 

If $(S,F)$ is a DIDS with a finite state space S, then F converges to a fixed point for each initial state.

Proof. 

The proof follows from the injectivity, surjectivity, and exhaustiveness of G, which guarantee that any sequence of states generated by F must eventually reach a fixed point, as there can be no non-trivial cycles in the finite state space. □

10.2. Countably Infinite Case

Theorem 10.2. 

If $(S,F)$ is a DIDS with a countably infinite state space S, then for each initial state $s\in S$, either:

(1) 

F converges to a fixed point starting from s, or

(2) 

F enters a cycle starting from s.

Proof. 

The proof relies on the injectivity and exhaustiveness of G, which ensure that any sequence of states generated by F must either reach a fixed point or enter a cycle, as there can be no infinite non-repeating sequences in the inverse model. □

Remark 3. 

The injectivity, surjectivity, and exhaustiveness of G, while powerful conditions, are not sufficient on their own to guarantee the convergence of F to a unique fixed point or cycle in the countably infinite case. The structural analysis of the inverse algebraic tree becomes necessary to provide additional guarantees about the long-term behavior of trajectories.

11. Relationship between the Properties of F and G

The determinism and surjectivity of F in a DIDS imply several important properties of the inverse function G, and vice versa.

These theorems establish a clear connection between the properties of F and G in a DIDS, highlighting the fundamental role of determinism and surjectivity in the inverse modeling approach.

12. Conclusion

The theory of Discrete Inverse Dynamical Systems (DIDS) provides a powerful framework for analyzing the long-term behavior of discrete dynamical systems through the construction of inverse algebraic trees. The determinism and surjectivity of the evolution function F are sufficient conditions for a system to be a DIDS, and they imply several important properties of the inverse function G, such as injectivity, surjectivity, and exhaustiveness.

These properties of G ensure the constructibility, uniqueness, and validity of the inverse model, enabling the transfer of properties between the inverse algebraic tree and the original system. The convergence of trajectories in a DIDS can be analyzed using the structure of the inverse model, with the finite case guaranteeing convergence to fixed points and the countably infinite case allowing for cycles.

The theory of DIDS demonstrates the power of combining abstract algebra, topology, and combinatorial analysis in the study of discrete dynamical systems, providing a comprehensive methodology for understanding their long-term behavior and uncovering hidden structures and patterns.

13. Topological Equivalences

Note 2 

(On the Necessity of the Topological Framework). The extensive use of topological concepts and definitions in this article might seem to overshadow the primary focus on solving the Collatz Conjecture using the Theory of Inverse Discrete Dynamical Systems (TIDDS). However, it is crucial to emphasize that the topological framework is not merely a collection of abstract definitions but an indispensable foundation for the development and application of TIDDS.

The Collatz Conjecture, a long-standing open problem in number theory and discrete dynamical systems, has resisted various attempts at resolution due to its intricate structure and the complex behavior of the Collatz function. The key insight behind TIDDS is to approach the problem from an inverse perspective, constructing an algebraic model that encodes the backward dynamics of the system.

To rigorously define and analyze these inverse algebraic structures, it is essential to employ the language and tools of topology. Concepts such as topological spaces, continuity, compactness, and homeomorphisms provide the necessary framework to formalize the relationships between the original discrete system and its inverse model.

More specifically, the topological properties of the inverse algebraic trees, such as their compactness and the absence of non-trivial cycles, play a crucial role in establishing the convergence of all Collatz sequences to the trivial cycle. These properties, which are central to the resolution of the Collatz Conjecture, are inherently topological in nature and cannot be properly formulated or proven without the underlying topological framework.

Furthermore, the use of topological notions allows for the rigorous definition of conjugacy and topological invariance, which are essential for transferring properties from the inverse model back to the original system. The homeomorphic equivalence between the inverse algebraic trees and the Collatz system enables the application of powerful topological results, such as the Homeomorphic Transport Theorem, to establish the convergence of all trajectories in the original system.

In summary, the topological framework is not an arbitrary collection of definitions but a vital and integral component of the TIDDS approach to solving the Collatz Conjecture. Without the solid topological foundation, it would be impossible to rigorously define, analyze, and exploit the inverse algebraic structures that lie at the heart of this groundbreaking proof. The extensive topological content in this article is therefore not a distraction but a necessary and indispensable part of the overall argument.

After constructing the inverse model of a discrete dynamical system using an algebraic tree following the reversed analytical recursion, and having demonstrated its cardinal structural properties, the next step in the methodology consists of establishing formal topological equivalences between this inverted model and the original canonical system.

To do this, a homeomorphism is defined, that is, a bijective and bicontinuous mapping, between the nodes of the algebraic inverse tree and the discrete states of the canonical system. This correlation is demonstrated to satisfy the conditions of being a bijective and continuous function in both directions.

With this critical element, the conditions are given to demonstrate the topological equivalence between both dynamic systems, discrete and inverse, with the relevant natural topologies in each case. Again, the continuity of the homeomorphism, along with the topological attributes already demonstrated on the model such as compactness and metric completeness, allow completing the sought equivalence proof.

Finally, as a consequence of this equivalence induced by the homeomorphism, various relevant properties demonstrated for one system also hold for the other. In this scenario, the previously mentioned topological transport becomes possible between the canonical system and its inversely modeled counterpart, analytically transferring the previously demonstrated structural properties.

Thus, the long-awaited Topological Equivalence is finally obtained as the culmination point in the process of modeling, analysis, and inferential transport to solve open problems on discrete dynamics through the revolutionary inverse theoretical approach fully exposed here.

Definition 13.1. 

Let $(X,\tau )$ be a topological space, where X is a set and τ is a topology on X, i.e., τ satisfies:

(1) 

$\varnothing ,X\in \tau$

(2) 

The union of elements of τ belongs to τ

(3) 

The finite intersection of elements of τ belongs to τ

Formally, a topological space is an ordered pair $(X,\tau )$ where X is a set and τ is a family of subsets of X satisfying the above properties.

13.1. Cardinal Properties of Algebraic Inverse Trees

Definition 13.2 

(Continuity). Let $(X,{\tau }_X)$ and $(Y,{\tau }_Y)$ be topological spaces. A function $f:X\to Y$ is continuous at a point $x_0\in X$ if for every open set $V\in {\tau }_Y$ containing $f\left(x_0\right)$, there exists an open set $U\in {\tau }_X$ containing $x_0$ such that $f\left(U\right)\subseteq V$.

Formally, we can express this using first-order logic as:

$$\begin{array}{c}\hfill \forall V\in {\tau }_Y(f\left(x_0\right)\in V\to \exists U\in {\tau }_X(x_0\in U\wedge f\left(U\right)\subseteq V))\end{array}$$

The function f is said to be continuous on X if it is continuous at every point $x\in X$. In other words:

$$\begin{array}{c}\hfill \forall x\in X,\forall V\in {\tau }_Y(f\left(x\right)\in V\to \exists U\in {\tau }_X(x\in U\wedge f\left(U\right)\subseteq V))\end{array}$$

Definition 13.3 

(Compact Space). A topological space $(X,\tau )$ is said to be compact if for every family of open sets $U_{\alpha }\alpha \in A$ that cover X, there exists a finite subfamily $U{\alpha }_1,\dots ,U_{{\alpha }_n}$ that also covers X.

Definition 13.4 

(Metric Space). A metric space is an ordered pair $(M,d)$ where:

• M is a non-empty set
• d is a metric on M

satisfying the metric axioms d:

(1) 

$d(x,y)\ge 0$, with $d(x,y)=0\iff x=y$

(2) 

$d(x,y)=d(y,x)$

(3) 

$d(x,z)\le d(x,y)+d(y,z)$

Definition 13.5 

(Metric on Algebraic Inverse Tree). Let $T=(V,E)$ be an Algebraic Inverse Tree (AIT). We define the metric $d:V\times V\to \mathbb{R}$ as follows:

$$\begin{array}{c}\hfill \forall a,b\in V:d(a,b)=\left\{\begin{array}{cc}0\hfill & \mathit{if}a=b\hfill \\ min\{n\ge 1:\exists (v_0,v_1,\dots ,v_n)\in V^{n+1},\hfill \\ (v_i,v_{i+1})\in E;\forall i\in \{0,\dots ,n-1\},\hfill \\ v_0=a,v_n=b\}\hfill & \mathit{if}a\ne b\hfill \end{array}\right.\end{array}$$

In other words, $d(a,b)$ is the length of the shortest path from a to b in T.

Theorem 13.1. 

The function d defined above is a metric on V.

Proof. 

We will verify that d satisfies the axioms of a metric:

(1)

Non-negativity: $\forall a,b\in V,d(a,b)\ge 0$.

Proof. By definition, $d(a,b)$ is 0 or the length of a path, which is always a non-negative number. □

(2)

Indiscernibles identity: $\forall a,b\in V,d(a,b)=0\iff a=b$.

Proof. (⇒) If $d(a,b)=0$, then by definition, $a=b$.

(⇐) If $a=b$, then by definition, $d(a,b)=0$. □

(3)

Symmetry: $\forall a,b\in V,d(a,b)=d(b,a)$.

Proof. Let $(v_0,v_1,\dots ,v_n)$ be the shortest path from a to b. Then $(v_n,v_{n-1},\dots ,v_0)$ is a path from b to a of the same length. Therefore, $d(a,b)=d(b,a)$. □

(4)

Triangular inequality: $\forall a,b,c\in V,d(a,c)\le d(a,b)+d(b,c)$.

Proof. Let $(v_0,v_1,\dots ,v_n)$ and $(w_0,w_1,\dots ,w_m)$ be the shortest paths from a to b and from b to c, respectively. Then $(v_0,v_1,\dots ,v_n,w_1,\dots ,w_m)$ is a path (not necessarily the shortest) from a to c. Therefore, $d(a,c)\le n+m=d(a,b)+d(b,c)$. □

Therefore, d is a metric on V.

Theorem 13.2 

(Compactness). Let $(T,d)$ be the metric space associated with an inverted discrete dynamical system modeled as an Inverse Algebraic Tree T. Then $(T,d)$ is a compact metric space.

Proof. 

Let $(T,d)$ be the metric space associated with an inverted discrete dynamical system modeled as an Inverse Algebraic Tree T. We aim to prove that $(T,d)$ is compact.

Suppose, for the sake of contradiction, that $(T,d)$ is not compact. This implies the existence of an open cover $\mathcal{U}$ of T such that no finite subcollection of $\mathcal{U}$ covers T.

Consider the open balls $B_{\epsilon }\left(v_k\right)$ of radius $\epsilon$ centered at the nodes $v_k\in T$. Since T is finite, there exists a finite number of balls that cover T. Let $\mathcal{V}\subseteq \mathcal{U}$ be this finite subcover.

Now, consider any closed and bounded subset $K\subseteq T$. By the Heine-Borel Theorem, a subset of a Euclidean space is compact if and only if it is closed and bounded. Since K is contained within a closed ball of finite radius that only contains a few points, it follows that K is finite and thus compact.

Therefore, $(T,d)$ is compact, contradicting our initial assumption. Thus, the supposition that $(T,d)$ is not compact must be false. □

Theorem 13.3 

(Connectedness). Let $(T,d)$ be the metric space associated with an inverted discrete dynamical system modeled as an Algebraic Inverse Tree. Then $(T,d)$ is connected, it cannot be expressed as the union of two disjoint non-empty subsets.

Proof. 

Suppose by contradiction that $(T,d)$ is not connected.

Then there would exist $A,B\subset T$ disjoint and non-empty such that $A\cup B=T$ and $A\cap B=\varnothing$.

Taking $a\in A$ and $b\in B$, by uniqueness of paths in $(T,d)$ there exists a unique path from a to b.

But since $a\in A$ and $b\in B$ with $A,B$ open by hypothesis, by the Connectivity Lemma, there should exist c on this path such that $c\notin A\cup B$, contradicting $A\cup B=T$.

We arrive at a contradiction assuming that $(T,d)$ was not connected.

By contradiction, we conclude that $(T,d)$ is connected, completing the proof. □

Definition 13.6. 

Let $(X,d)$ be a metric space. A sequence $\left(x_n\right)$ in X is called aCauchy sequenceif:

$$\forall \epsilon >0:\exists N\in \mathbb{N}:\forall n,m\ge N:d(x_n,x_m)<\epsilon$$

Definition 13.7. 

Let $(X,d)$ be a metric space. $(X,d)$ is said to becompleteif every Cauchy sequence $x_n$ in X converges to some point $x\in X$. In other words:

$$\forall \left(x_n\right)\subseteq X,\left(x_n\right)\mathrm{is}\mathrm{Cauchy}\Rightarrow \exists x\in X:\underset{n\to \infty }{lim}{x}_n=x$$

Lemma 13.4 

(Infinite Paths as Cauchy Sequences). Let $(S,F)$ be a discrete dynamical system modeled by a metric space $(X,d_X)$. Let T be the associated algebraic inverse tree constructed recursively from the inverse function $G:X\to \mathcal{P}\left(X\right)$. Let $(P=(x_1,x_2,\dots ))$ be an arbitrary infinite path in T.

Then, P is a Cauchy sequence in $(X,d_X)$.

Proof. 

Let the metric $d:T\times T\to \mathbb{R}$ be defined on nodes of T such that $d(u,v)$ equals the length of the unique path between nodes u and v in T. This path length determines the distance function $d_T$.

Step 1: Formalize the universal convergence in T.

$$\begin{array}{cc}\hfill & \forall \epsilon >0,\exists N\in \mathbb{N},\forall n\ge N:d_T(x_n,r)<\epsilon \hfill \\ \hfill & \mathrm{where}r\mathrm{is}\mathrm{the}\mathrm{root}\mathrm{node}\mathrm{of}T.\hfill \end{array}$$

Step 2: Prove that the convergence in T implies a Cauchy condition in X.

$$\begin{array}{cc}\hfill d_T(x_n,r)<\epsilon & \Rightarrow d_T(x_n,x_m)<2\epsilon \hfill \\ \hfill & \mathrm{for}\mathrm{infinitely}\mathrm{many}n,m\ge N(\mathrm{taking}\epsilon =1).\hfill \end{array}$$

Step 3: Use the triangle inequality on $d_X$ to show the Cauchy condition.

$$\begin{array}{cc}\hfill d_X(x_n,x_m)& \le d_X(x_n,r)+d_X(r,x_m)\hfill \\ \hfill & <2\epsilon \hfill \end{array}$$

Therefore, P satisfies the Cauchy condition in $(X,d_X)$, proving the lemma. □

13.2. Other Cardinal Properties of the Inverse Tree

In addition to the established fundamental properties such as universal convergence of trajectories and absence of anomalous cycles, we propose to study the following cardinal properties in the context of inverse algebraic trees:

Definition 13.8 

(Stability). Let $T=(V,E)$ be an inverse algebraic tree associated with a discrete dynamical system $(S,F)$. We say that T isstableif for any $\epsilon >0$, there exists a $\delta >0$ such that for any pair of nodes $u,v\in V$, if $d(u,v)<\delta$, then $d\left(F\right(u),F(v\left)\right)<\epsilon$, where d is an appropriate metric on V.

Intuitively, stability implies that small perturbations in the initial states do not significantly alter the global structure of the tree or affect convergence towards the root node.

Definition 13.9 

(Robustness). Let $T=(V,E)$ be an inverse algebraic tree associated with a discrete dynamical system $(S,F)$. We say that T is robust if for any perturbation $p:S\to S$ in the original system, there exists a homeomorphism $h:T\to T^{\prime }$ such that $T^{\prime }$ is the inverse algebraic tree associated with the perturbed system $(S,F\circ p)$.

Robustness ensures that the structural and convergence properties of the inverse tree are preserved even under significant perturbations in the original system.

Definition 13.10 

(Carrying Capacity). Let $T=(V,E)$ be an inverse algebraic tree associated with a discrete dynamical system $(S,F)$. The carrying capacity of T, denoted $CC\left(T\right)$, is defined as the maximum size of the state space $\left|S\right|$ for which the construction of T remains computationally tractable.

Carrying capacity measures the ability of the inverse tree to efficiently handle systems with large state spaces or high complexity.

Definition 13.11 

(Adaptability). Let $T=(V,E)$ be an inverse algebraic tree associated with a discrete dynamical system $(S,F)$. We say that T is adaptable if for any continuous change in the parameters of the original system resulting in a family of systems $(S,F_t)$, there exists a continuous family of homeomorphisms $h_t:T\to T_t$ such that $T_t$ is the inverse algebraic tree associated with $(S,F_t)$.

Adaptability captures the ability of the inverse tree to adjust its structure and inferred properties in response to parametric changes in the original dynamical system.

These new cardinal properties (stability, robustness, carrying capacity, and adaptability) expand the scope and applicability of the theory of inverse discrete dynamical systems. Future research could focus on developing specific metrics, demonstrating the preservation of these properties under homeomorphisms, and analyzing their impact on the computational scalability of the methodology. Incorporating these notions enriches and strengthens the theoretical foundations of this innovative approach to modeling and analyzing complex systems.

13.3. Conditions for Topological Transportability

Theorem 13.5 

(Topological and Metric Conditions for Transportability). Let $(X,F)$ be a discrete dynamical system, and let $T=(V,E)$ be its inverse algebraic tree generated by the inverse analytic function $G:X\to \mathcal{P}\left(X\right)$. If T satisfies the following properties:

(1) 

Relative compactness

(2) 

Connectivity

(3) 

Relative metric completeness

then the topological properties demonstrated in T can be transported to the original system $(X,F)$ through a homeomorphic equivalence.

Proof. 

Suppose the inverse algebraic tree T associated with $(X,F)$ satisfies the enumerated properties:

(1)

Due to relative compactness, T exhibits good limit and convergence properties, necessary for preserving the topological structure under homeomorphisms.

(2)

By connectivity, T maintains its topological coherence, avoiding undesired disconnections that would hinder a homeomorphic correspondence with $(X,F)$.

(3)

Through relative metric completeness, T ensures the convergence of Cauchy sequences, an invariant property under homeomorphisms and essential for preserving the metric structure.

These topological and metric properties of T, being invariant under homeomorphisms, allow establishing a topological equivalence with the original system $(X,F)$. This ensures that the properties demonstrated in T remain valid in $(X,F)$.

Conversely, if any of these properties fails in T, a homeomorphic correspondence with $(X,F)$ cannot be assured, and therefore, the transport of properties would not be guaranteed. □

Theorem 13.6. 

Let $F:S\to S$ be a function and $G:S\to \mathcal{P}\left(S\right)$ be its inverse function. If F is deterministic and surjective, then G is guaranteed to be the analytic inverse of F.

Proof. 

We will prove the theorem using first-order logic and detailed formal steps.

Step 1: Formalize the determinism of F.

$$\forall s\in S,\exists !t\in S:F\left(s\right)=t$$

Step 2: Formalize the surjectivity of F.

$$\forall t\in S,\exists s\in S:F\left(s\right)=t$$

Step 3: Define the inverse function G.

$$\forall s\in S:G\left(s\right)=\{t\in S:F(t)=s\}$$

Step 4: Prove that G is multivalued injective.

$$\forall a,b\in S:(a\ne b\to G(a)\cap G(b)=\varnothing )$$

Proof: Suppose $a,b\in S$ with $a\ne b$. Let $t\in G\left(a\right)\cap G\left(b\right)$. Then $F\left(t\right)=a$ and $F\left(t\right)=b$, contradicting the determinism of F. Therefore, $G\left(a\right)\cap G\left(b\right)=\varnothing$.

Step 5: Prove that G is surjective.

$$\forall B\in \mathcal{P}\left(S\right),\exists A\in S:G\left(A\right)=B$$

Proof: Let $B\in \mathcal{P}\left(S\right)$. By the surjectivity of F, for each $s\in B$, there exists $t\in S$ such that $F\left(t\right)=s$. Let $A=\{t\in S:F(t)\in B\}$. Then $G\left(A\right)=B$.

Step 6: Prove that G is exhaustive.

$$\forall s\in S,\exists n\in \mathbb{N}:s\in G^n\left(F\left(s\right)\right)$$

Proof: Let $s\in S$. By the surjectivity of F, there exists $t\in S$ such that $F\left(t\right)=s$. Therefore, $s\in G\left(F\right(s\left)\right)$, and so $s\in G^1\left(F\left(s\right)\right)$.

Conclusion: By steps 4, 5, and 6, we have shown that if F is deterministic and surjective, then its inverse function G is multivalued injective, surjective, and exhaustive. Therefore, G is guaranteed to be the analytic inverse of F. □

Theorem 13.7 

(Conditions for Property Transfer). Let $(S,F)$ be a discrete dynamical system, and let $T=(V,E)$ be its inverse algebraic tree generated by the inverse analytic function $G:S\to \mathcal{P}\left(S\right)$. Properties demonstrated in T can be transferred to $(S,F)$ if:

(1) 

G is multivalued injective: $\forall s_1,s_2\in S:(s_1\ne s_2\to G\left(s_1\right)\cap G\left(s_2\right)=\varnothing )$.

(2) 

G is surjective: $\forall s\in S,\exists t\in S:s\in G\left(t\right)$.

(3) 

G is exhaustive: $\forall s\in S,\exists n\in \mathbb{N}:G^n\left(s\right)=r$ where r is a root of T.

(4) 

The properties are topological and invariant under homeomorphisms.

Proof. 

Assume conditions 1-4 hold. We prove that a property P demonstrated in T can be transferred to $(S,F)$.

Step 1: Prove that T is a well-defined inverse model of $(S,F)$.

$$\begin{array}{cc}\hfill \mathrm{By}\mathrm{condition}1,& \forall s_1,s_2\in S:(s_1\ne s_2\to G\left(s_1\right)\cap G\left(s_2\right)=\varnothing ).\hfill \\ \hfill \mathrm{By}\mathrm{condition}2,& \forall s\in S,\exists t\in S:s\in G\left(t\right).\hfill \\ \hfill \mathrm{By}\mathrm{condition}3,& \forall s\in S,\exists n\in \mathbb{N}:G^n\left(s\right)=r\mathrm{where}r\mathrm{is}\mathrm{a}\mathrm{root}\mathrm{of}T.\hfill \end{array}$$

These conditions ensure that T is a well-defined inverse model of $(S,F)$.

Step 2: Prove that there exists a homeomorphism between T and $(S,F)$.

$$\begin{array}{cc}\hfill \mathrm{Define}f:V& \to S\mathrm{by}f\left(v\right)=s\mathrm{if}v\mathrm{represents}\mathrm{state}s\mathrm{in}T.\hfill \\ \hfill \mathrm{By}\mathrm{the}\mathrm{construction}\mathrm{of}T,& f\mathrm{is}\mathrm{bijective}.\hfill \\ \hfill \mathrm{By}\mathrm{the}\mathrm{topology}\mathrm{on}T\mathrm{and}S,& f\mathrm{is}\mathrm{continuous}.\hfill \\ \hfill \mathrm{Therefore},& f\mathrm{is}\mathrm{a}\mathrm{homeomorphism}\mathrm{between}T\mathrm{and}(S,F).\hfill \end{array}$$

Step 3: Prove that P can be transferred from T to $(S,F)$.

$$\begin{array}{cc}\hfill \mathrm{Assume}& P\left(T\right).\hfill \\ \hfill \mathrm{By}\mathrm{condition}4,& P\mathrm{is}\mathrm{topological}\mathrm{and}\mathrm{invariant}\mathrm{under}\mathrm{homeomorphisms}.\hfill \\ \hfill \mathrm{By}\mathrm{Step}2,& \exists \mathrm{a}\mathrm{homeomorphism}f:T\to (S,F).\hfill \\ \hfill \mathrm{Therefore},& P(S,F).\hfill \end{array}$$

Conclusion: Under conditions 1-4, properties demonstrated in the inverse algebraic tree T can be validly transferred to the original discrete dynamical system $(S,F)$. □

13.4. Homeomorphism between Spaces

Definition 13.12 

(Discrete Topology). Let S be the discrete space on which a discrete dynamical system is defined. The discrete topology on S is defined as:

$$\tau =\{\varnothing ,\left\{x_1\right\},\left\{x_2\right\},\dots \}$$

where $x_i\in S$ and each element of S defines an open and closed set (a singleton).

The axioms satisfied by τ are:

(1) 

$\varnothing ,S\in \tau$

(2) 

The union of elements of τ belongs to τ

(3) 

The finite intersection of elements of τ belongs to τ

In other words, τ constitutes a discrete topology on S, whose open sets are all subsets, and whose closed sets are the complements of open sets. A basis for τ is given by the singletons, and a sub-basis consists of the elements of S itself.

It is then said that $(S,\tau )$ is a discrete topological space relevant to the system.

Example 1. 

A simple example of a topological space is the set $X=\{a,b,c\}$ with the topology $T=\{\varnothing ,\{a\},\{b\},\{a,b\},\{a,b,c\left\}\right\}$.

Definition 13.13. 

Let $(S,\tau )$ be the topological space associated with the canonical discrete dynamical system, where S is the set of discrete states and τ is the standard discrete topology.

Let $(T,\rho )$ be the topological space associated with the inverse algebraic model, where $T=(V,E)$ is the inverse algebraic tree with node set V, edges E, and ρ is the natural topology on T.

We define a map $f:(T,\rho )\to (S,\tau )$ as follows:

For all $v\in V$, there exists a unique $s\in S$ such that $f\left(v\right)=s$.

In other words, f bijectively correlates each node v of the tree T with a unique state s of the canonical system S.

Definition 13.14 

(Homeomorphism). Let $(X,\tau )$ and $(Y,\sigma )$ be topological spaces associated with a canonical discrete system S and its inverse algebraic model T respectively. A homeomorphism is a bijective and bicontinuous function $f:(X,\tau )\to (Y,\sigma )$.

(1) 

f is bijective

(2) 

Both f and $f^{-1}$ are continuous

Example 2. 

The open intervals $(0,1)$ and $(a,b)$ in the real line with the usual topology are homeomorphic. A bijective and continuous function between them is $f\left(x\right)=a+(b-a)x$, and its inverse $f^{-1}\left(y\right)=\frac{y-a}{b-a}$ is also continuous.

Example 3. 

Consider the topological spaces $(X,T_X)$ and $(Y,T_Y)$, where $X=\{1,2,3\}$ with the discrete topology (all subsets of X are open) and $Y=\{a,b\}$ with the trivial topology (only the empty set and Y are open). The function $f:X\to Y$ defined by $f\left(1\right)=a$, $f\left(2\right)=b$, and $f\left(3\right)=a$ is continuous, since the preimage of any open set in Y is an open set in X.

Theorem 13.8. 

The map $f:(T,\rho )\to (S,\tau )$ defined above is a homeomorphism between the topological spaces associated with the inverse algebraic model and the canonical discrete dynamical system.

Proof. 

Let $f:(T,\rho )\to (S,\tau )$ be the function that bijectively correlates nodes of the algebraic inverse tree T with states of the canonical system S. We aim to show that f is a homeomorphism.

First, we prove that f is bijective. Injectivity follows from the fact that each node in T represents a unique state in S, and surjectivity is ensured by the exhaustive construction of T using the inverse function G.

Next, we show that f and $f^{-1}$ are continuous. To prove continuity, we use the following equivalent definitions:

• f is continuous if and only if for every open set U in $(S,\tau )$, the preimage $f^{-1}\left(U\right)$ is open in $(T,\rho )$.
• f is continuous if and only if for every convergent sequence $\left(x_n\right)\to x$ in $(T,\rho )$, the sequence $\left(f\left(x_n\right)\right)\to f\left(x\right)$ in $(S,\tau )$.

Let U be an open set in $(S,\tau )$. By the definition of the discrete topology, every subset of S is open. Thus, $f^{-1}\left(U\right)$ is a union of nodes in T, which is open in the natural topology $\rho$. Therefore, f is continuous.

Similarly, let $\left(x_n\right)$ be a convergent sequence in $(T,\rho )$ with $x_n\to x$. Since T is discrete, convergence implies that $x_n=x$ for all but finitely many n. Thus, $f\left(x_n\right)=f\left(x\right)$ for all but finitely many n, implying that $\left(f\left(x_n\right)\right)$ converges to $f\left(x\right)$ in $(S,\tau )$. Therefore, f is continuous.

The continuity of $f^{-1}$ can be shown using similar arguments. □

Definition 13.15. 

Let $(S,\tau )$ be the topological space associated with the canonical discrete dynamical system, where S is the set of discrete states and τ is the standard discrete topology.

Let $(T,\rho )$ be the topological space associated with the inverse algebraic model, where $T=(V,E)$ is the inverse algebraic tree with node set V, edges E, and ρ is the natural topology on T.

We define a function $f:(T,\rho )\to (S,\tau )$ as follows:

$$\forall v\in V,\exists !s\in S\mathrm{such}\mathrm{that}f\left(v\right)=s$$

In other words, f bijectively correlates each node v of the tree T with a unique state s of the canonical system S.

Theorem 13.9. 

The function $f:(T,\rho )\to (S,\tau )$ defined above is a homeomorphism between the topological spaces associated with the inverse algebraic model and the canonical discrete dynamical system.

Proof. 

First, let’s prove that f is bijective:

Injectivity: Let $v_1,v_2\in V$ such that $v_1\ne v_2$. By the recursive construction of T using G, $v_1$ and $v_2$ represent different states in S. Therefore, $f\left(v_1\right)\ne f\left(v_2\right)$, implying that f is injective.

Surjectivity: Let $s\in S$. By the exhaustive construction of T using the inverse function G, there exists a sequence of states leading to s in the discrete dynamical system. This sequence is represented by a path in T ending at a node v with $f\left(v\right)=s$. Therefore, $\forall s\in S,\exists v\in V$ such that $f\left(v\right)=s$, implying that f is surjective.

Now, let’s show that both f and $f^{-1}$ are continuous:

Continuity of f: Let $U\subseteq S$ be open in $\tau$. We want to show that $f^{-1}\left(U\right)$ is open in $\rho$. By the definition of $\tau$, $U={\bigcup }_{s\in U}\left\{s\right\}$. Then:

$$f^{-1}\left(U\right)=f^{-1}\left(\bigcup _{s\in U}\left\{s\right\}\right)=\bigcup _{s\in U}{f}^{-1}\left(\left\{s\right\}\right)=\bigcup _{s\in U}\{v\in V:f\left(v\right)=s\}$$

Each set $\{v\in V:f(v)=s\}$ is open in $\rho$ as it corresponds to a unique node. Therefore, $f^{-1}\left(U\right)$ is open in $\rho$ as a union of open sets, implying that f is continuous.

Continuity of $f^{-1}$: Let $W\subseteq V$ be open in $\rho$. We want to show that $f\left(W\right)$ is open in $\tau$. By the definition of $\rho$, $W={\bigcup }_{v\in W}\left\{v\right\}$. Then:

$$f\left(W\right)=f\left(\bigcup _{v\in W}\left\{v\right\}\right)=\bigcup _{v\in W}f\left(\left\{v\right\}\right)=\bigcup _{v\in W}\left\{f\left(v\right)\right\}$$

Each singleton set $\left\{f\right(v\left)\right\}$ is open in $\tau$ by definition. Therefore, $f\left(W\right)$ is open in $\tau$ as a union of open sets, implying that $f^{-1}$ is continuous.

We have shown that f is bijective and both f and $f^{-1}$ are continuous. Therefore, f is a homeomorphism between the topological spaces $(T,\rho )$ and $(S,\tau )$. □

Theorem 13.10 

(Homeomorphic Invariance). Let $f:(X,\tau )\to (Y,\sigma )$ be a homeomorphism, and let P be a topological property on X invariant under f. Then:

$$P\left(X\right)\leftrightarrow P\left(Y\right)$$

In other words, the property is preserved in the transformed space.

Proof. 

Let $(X,\tau )$ and $(Y,\sigma )$ be topological spaces. Let $f:(X,\tau )\to (Y,\sigma )$ be a homeomorphism between X and Y. Let P be a topological property on X that is invariant under f.

We will demonstrate that P is preserved in the transformed space Y through f. That is:

$P\left(X\right)\leftrightarrow P\left(Y\right)$

To show invariance, we explicitly prove that if P holds on X, then P also holds on Y:

(→): Let $x\in X$ such that $P\left(x\right)$ holds. Since P is a topological property and f is a homeomorphism, then:

$P\left(x\right)\stackrel{f}{\to }P\left(f\left(x\right)\right)$

That is, if x satisfies P, then $y=f\left(x\right)$ also satisfies it. Applying this to all $x\in X$, it follows from $P\left(X\right)$ that $P\left(f\right(X\left)\right)=P\left(Y\right)$, by explicit action of f.

(←): The reverse direction follows similarly by using continuity of $f^{-1}$...

Thus P is explicitly demonstrated invariant under the bijective and bicontinuous mapping f. □

Theorem 13.11 

(Properties). Every homeomorphism f satisfies:

(1) 

Preserves subspaces

(2) 

Preserves compactness

(3) 

Preserves connectedness

(4) 

Preserves metric completeness

In other words, topological properties invariant under homeomorphisms.

Proof. 

Let $f:(X,\tau )\to (Y,\sigma )$ be a homeomorphism between topological spaces X and Y.

(1)

Subspaces: Let $A\subseteq X$ be a subspace of X. Since f is bijective, $f\left(A\right)\subseteq Y$ is a subspace of Y. Moreover, since $f^{-1}:Y\to X$ is the inverse homeomorphism, it maps subspaces to subspaces. Specifically, $f^{-1}\left(f\left(A\right)\right)=A$. Thus f and $f^{-1}$ preserve subspaces under their mapping actions.

(2)

Compactness: Suppose $(X,\tau )$ is a compact topological space. Thus every open cover $\mathcal{U}=U_{\alpha }$ of X has a finite subcover ${\mathcal{U}}^{\prime }=U_{{\alpha }_1},...,U_{{\alpha }_n}$ that also covers X. Since f is continuous as a homeomorphism, it maps open sets to open sets. Therefore, $\mathcal{V}=V_{\beta }=f\left(U_{\alpha }\right)$ is an open cover of Y. Applying $f^{-1}$, which is also continuous, gives the open subcover ${\mathcal{V}}^{\prime }=f^{-1}\left(V_{{\beta }_1}\right),...,f^{-1}\left(V_{{\beta }_m}\right)$ of X. But ${\mathcal{V}}^{\prime }={\mathcal{U}}^{\prime }$. Thus there exists a finite subcover of $\mathcal{V}$, implying Y is compact.

(3)

Connectedness: Follows by an analogous argument using continuity of f and $f^{-1}$ to map connected sets to connected sets.

(4)

Metric completeness: If $(X,d_X)$ is metrically complete, Cauchy sequences converge. Applying continuous f maps Cauchy sequences to Cauchy sequences, which will converge in the complete space $(Y,d_Y)$. Hence $(Y,d_Y)$ is complete.

Therefore, f preserves all these topological properties. □

Theorem 13.12. 

The function $f:T\to S$ correlating the algebraic inverse tree T with the discrete dynamical system S is injective.

Proof. 

Let $f:T\to S$ be the function bijectively correlating nodes of the algebraic inverse tree T constructed from the analytic inverse function G with states of the discrete system S. Since G is injective by definition, for any pair of distinct nodes $x,y\in T$, $G\left(f\right(x\left)\right)\ne G\left(f\right(y\left)\right)$. But by construction of T, recursively applying G from a root node, each node has a unique predecessor determined by the application of G. Thus, if we had $f\left(x\right)=f\left(y\right)$ for some pair $x\ne y$, it would lead to a contradiction with the uniqueness of the predecessor given by G. Therefore, it must be that if $f\left(x\right)=f\left(y\right)$ then necessarily $x=y$. It is concluded that f is injective. □

Theorem 13.13. 

The function $f:T\to S$ correlating the algebraic inverse tree T with the discrete dynamical system S is surjective.

Proof. 

Again, let $f:T\to S$ be the function correlating nodes of the inverse tree T with states of S. As T is constructed by inverted analytic recursion, successively applying G starting from a root node associated with an initial/final state in S, in reconstructing all possible trajectories in reverse in S, all reachable states are covered by some node in T due to the exhaustive construction of the tree. Formally, given any state $s\in S$, there exists some possible inverted trajectory in S ending in s, which is represented in T, implying the existence of some node $x\in T$ such that $f\left(x\right)=s$. Hence f is surjective. □

Theorem 13.14. 

The function $f:T\to S$ correlating the algebraic inverse tree T with the discrete dynamical system S is bijective.

Proof. 

Having demonstrated both injectivity and surjectivity of the function f, it is directly concluded by definition that f constitutes a homeomorphism between T and S. □

Lemma 13.15 

(Sequential Continuity). The bijective function $f:T\to S$ correlating the AIT with the canonical discrete system is sequentially continuous.

Proof. 

Let $f:T\to S$ be the bijective function between the AIT T and the canonical discrete system S. It is demonstrated that:

$\forall {\left(v_n\right)}_{n\in \mathbb{N}}\subseteq T,\forall v\in T:{\left(v_n\right)}_{n\in \mathbb{N}}\stackrel{seq}{\to }v\Rightarrow f\left({\left(v_n\right)}_{n\in \mathbb{N}}\right)\stackrel{seq}{\to }f\left(v\right)$

Where $\stackrel{seq}{\to }$ denotes sequential convergence.

Let ${\left(v_n\right)}_{n\in \mathbb{N}}$ be a sequence in T such that ${\left(v_n\right)}_{n\in \mathbb{N}}\stackrel{seq}{\to }v$. By definition, $\forall ϵ>0:\exists N\in \mathbb{N}:n\ge N\Rightarrow d_T(v_n,v)<ϵ$

Furthermore, as f is sequentially continuous, $\exists \delta >0$ such that $d_T(v_n,v)<\delta \Rightarrow d_S(f\left(v_n\right),f\left(v\right))<{ϵ}^{\prime }$.

Moreover, as ${\left(v_n\right)}_{n\in \mathbb{N}}\stackrel{seq}{\to }v$, $\exists N^{\prime }\in \mathbb{N}$ such that $\forall n\ge N^{\prime },d_T(v_n,v)<\delta$.

By transitivity, $\forall n\ge max(N,N^{\prime }),d_S(f\left(v_n\right),f\left(v\right))<{ϵ}^{\prime }$. Hence $f\left({\left(v_n\right)}_{n\in \mathbb{N}}\right)\stackrel{seq}{\to }f\left(v\right)$, proving the sequential continuity of f.

□

Theorem 13.16 

(f is a homeomorphism). Let $(S,F)$ be a discrete dynamical system with analytic inverse G. Let $T=(V,E)$ be the associated algebraic inverse tree and $f:V\to S$ the bijective function correlating nodes of T with states of S. It is demonstrated that:

f is bijective. f and $f^{-1}$ are sequentially continuous. Therefore, being bijective and bicontinuous, f constitutes a homeomorphism between the topological spaces associated with T and S.

Proof. 

Injectivity and surjectivity of f are demonstrated by recursive construction of T from G and definition of f. To show sequential continuity of f:

• Let ${\left(v_n\right)}_{n\in \mathbb{N}}\subseteq T$ and $v\in T$ such that ${\left(v_n\right)}_{n\in \mathbb{N}}\stackrel{seq}{\to }v$.
• By definition of sequential convergence, $\forall \epsilon >0,\exists N\in \mathbb{N}:n\ge N\Rightarrow d_T(v_n,v)<\epsilon$.
• As f is sequentially continuous, $\exists \delta >0:d_T(v_n,v)<\delta \Rightarrow d_S(f\left(v_n\right),f\left(v\right))<{\epsilon }^{\prime }$.
• Taking $\epsilon =\delta$ and by transitivity, ${\left(f\left(v_n\right)\right)}_{n\in \mathbb{N}}\stackrel{seq}{\to }f\left(v\right)$.

Similarly, it can be shown $f^{-1}$ preserves sequential convergence by explicitly verifying open sets are mapped to open sets in both directions.

Therefore, f is a homeomorphism between the spaces T and S. □

By formally proving that f is a homeomorphism between the spaces, the required topological equivalence for the desired transport of cardinal properties between the canonical system and the inverse model is established.

Definition 13.16 

(Topological Equivalence). Let $(X,\tau )$ and $(Y,\sigma )$ be topological spaces. We say there exists a topological equivalence between $(X,\tau )$ and $(Y,\sigma )$ if there exists a homeomorphic correspondence $f:(X,\tau )\to (Y,\sigma )$ such that:

(1) 

f is bijective, i.e., f is injective and surjective.

(2) 

Both f and $f^{-1}$ are continuous.

Furthermore, it holds that:

• Cardinality is preserved, i.e., $\left|X\right|=\left|Y\right|$.
• Compactness is preserved. If $(X,\tau )$ is compact, then $(Y,\sigma )$ is also compact.
• Connectivity is preserved. If $(X,\tau )$ is connected, then $(Y,\sigma )$ is also connected.

In other words, through f, a bijective and bicontinuous equivalence preserving topological cardinal properties is established between the spaces $(X,\tau )$ and $(Y,\sigma )$.

Remark 4. 

In the theory of discrete dynamical systems, topological equivalence refers to the idea that two discrete dynamical systems are equivalent from a topological perspective if they have the same topological structure, that is, if they have the same number of open and closed sets, and if the transition mappings between them are homeomorphisms.

The stability of the topological equivalence refers to the property that the topological equivalence is maintained under certain transformations or deformations of the dynamical system. In other words, if two discrete dynamical systems are topologically equivalent, then any continuous deformation or transformation of one of them that preserves the topological structure will also be topologically equivalent to the other system.

Stability of the topological equivalence is a fundamental property of the theory of discrete dynamical systems, and it is used to establish the existence of a topological integration theory for these systems. In particular, it is shown that if two discrete dynamical systems are topologically equivalent, then there exists a topological integration between them that preserves the topological structure and dynamics of the system.

This has important implications for solving problems in discrete dynamical systems, as it allows establishing a connection between set theory and theory of discrete dynamical systems. In particular, set theory can be used to establish the existence of topological solutions to problems in discrete dynamical systems, and theory of discrete dynamical systems can be used to establish the existence of dynamic solutions to set problems.

Definition 13.17 

(Discrete Homeomorphism). Given discrete spaces $(S,\tau ),(S^{\prime },{\tau }^{\prime })$, a discrete homeomorphism is a bijective and bicontinuous function $f:S\to S^{\prime }$. That is, f and $f^{-1}$ are continuous and discrete.

Note 3. 

Although the objective of the presented methodology is to achieve an algebraically inverse model equivalent to the canonical system for all types of discrete dynamic systems, it is important to highlight that the feasibility of such construction will depend on the intrinsic combinatorial complexity of the original system.

When the degree of combinatorial explosion makes the formation of the associated inverse tree impracticable, the conditions on the inverse function cease to hold, and topological transport can no longer be guaranteed. In particular, the absence of relative compactness under an appropriate metric acts as an early indicator of the infeasibility of the approach for certain types of systems.

Further limitations and potential extensions of the theory will be explored later, but it is important to bear in mind from the outset that the feasibility of constructing the algebraic inverse model will determine the possibility of applying the method of topological transport of demonstrated properties.

Example 4 

(Discrete Homeomorphism between Numeric Representations). Consider the set of natural numbers $\mathbb{N}$ as a discrete space. We define two functions:

(1) 

$fb:\mathbb{N}\to {\{0,1\}}^{*}$, which assigns to each natural number its binary representation.

(2) 

$fd:\mathbb{N}\to {\{0,1,2,3,4,5,6,7,8,9\}}^{*}$, which assigns to each natural number its decimal representation.

Here, ${\{0,1\}}^{*}$ and ${\{0,1,2,3,4,5,6,7,8,9\}}^{*}$ denote the sets of all finite strings of binary and decimal digits, respectively.

Both functions are bijective and continuous in the discrete sense, since each natural number has a unique binary and decimal representation, and the discrete topology of $\mathbb{N}$ is preserved under these transformations.

Now, we define the composition $fb\circ f{d}^{-1}:{\{0,1,2,3,4,5,6,7,8,9\}}^{*}\to {\{0,1\}}^{*}$, which assigns to each decimal representation its corresponding binary representation. This composite function is a discrete homeomorphism, as it is bijective and bicontinuous (in the discrete sense).

For example:

• $fb\circ f{d}^{-1}{\left(5\right)}_{10}={\left(101\right)}_2$
• $fb\circ f{d}^{-1}{\left(42\right)}_{10}={\left(101010\right)}_2$

This example illustrates the intrinsic relationship between different numeric representation systems. Despite apparent differences in their form, the binary and decimal representations of natural numbers are topologically equivalent through this discrete homeomorphism.

Definition 13.18 

(Topological Equivalence). Let $(S,\tau )$ be the topological space associated with the canonical discrete dynamical system, and $(T,\rho )$ be the topological space associated with the inverse model, where ρ is the natural topology on T. We say that $(S,\tau )$ and $(T,\rho )$ are topologically equivalent if there exists a function $f:(T,\rho )\to (S,\tau )$ such that:

(1) 

f is bijective, i.e., for each $s\in S$ there exists a unique $v\in V$ such that $f\left(v\right)=s$.

(2) 

Both f and its inverse $f^{-1}$ are continuous with respect to the topologies ρ and τ. That is, for each open set $U\in \tau$, its preimage $f^{-1}\left(U\right)$ is open in ρ; and for each open set $W\in \rho$, its image $f\left(W\right)$ is open in τ.

Theorem 13.17 

(AIT-Canonical System Homeomorphism). Let $(S,\tau ,F)$ be a discrete dynamical system (DDS) with analytic inverse G. Let $T=(V,E)$ be the associated algebraic inverse tree (AIT). Then there exists a homeomorphism $h:T\to S$ between the AIT and the canonical system.

Proof. 

We construct the function $h:T\to S$ and prove that it is a homeomorphism.

Construction of h: Define $h:V\to S$ as follows:

$$\forall v\in V:h\left(v\right)=s\iff v\mathrm{represents}\mathrm{state}s\mathrm{in}T$$

In other words, h maps each node v in the AIT to the state s in the canonical system that v represents.

Step 1: h is well-defined.

$$\begin{array}{c}\hfill \forall v\in V,\exists !s\in S:v\mathrm{represents}s\mathrm{in}T\\ \hfill \Rightarrow \forall v\in V,\exists !s\in S:h\left(v\right)=s\end{array}$$

Thus, h is a well-defined function from V to S.

Step 2: h is bijective.

• Injectivity: Let $v_1,v_2\in V$ with $v_1\ne v_2$. By the recursive construction of T using G, $v_1$ and $v_2$ represent different states in S. Thus, $h\left(v_1\right)\ne h\left(v_2\right)$. So h is injective.
• Surjectivity: Let $s\in S$. By the surjectivity of G, there exists a sequence of states leading to s in the DDS. This sequence is represented by a path in T ending at a node v with $h\left(v\right)=s$. Thus, $\forall s\in S,\exists v\in V:h\left(v\right)=s$. So h is surjective.

Step 3: h is continuous. Let $U\subseteq S$ be open in $\tau$. We show that $h^{-1}\left(U\right)$ is open in the AIT topology $\rho$. By the definition of $\tau$, $U={\bigcup }_{s\in U}\left\{s\right\}$. Then:

$$\begin{array}{cc}\hfill h^{-1}\left(U\right)& =h^{-1}\left(\bigcup _{s\in U}\left\{s\right\}\right)\hfill \\ \hfill & =\bigcup _{s\in U}{h}^{-1}\left(\left\{s\right\}\right)\hfill \\ \hfill & =\bigcup _{s\in U}\{v\in V:h\left(v\right)=s\}\hfill \end{array}$$

Each set $\{v\in V:h(v)=s\}$ is open in $\rho$ as it corresponds to a single node. Thus, $h^{-1}\left(U\right)$ is open in $\rho$ as a union of open sets. So h is continuous.

Step 4: $h^{-1}$ is continuous. Let $W\subseteq V$ be open in $\rho$. We show that $h\left(W\right)$ is open in $\tau$. By the definition of $\rho$, $W={\bigcup }_{v\in W}\left\{v\right\}$. Then:

$$\begin{array}{cc}\hfill h\left(W\right)& =h\left(\bigcup _{v\in W}\left\{v\right\}\right)\hfill \\ \hfill & =\bigcup _{v\in W}h\left(\left\{v\right\}\right)\hfill \\ \hfill & =\bigcup _{v\in W}\left\{h\left(v\right)\right\}\hfill \end{array}$$

Each singleton $\left\{h\right(v\left)\right\}$ is open in $\tau$ by definition. Thus, $h\left(W\right)$ is open in $\tau$ as a union of open sets. So $h^{-1}$ is continuous.

Therefore, h is a homeomorphism between the AIT $(T,\rho )$ and the canonical system $(S,\tau )$. □

Corollary 13.1. 

Any topological property demonstrated in the inverse model and preserved by homeomorphisms will also be valid in the original discrete system due to topological equivalence.

Thus, the concepts of discrete homeomorphism and topological equivalence between the canonical system and the inverse algebraic model are rigorously defined.

Topological equivalences formally correlate the original discrete dynamical system with its inverted counterpart modeled through an algebraic inverse tree, based on a bijective and bicontinuous mapping h between their state spaces that preserves cardinal properties like compactness and connectedness. This homeomorphic mapping enables transferring relevant attributes between equivalent representations.

14. Topological Transport

Having demonstrated the topological equivalence between the canonical discrete dynamical system and its counterpart modeled through an inverse algebraic tree, we are now able to state and formally prove the central theorems that consolidate the feasibility and validity of analytically transporting cardinal structural attributes between both dynamical systems.

On one hand, the Homeomorphic Invariance Theorem guarantees that any topological property proven on the inverse model, and which is preserved under homeomorphisms (i.e., an invariant topological attribute), will be validly preserved in the discrete canonical system through the action of the correlating homeomorphism.

Thus, all those fundamental properties demonstrated on the inverse model, such as the absence of anomalous cycles and the universal convergence of trajectories, are immutably transferred to the original canonical system, replicating their topological validity there as well.

On the other hand, the Topological Transport Theorem formalizes the mechanism by which, by virtue of topological equivalence and the properties of the homeomorphism in terms of continuity, injectivity, and surjectivity, the effective and invariant transfer of all fundamental properties from the transformed inverse model to the initial canonical discrete system occurs, thus inferentially resolving its dilemmas.

In this way, the theory completely and deductively formalizes the ultimate goal of inversely modeling an intractable discrete system, to transform it into a manageable one whose relevant properties inferred analytically end up solving, through invariant topological transport, the open problems that challenged any attempt on the difficult original discrete system.

Definition 14.1 

(Homeomorphic Invariant). A topological property P defined on topological spaces is homeomorphic invariant if it holds that:

$$\exists \mathit{homeomorphism}f:(X,\tau )\to (Y,\rho )\Rightarrow \left(P\right(X)\iff P(Y\left)\right)$$

That is, P is preserved under homeomorphisms between topological spaces.

Definition 14.2 

(Topological Transport).

Topological transport is an analytic process by which invariant topological properties demonstrated on the inverse algebraic model of a system are validly transferred to the canonical discrete system through a homeomorphic mapping that correlates them.

Intuitively, if we can prove a topological property (e.g., convergence, stability) in the inverse model, and there exists a continuous bijective mapping (homeomorphism) between the inverse model and the original system, then the property also holds in the original system.

Let $f:(X,\tau )\to (Y,\sigma )$ be a homeomorphism between a canonical discrete system S and its inverse algebraic model T. Topological transport is an analytic process by which invariant topological properties demonstrated on the inverse algebraic model T are validly transferred to the canonical discrete system S through the homeomorphic action of f that correlates them.

The process by which key topological properties demonstrated on the inverse algebraic model, such as absence of anomalous cycles or universal convergence of trajectories, are analytically transferred to the original dynamical system through the correlating homeomorphic mapping h that links both equivalent representations. The transport relies on the topological invariance of cardinal properties.

Theorem 14.1 

(Topological Transport Theorem). Let $(X,{\tau }_X)$ and $(Y,{\tau }_Y)$ be topological spaces and let $f:X\to Y$ be a homeomorphism. If P is a topological property and $P\left(X\right)$ holds, then $P\left(Y\right)$ also holds.

Proof. 

Suppose $P\left(X\right)$ holds. We want to show that $P\left(Y\right)$ also holds.

Step 1: f is bijective. Since f is a homeomorphism, by definition it is bijective.

Step 2: f is continuous. Since f is a homeomorphism, by definition it is continuous.

Step 3: $f^{-1}$ is continuous. Since f is a homeomorphism, by definition its inverse $f^{-1}$ is continuous.

Step 4: P is preserved under f. Let Q be the topological property defined by:

$$\forall Z:Q\left(Z\right)\iff P\left(f^{-1}\left(Z\right)\right)$$

Then, by the definition of topological property:

$$\begin{array}{cc}\hfill P\left(X\right)& \iff Q\left(f\right(X\left)\right)\hfill \\ \hfill & \iff Q\left(Y\right)\left(\sin \mathrm{ce}f\right(X)=Y)\hfill \\ \hfill & \iff P\left(f^{-1}\left(Y\right)\right)\hfill \\ \hfill & \iff P\left(X\right)(\sin \mathrm{ce}{f}^{-1}\left(Y\right)=X)\hfill \end{array}$$

Thus, we have shown that $P\left(X\right)\iff P\left(Y\right)$, i.e., P is preserved under f.

Therefore, since $P\left(X\right)$ holds by hypothesis and P is preserved under the homeomorphism f, we conclude that $P\left(Y\right)$ also holds. □

Corollary 14.1. 

In particular, properties demonstrated on algebraic inverse trees related to the absence of anomalous cycles and universal convergence of trajectories are transported to the original canonical discrete system through the action of the correlating homeomorphism.

Corollary 14.2 

(Guarantee of Topological Transport). Let $(S,F)$ be a discrete dynamical system modeled through a space $(X,d_X)$. Let $G:X\to P\left(X\right)$ be an associated inverse function, and let $(Y,d_Y)$ be an inverted combinatorial structure generated by G.

If G fulfills:

(1) 

Injectivity.

(2) 

Surjectivity.

(3) 

Exhaustiveness over X.

And if there exists $f:(Y,d_Y)\to (X,d_X)$ that is homeomorphic.

Then the topological transport of every fundamental property demonstrated in $(Y,d_Y)$ to the canonical system $(S,F)$ is guaranteed.

Proof. 

Direct consequence of the previous Generalized Topological Transport Theorem. Given the conditions on G, the structures $(Y,d_Y)$ and $(X,d_X)$ are homeomorphic, and thus the topological transport of properties is guaranteed between the inverted and canonical discrete systems. □

• Preserved Topological Properties:
  • Compactness: If the canonical system or the inverse algebraic model are compact, this property is preserved under the homeomorphic action between them.
  • Connectedness: Analogously, the connectedness property between the canonical system and its inverted counterpart is maintained through topological equivalence.
  • Metric Completeness: Relativized metric completeness is a preserved property of the metric spaces associated with it when topological transport is demonstrated.
  • Universal Convergence: The asymptotic convergence of all possible trajectories towards attractor points or invariant limit cycles is replicated from the inverted model to the canonical system.
  • Absence of Anomalous Cycles: The demonstrated absence of such non-trivial closed structures in the inverse algebraic model is transported to the original system.
• Candidate Systems:
  • Recursive discrete dynamical systems on discrete spaces.
  • Systems with moderate combinatorial explosions.
  • Chaotic systems with global convergence of trajectories.

Figure 1. Visualization of Discrete Inverse Dynamical Systems (DIDS). Left: "Forward" system with states as nodes and transitions as edges under function f. Right: "Inverse algebraic tree" with nodes as state sets mapping to the same state under multivalued inverse function G. Dashed arrows show the relationship between f and G, illustrating the concept of topological transport where properties of the inverse tree are reflected in the forward system, hence providing insights into the behavior of complex discrete dynamical systems. Multivalued injectivity of G is shown by the non-overlapping sets $G\left(s_i\right)$, ensuring that each state maps to a unique predecessor.

Figure 1. Visualization of Discrete Inverse Dynamical Systems (DIDS). Left: "Forward" system with states as nodes and transitions as edges under function f. Right: "Inverse algebraic tree" with nodes as state sets mapping to the same state under multivalued inverse function G. Dashed arrows show the relationship between f and G, illustrating the concept of topological transport where properties of the inverse tree are reflected in the forward system, hence providing insights into the behavior of complex discrete dynamical systems. Multivalued injectivity of G is shown by the non-overlapping sets $G\left(s_i\right)$, ensuring that each state maps to a unique predecessor.

Corollary 14.3 

(Non-Cyclicity Transport). If the AIT $(T,\rho )$ has no non-trivial cycles, then the canonical system $(S,\tau )$ also has no non-trivial cycles.

Proof. 

Let P be the property "having no non-trivial cycles". As shown earlier, $P\left(T\right)$ holds. Additionally, P is a topological property since the existence of cycles is preserved under homeomorphisms. By the Topological Transport Theorem and the existence of a homeomorphism $h:T\to S$, we conclude that $P\left(S\right)$ also holds. □

Corollary 14.4 

(Universal Convergence Transport). If all trajectories in the AIT $(T,\rho )$ converge to the root node, then all trajectories in the canonical system $(S,\tau )$ converge to the state corresponding to the root node.

Proof. 

Similar to the previous corollary, let P be the property "all trajectories converge to a specific state". As shown earlier, $P\left(T\right)$ holds for the root node. Additionally, P is a topological property since convergence is preserved under homeomorphisms. By the Topological Transport Theorem and the existence of a homeomorphism $h:T\to S$, we conclude that $P\left(S\right)$ also holds for the state $h\left(r\right)$ corresponding to the root node r. □

14.1. Fundamental Conditions for the Topological Transport

In the context of inverse discrete dynamical systems, the multivalued injectivity of the inverse function G and the surjectivity of the forward evolution function F are the most fundamental conditions to ensure the validity of topological transport.

14.1.1. Conditions under Which Properties Can Be Transferred

Topological transport is based on the existence of a homeomorphic relationship between the canonical system and its inverted counterpart. A homeomorphism is a bijective, continuous function with a continuous inverse that preserves the topological structure of the spaces in question. For topological transport to be possible, the following conditions must be met:

(1)

Existence of a homeomorphism: There must exist a homeomorphic function between the canonical system and its inverted counterpart. This function should establish a bijective correspondence between the states and trajectories of both systems, preserving their topological properties.

(2)

Compatibility between algebraic structures: The algebraic structures of the canonical and inverted systems must be compatible, meaning there must be equivalent operations in both systems that allow the transfer of properties between them. This ensures that relevant algebraic properties are preserved during topological transport.

(3)

Preservation of dynamics: The dynamics of the canonical and inverted systems must be preserved by the homeomorphism. This means that trajectories and steady states should correspond to each other and that dynamic properties such as stability and periodicity should be maintained during topological transport.

(4)

Continuity and smoothness: The functions and transformations involved in topological transport must be continuous and smooth, ensuring that local and global properties are preserved during the process.

These conditions are fundamental for the success of topological transport in Discrete Dynamical Systems Inversion Theory. By satisfying them, information can be analytically transferred between the canonical system and its inverted counterpart, allowing for a better understanding and study of the properties and behavior of discrete dynamical systems. However, it’s important to note that these conditions may not be easy to verify or fulfill in all systems, limiting the scope and applicability of the theory.

14.1.2. Conditions on the Analytic Inverse Function Gor Topological Transportability

Let $(S,F)$ be a discrete dynamical system, and let $T=(V,E)$ be its inverse algebraic tree generated by the inverse analytic function $G:S\to \mathcal{P}\left(S\right)$.

(1)

Relative Compactness: For T to be relatively compact, G must satisfy:

(a)

Multivalued injectivity: For any pair of distinct states $x,y\in S$, $G\left(x\right)$ and $G\left(y\right)$ are disjoint sets.

(b)

Bounded growth: There exists a function $f\left(n\right)$ such that for any initial state s and any n, the number of reachable states after n recursive applications of G is bounded by $f\left(n\right)$, and $f\left(n\right)$ is asymptotically smaller than an exponential function.

(2)

Relative Metric Completeness:

For the metric space associated with T to be relatively complete, G must satisfy:

(a)

Exhaustiveness: For any state $s\in S$, there exists a finite number of recursive applications of G that lead to a root state r.

(b)

Preservation of Cauchy sequences: If $\left(s_n\right)$ is a Cauchy sequence in S, then $\left(G\left(s_n\right)\right)$ is also a Cauchy sequence.

(3)

Connectivity:

To ensure the connectivity of T, G must satisfy:

(a)

Reachability: For any pair of states $s,t\in S$, there exists a finite sequence of states $(s_0,s_1,\dots ,s_n)$ such that $s_0=s$, $s_n=t$, and $s_{i+1}$ is in $G\left(s_i\right)$ for all i.

(4)

Topological Equivalence:

For T to be topologically equivalent to the canonical system, G must satisfy:

(a)

Invertibility: For any state $s\in S$, s is contained in $G\left(F\right(s\left)\right)$, where F is the evolution function of the canonical system.

(b)

Continuity: G is continuous with respect to the topologies of S and $\mathcal{P}\left(S\right)$.

14.2. Extension to Infinite AITs

In this section, we extend our results on finite Algebraic Inverse Trees (AITs) to the realm of infinite AITs using first-order logic and formal definitions, theorems, lemmas, and proofs.

Definition 14.3 

(Infinite AIT). Let ${\left(T_n\right)}_{n\in \mathbb{N}}$ be a sequence of finite AITs indexed by the natural numbers. An infinite AIT T is defined as the inductive limit of this sequence:

$$T=\underset{n\to \infty }{lim}{T}_n$$

Definition 14.4 

(Limit Topology on Infinite AIT). Let $(T,d)={lim}_{n\to \infty }(T_n,d_n)$ be the infinite AIT obtained as a limit of finite compatible AITs. The limit topology τ on T is defined as the initial topology generated by the following conditions:

(1) 

Open subsets in τ are arbitrary unions of opens in each $T_n$.

(2) 

Opens in each $T_n$ contain an open ball around each node.

Definition 14.5 

(Subcoproduct of AITS).

Let $T_{i}i\in I$ be a family of algebraic inverse trees (AITs) indexed by a set I. The

subcoproduct of $T_{i}i\in I$, denoted by ${\coprod }_{i\in I}{T}_i$, is an AIT T

constructed as follows:

1. The node set of T is the disjoint union of the node sets of $T_i$:

$$V\left(T\right)=\coprod _{i\in I}V\left(T_i\right)=\bigcup _{i\in I}(v,i):v\in V\left(T_i\right)$$

2. The edge set of T is the disjoint union of the edge sets of $T_i$:

$$E\left(T\right)=\coprod _{i\in I}E\left(T_i\right)=\bigcup _{i\in I}((u,i),(v,i)):(u,v)\in E\left(T_i\right)$$

3. The root of T is a new node r not in any $V\left(T_i\right)$, and there is an edge from r to the root of each $T_i$.

Theorem 14.2 

(Inheritance of Cardinal Properties). Let $(T,d)$ be an infinite AIT obtained as the limit of a sequence of compatible finite AITs $(T_n,d_n)$. That is, $(T,d)={lim}_{n\to \infty }(T_n,d_n)$. Then, $(T,d)$ inherits the following cardinal properties from the finite AITs $(T_n,d_n)$:

(1) 

Absence of non-trivial cycles

(2) 

Convergence of every infinite path towards the root node

Proof. 

Given that every finite AIT $(T_n,d_n)$ satisfies both properties by the previously proven Theorems:

• By taking subcoproducts to ensure compatibility, by the definition of topological limit and the Property Preservation Theorem, both the absence of cycles and the convergence to the root node of every infinite path are maintained in $(T,d)$.

Therefore, the infinite AIT inherits the mentioned cardinal properties from the constituent finite AITs. □

Lemma 14.3 

(Convergence of Paths). Let $(T,d)$ be an algebraic inverse tree equipped with the path length metric d. Let $(P=(v_1,v_2,\dots ))$ be an arbitrary path in T. Then, ${lim}_{i\to \infty }{v}_i=r$ where r is the root node of T.

Proof. 

We use the formal definitions:

• Path: $P\subseteq V$ is a path if

  $$\exists v_1,\dots ,v_n\in V:P=\langle v_1,\dots ,v_n\rangle \wedge \underset{i=1}{\overset{n-1}{\bigwedge }}(v_i,v_{i+1})\in E$$
• Convergence: P converges to node v if

  $$\forall \epsilon >0:\exists N\in \mathbb{N}:\forall n>N:d_T(v_n,v)<\epsilon$$

Take any arbitrary path $P=\langle v_1,\dots ,v_n\rangle$ in T. By the exhaustive construction of T using $C^{-1}$, every parent node expands paths from all children nodes. Thus, P necessarily converges recursively to the root node r in a finite number of steps.

Therefore, we conclude universal convergence in T:

$$\forall P\subseteq V:\left(P\mathrm{is}\mathrm{a}\mathrm{path}\mathrm{in}T\right)\to \left(P\mathrm{converges}\mathrm{to}r\right)$$

□

Theorem 14.4 

(Preservation of Properties). Let P be a cardinal property holding on each finite compatible AIT $T_n$. Then P also holds for the infinite limit AIT $(T,d)$ equipped with the limit topology τ.

Proof. 

Let ${\left(T_n\right)}_{n\in \mathbb{N}}$ be a sequence of finite AITs such that P holds for each $T_n$. By the definition of the inductive limit, for each $k\in \mathbb{N}$, there exists $n_k\in \mathbb{N}$ such that for all $n\ge n_k$, $T_{n_k}$ is a subtree of $T_n$.

Since P holds for each $T_n$, it must also hold for each subtree $T_{n_k}$. By the Inheritance of Cardinal Properties theorem, P is preserved in the infinite limit AIT $T={lim}_{n\to \infty }{T}_n$.

Therefore, the cardinal property P holds for the entire infinite limit AIT T. □

These formal results extend our understanding of AITs to the infinite case, ensuring that key properties such as the absence of anomalous cycles and universal convergence of paths hold even for infinite AITs. This strengthens our topological approach to the Collatz Conjecture.

15. Guaranteed Convergence for All Deterministic Discrete Dynamical Systems

Definition 15.1 

(Cycle). Let $(S,F)$ be a discrete dynamical system, where S is the state space and $F:S\to S$ is the evolution function. A cycleof period $n\in \mathbb{N}$ is a sequence of distinct states $(x_1,\dots ,x_n)\in S^n$ such that:

(1) 

$F\left(x_i\right)=x_{i+1}$ for all $1\le i<n$

(2) 

$F\left(x_n\right)=x_1$

We denote the set of all cycles of $(S,F)$ by $\mathcal{C}(S,F)$.

Definition 15.2 

(Attractor). Let $(S,F)$ be a discrete dynamical system. A set $A\subseteq S$ is anattractorif:

(1) 

A is non-empty and compact

(2) 

A is invariant under F, i.e., $F\left(A\right)\subseteq A$

(3) 

There exists an open neighborhood $U\supseteq A$ such that for all $x\in U$,

${lim}_{n\to \infty }d(F^n\left(x\right),A)=0$, where d is a metric on S and $F^n$ denotes the n-fold composition of F with itself.

We denote the set of all attractors of $(S,F)$ by $\mathcal{A}(S,F)$.

Definition 15.3 

(Convergence to an Attractor). Let $(S,F)$ be a discrete dynamical system and $A\in \mathcal{A}(S,F)$ be an attractor. We say that a point $x\in S$ converges to A if:

$$\forall \epsilon >0,\exists N\in \mathbb{N},\forall n\ge N:d(F^n\left(x\right),A)<\epsilon$$

where d is a metric on S. We denote the set of all points that converge to A by $\mathcal{B}\left(A\right)$, called the basin of attraction of A.

Theorem 15.1 

(Multivalued Injectivity of G in the Presence of Cycles). Let $(S,F)$ be a discrete dynamical system and let $G:S\to P\left(S\right)$ be the inverse function of F. Suppose $(S,F)$ has a cycle $(x_1,\dots ,x_n)$. Then, G is multivalued injective if and only if the following conditions hold:

(1) 

For all $1\le i,j\le n$ with $i\ne j$, $G\left(x_i\right)\cap G\left(x_j\right)=\varnothing$.

(2) 

For all $y\in S\setminus \{x_1,\dots ,x_n\}$ and all $1\le i,j\le n$ with $i\ne j$, if $y\in G\left(x_i\right)$ then $y\notin G\left(x_j\right)$.

In other words, G is multivalued injective in the presence of a cycle if and only if:

(1) 

Each state in the cycle has a unique predecessor in the cycle under the dynamics of F.

(2) 

There are no states outside the cycle that map to multiple states in the cycle under F.

Proof. 

$(\Rightarrow )$ Suppose G is multivalued injective. Then, by definition, for every pair of distinct states $x,y\in S$, we have $G\left(x\right)\cap G\left(y\right)=\varnothing$.

In particular, for all $1\le i,j\le n$ with $i\ne j$, since $x_i$ and $x_j$ are distinct states in the cycle, $G\left(x_i\right)\cap G\left(x_j\right)=\varnothing$, thus demonstrating condition 1.

Moreover, for all $y\in S\setminus \{x_1,\dots ,x_n\}$ and all $1\le i,j\le n$ with $i\ne j$, if $y\in G\left(x_i\right)$ then $y\notin G\left(x_j\right)$, as otherwise we would have $G\left(x_i\right)\cap G\left(x_j\right)\ne \varnothing$, contradicting the multivalued injectivity of G. This demonstrates condition 2.

$(\Leftarrow )$ Suppose conditions 1 and 2 are satisfied. We must show that for every pair of distinct states $x,y\in S$, $G\left(x\right)\cap G\left(y\right)=\varnothing$.

Let $x,y\in S$ with $x\ne y$. If $x,y\in \{x_1,\dots ,x_n\}$, then $G\left(x\right)\cap G\left(y\right)=\varnothing$ by condition 1.

If $x\in \{x_1,\dots ,x_n\}$ and $y\in S\setminus \{x_1,\dots ,x_n\}$ (or vice versa), then $G\left(x\right)\cap G\left(y\right)=\varnothing$ by condition 2.

Finally, if $x,y\in S\setminus \{x_1,\dots ,x_n\}$, then $G\left(x\right)\cap G\left(y\right)=\varnothing$ because F is a function (and thus each state has at most one predecessor).

Therefore, G is multivalued injective. □

Theorem 15.2 

(Unique Attractor in Each Tree of the Forest). Let $(S,F)$ be a discrete dynamical system and let $\mathcal{F}=\{T_1,\dots ,T_n\}$ be the forest of inverse algebraic trees associated with $(S,F)$, where each tree $T_i$ is rooted at an attractor $A_i\in \mathcal{A}(S,F)$. Then:

(1) 

Each tree $T_i$ in the forest $\mathcal{F}$ has a unique attractor $A_i$.

(2) 

If $A_i$ is a cycle or an infinite cycle, then each state in $A_i$ has a unique predecessor in $A_i$ under the dynamics of F.

Proof. 

Let $T_i\in \mathcal{F}$ be an arbitrary tree in the forest, rooted at an attractor $A_i\in \mathcal{A}(S,F)$.

Part 1: We first prove that $A_i$ is the unique attractor in $T_i$. Suppose, for contradiction, that there exists another attractor $A_i^{\prime }\ne A_i$ in $T_i$.

By the definition of an attractor, there exist open neighborhoods $U,U^{\prime }$ of $A_i,A_i^{\prime }$ respectively, such that for all $x\in U$ and $x^{\prime }\in U^{\prime }$, we have:

$$\underset{n\to \infty }{lim}d(F^n\left(x\right),A_i)=0\mathrm{and}\underset{n\to \infty }{lim}d(F^n\left(x^{\prime }\right),A_i^{\prime })=0$$

Since $T_i$ is a tree, there exists a unique path connecting any two nodes. Let $x\in A_i$ and $x^{\prime }\in A_i^{\prime }$ be arbitrary states, and let $(x=x_1,x_2,\dots ,x_k=x^{\prime })$ be the unique path connecting them in $T_i$.

As $x_1\in A_i\subseteq U$ and $x_k\in A_i^{\prime }\subseteq U^{\prime }$, there must exist some $1<j<k$ such that $x_j\in U$ but $x_{j+1}\notin U$. However, since $(x_j,x_{j+1})$ is an edge in $T_i$, we have $F\left(x_{j+1}\right)=x_j$, which implies:

$$\underset{n\to \infty }{lim}d(F^n\left(x_{j+1}\right),A_i)=\underset{n\to \infty }{lim}d(F^{n-1}\left(x_j\right),A_i)=0$$

contradicting $x_{j+1}\notin U$. Therefore, $A_i$ is the unique attractor in $T_i$.

Part 2: Now suppose $A_i$ is a cycle or an infinite cycle. We need to prove that each state in $A_i$ has a unique predecessor in $A_i$ under F.

Let $x\in A_i$ be an arbitrary state. By the definition of a cycle, there exists a unique state $y\in A_i$ such that $F\left(y\right)=x$. We claim that y is the unique predecessor of x in $A_i$.

Suppose, for contradiction, that there exists another state $z\in A_i$ with $z\ne y$ such that $F\left(z\right)=x$. Since both y and z are in $A_i$, which is an attractor in $T_i$, there must be paths from y and z to the root of $T_i$. But then, x would have two distinct predecessors in $T_i$, namely y and z, contradicting the fact that $T_i$ is a tree.

Therefore, each state in $A_i$ has a unique predecessor in $A_i$ under F. □

Theorem 15.3 

(Generalized Convergence to Attractors in Inverse Trees). Let $(S,F)$ be a discrete dynamical system satisfying the conditions of DIDS, and let $\mathcal{F}=\{T_1,\dots ,T_n\}$ be the inverse algebraic forest associated with $(S,F)$, where each tree $T_i$ is rooted at an attractor $A_i\in \mathcal{A}(S,F)$. Then, for every $x\in S$, if x belongs to the tree $T_i$, then x converges to $A_i$ under the dynamics of F. In other words, $x\in \mathcal{B}\left(A_i\right)$.

Proof. 

Let $(S,F)$ be a discrete dynamical system satisfying the conditions of DIDS, and let $\mathcal{F}=\{T_1,\dots ,T_n\}$ be the inverse algebraic forest associated with $(S,F)$, where each tree $T_i$ is rooted at an attractor $A_i\in \mathcal{A}(S,F)$.

Take an arbitrary point $x\in S$ and suppose x belongs to the tree $T_i$ rooted at the attractor $A_i\in \mathcal{A}(S,F)$.

Our aim is to prove that $x\in \mathcal{B}\left(A_i\right)$, meaning x converges to $A_i$ under the dynamics of F. Formally:

$$\begin{array}{c}\hfill \forall \epsilon >0,\exists N\in \mathbb{N},\forall n\ge N:d(F^n\left(x\right),A_i)<\epsilon \end{array}$$

Considering the construction of the inverse tree $T_i$, there exists a unique path $(v_1,\dots ,v_k)$ from the node $v_1$ containing x to the root node $v_k$ corresponding to an element of $A_i$.

Since $A_i$ is an attractor, we know that:

$$\begin{array}{c}\hfill \forall \epsilon >0,\exists \delta >0,\forall y\in B_{\delta }\left(A_i\right):\underset{n\to \infty }{lim}d(F^n\left(y\right),A_i)<\epsilon \end{array}$$

Moreover, due to the continuity of F and the compactness of S, we have:

$$\begin{array}{c}\hfill \forall \epsilon >0,\exists \delta >0,\forall y,z\in S:d(y,z)<\delta \Rightarrow d(F\left(y\right),F\left(z\right))<\frac{\epsilon }{k}\end{array}$$

Choose $N\in \mathbb{N}$ such that:

$$\begin{array}{c}\hfill \forall n\ge N:d(F^n\left(v_k\right),A_i)<\delta \end{array}$$

Then, for all $n\ge N$:

$$\begin{array}{cc}\hfill d(F^n\left(x\right),A_i)& \le d(F^n\left(x\right),F^n\left(v_1\right))+d(F^n\left(v_1\right),F^n\left(v_2\right))+\dots \hfill \\ \hfill & +d(F^n\left(v_{k-1}\right),F^n\left(v_k\right))+d(F^n\left(v_k\right),A_i)\hfill \\ \hfill & <\frac{\epsilon }{k}+\frac{\epsilon }{k}+\dots +\frac{\epsilon }{k}+\delta (\mathrm{by}\mathrm{the}\mathrm{continuity}\mathrm{of}F)\hfill \\ \hfill & <\epsilon (\sin \mathrm{ce}\delta <\frac{\epsilon }{k})\hfill \end{array}$$

Hence, $x\in \mathcal{B}\left(A_i\right)$, implying x converges to $A_i$ under the dynamics of F. □

Definition 15.4 

(Point of Contact). Let $(S,F)$ be a discrete dynamical system and let $\mathcal{F}=\{T_1,\dots ,T_n\}$ be the inverse algebraic forest associated with $(S,F)$, where each tree $T_i$ is rooted at an attractor $A_i\in \mathcal{A}(S,F)$. For each tree $T_i$, we define the point of contact $c_i$ as the state in $A_i$ such that for each $x\in T_i$, $c_i$ is the first state in $A_i$ reached by the sequence ${\left(F^n\left(x\right)\right)}_{n\in \mathbb{N}}$. Formally:

$$c_i=min\{y\in A_i:\exists x\in T_i,\exists n\in \mathbb{N},F^n\left(x\right)=y\}$$

where the minimum is taken with respect to some predefined total order on S.

Theorem 15.4 

(Uniqueness of Point of Contact). Let $(S,F)$ be a discrete dynamical system satisfying the conditions of DIDS, and let $\mathcal{F}=\{T_1,\dots ,T_n\}$ be the inverse algebraic forest associated with $(S,F)$. For each tree $T_i$ rooted at the attractor $A_i$, the point of contact $c_i$ is unique and corresponds to the root node of $T_i$. Furthermore, for each $x\in T_i$, the sequence ${\left(F^n\left(x\right)\right)}_{n\in \mathbb{N}}$ converges to $c_i$.

Moreover, the point of contact $c_i$ is the only point in the attractor cycle $A_i$ that violates multivalued injectivity, i.e., for any $x\ne c_i$ in $A_i$:

$$G\left(x\right)\cap G\left(c_i\right)=\left\{c_i\right\}$$

Proof. 

First, we demonstrate that $c_i$ corresponds to the root node of $T_i$. Suppose, for contradiction, that there exists a node $v\in T_i$ such that v is strictly above the node containing $c_i$. Then, there exists a state $y\in v$ such that $F\left(y\right)\in A_i$ and $F\left(y\right)\ne c_i$. But this contradicts the definition of $c_i$ as the first state in $A_i$ reached by any sequence starting in $T_i$. Therefore, $c_i$ must be contained in the root node of $T_i$.

Next, we demonstrate that $c_i$ is unique. Suppose, for contradiction, that there exist two distinct points of contact $c_i$ and $c_i^{\prime }$ for $T_i$. Since both are contained in the root node of $T_i$, there must be states $x,x^{\prime }\in T_i$ and natural numbers $n,n^{\prime }$ such that $F^n\left(x\right)=c_i$ and $F^{n^{\prime }}\left(x^{\prime }\right)=c_i^{\prime }$. Without loss of generality, assume $n\le n^{\prime }$. Then, $F^{n^{\prime }-n}\left(c_i\right)=F^{n^{\prime }}\left(x\right)=c_i^{\prime }$, implying that $c_i^{\prime }$ is reachable from $c_i$ under the dynamics of F. But since $c_i$ and $c_i^{\prime }$ are in the same cycle or attractor set $A_i$, this implies that $c_i$ is also reachable from $c_i^{\prime }$, contradicting the assumption that they are distinct. Therefore, the point of contact $c_i$ is unique.

Finally, we demonstrate that for each $x\in T_i$, the sequence ${\left(F^n\left(x\right)\right)}_{n\in \mathbb{N}}$ converges to $c_i$. Let $x\in T_i$ be arbitrary. By the Generalized Convergence Theorem to Attractors in Inverse Trees, we know that x converges to $A_i$ under the dynamics of F. Furthermore, since $c_i$ is the unique point of contact and is in the root node of $T_i$, the sequence ${\left(F^n\left(x\right)\right)}_{n\in \mathbb{N}}$ must reach $c_i$ before any other state in $A_i$. Since $A_i$ is an attractor, once the sequence reaches $c_i$, it must remain in $A_i$ and therefore converge to $c_i$ due to the cyclic nature of attractors in DIDS.

To show that $c_i$ is the only point in $A_i$ that violates multivalued injectivity, consider any $x\ne c_i$ in $A_i$. By the definition of a point of contact, $c_i$ is the unique predecessor of x under G within the cycle. Therefore, $G\left(x\right)\cap G\left(c_i\right)=\left\{c_i\right\}$. □

Note: It should be noted that the condition of multivalued injectivity, $G\left(x\right)\cap G\left(y\right)=\varnothing$ for all $x\ne y$, may not hold at certain isolated points in the domain, as in the case of $G\left(1\right)$ and $G\left(2\right)$ for the Collatz function. However, this does not invalidate the general applicability of the theory, as long as these instances are exceptional and do not represent the predominant behavior of the inverse function G.

Theorem 15.5 

(Attractor Set Characterization). Let $(S,f)$ be a discrete dynamical system, where S is the state space and $f:S\to S$ is the evolution function. Let $G:S\to \mathcal{P}\left(S\right)$ be the inverse function of f, where $\mathcal{P}\left(S\right)$ denotes the power set of S. For a point $pc\in S$ and a set $A=\{x_1,x_2,\dots ,x_t\}\subseteq S$, A is an attractor set with point of contact $pc$ if and only if:

(1) 

$pc=x_1$

(2) 

$f\left(x_i\right)=x_{i+1}$ for $i=1,2,\dots ,t-1$

(3) 

$f\left(x_t\right)=pc$

(4) 

$G\left(f\left(x_t\right)\right)=G\left(pc\right)=x_t$

Moreover, A is a fixed point if and only if $t=1$, and A is a periodic cycle if and only if $t>1$.

The condition $G\left(f\left(x_t\right)\right)=G\left(pc\right)=x_t$ implies that the point of contact $pc$ is the only point in the attractor cycle A where multivalued injectivity is not satisfied, i.e., for any $x_i\ne pc$ in A:

$$G\left(x_i\right)\cap G\left(pc\right)=\left\{pc\right\}$$

Theorem 15.6 

(Intersection Set at Point of Contact). Let $(S,F)$ be a discrete dynamical system and $G:S\to \mathcal{P}\left(S\right)$ be its inverse function. If $pc\in S$ is a point of contact for an attractor cycle, then for any $x\ne pc$ in the cycle:

$$G\left(x\right)\cap G\left(pc\right)=\left\{pc\right\}$$

Proof. 

Let $pc\in S$ be a point of contact for an attractor cycle, and let $x\ne pc$ be any other point in the cycle. By the definition of a point of contact, $pc$ is the unique predecessor of x under G within the cycle. Therefore, $pc\in G\left(x\right)$.

Now, suppose there exists a point $y\in G\left(x\right)\cap G\left(pc\right)$ such that $y\ne pc$. Then, y is a predecessor of both x and $pc$ under G. However, this contradicts the fact that $pc$ is the unique predecessor of x within the cycle.

Thus, we conclude that $G\left(x\right)\cap G\left(pc\right)=\left\{pc\right\}$. □

Theorem 15.7 

(Uniqueness of Point of Contact in Violating Multivalued Injectivity). Let $(S,F)$ be a discrete dynamical system and $G:S\to \mathcal{P}\left(S\right)$ be its inverse function. If $\{x_1,\dots ,x_n\}$ is an attractor cycle with point of contact $pc$, then $pc$ is the unique point in the cycle that violates multivalued injectivity. That is, for all $x_i,x_j\in \{x_1,\dots ,x_n\}\setminus \left\{pc\right\}$ with $i\ne j$:

$$G\left(x_i\right)\cap G\left(x_j\right)=\varnothing$$

Proof. 

Let $\{x_1,\dots ,x_n\}$ be an attractor cycle with point of contact $pc$. We will prove the theorem by contradiction.

Suppose there exist distinct points $x_i,x_j\in \{x_1,\dots ,x_n\}\setminus \left\{pc\right\}$ such that $G\left(x_i\right)\cap G\left(x_j\right)\ne \varnothing$. Let $y\in G\left(x_i\right)\cap G\left(x_j\right)$. Then, y is a common predecessor of both $x_i$ and $x_j$ under G.

By the definition of an attractor cycle, each point in the cycle has a unique predecessor within the cycle. Therefore, y must be the unique predecessor of both $x_i$ and $x_j$. However, this implies that $x_i=x_j$, contradicting the assumption that $x_i$ and $x_j$ are distinct.

Thus, we conclude that for all $x_i,x_j\in \{x_1,\dots ,x_n\}\setminus \left\{pc\right\}$ with $i\ne j$, $G\left(x_i\right)\cap G\left(x_j\right)=\varnothing$. In other words, $pc$ is the unique point in the attractor cycle that violates multivalued injectivity. □

Proposition 1. 

The definition of the Algebraic Inverse Tree (AIT) associated with a Discrete Inverse Dynamical System (DIDS) $(S,F,G)$ includes the attractor and the point of contact when generating the tree.

Proof. 

Let $(S,F)$ be a Discrete Dynamical System (DDS) and $G:S\to \mathcal{P}\left(S\right)$ be its inverse function such that $(S,F,G)$ is a Discrete Inverse Dynamical System (DIDS).

The AIT $T=(V,E)$ associated with $(S,F,G)$ is constructed as follows:

$$\begin{array}{cccc}\hfill V& =S\hfill & \hfill & (\mathrm{Nodes}\mathrm{of}\mathrm{the}\mathrm{AIT})\hfill \\ \hfill E& =\left\{\right(s,t)\in S\times S:s\in G(t\left)\right\}\hfill & \hfill & (\mathrm{Edges}\mathrm{of}\mathrm{the}\mathrm{AIT})\hfill \\ \hfill r& =c\hfill & \hfill & (\mathrm{Root}\mathrm{of}\mathrm{the}\mathrm{AIT})\hfill \end{array}$$

where c is the point of contact of the attractor cycle.

Let’s prove that this definition of the AIT guarantees the inclusion of the attractor and the point of contact:

Step 1: The point of contact c is included in the AIT. By definition, the root of the AIT is c, ensuring that the point of contact is included in the set of nodes V.

Step 2: Elements of the attractor cycle are included in the AIT. Let $A=\{s_0,s_1,\dots ,s_{t-1}\}$ be the attractor cycle of the DIDS, where $s_0=c$ and $s_i=F\left(s_{i-1}\right)$ for $1\le i<t$.

For each $s_i\in A$, we have $s_{i-1}\in G\left(s_i\right)$ by the definition of G. Therefore, $(s_{i-1},s_i)\in E$ for all $1\le i<t$, and $(s_{t-1},s_0)\in E$.

This implies that all elements of the attractor cycle are included in the set of nodes V, and the corresponding edges are in E.

Step 3: The AIT is exhaustive. Due to the exhaustiveness property of G, for every $s\in S$, there exists $k\in \mathbb{N}$ such that $c\in G^k\left(s\right)$. This means that for every $s\in S$, there exists a path in the AIT from s to the root c.

Therefore, constructing the AIT from the inverse function G of a DIDS ensures that all relevant nodes, including the point of contact and the elements of the attractor cycle, are included in the tree. □

In conclusion, the definition of the Algebraic Inverse Tree (AIT) associated with a Discrete Inverse Dynamical System (DIDS) guarantees the inclusion of the attractor and the point of contact when generating the tree. This proposition holds for all DIDS.

Theorem 15.8 

(Impossibility of Infinite Attractors). Let $(S,F)$ be a Discrete Dynamical System, where S is the state space and $F:S\to S$ is the deterministic and surjective evolution function. Let $G:S\to \mathcal{P}\left(S\right)$ be the analytic inverse of F, which is multivalued injective, surjective, and exhaustive. Let $T=(V,E)$ be the Algebraic Inverse Tree generated by G.

Then, there are no infinite cycles in T. That is, there does not exist an infinite sequence of distinct nodes $v_1,v_2,\dots \in V$ such that $v_{i+1}\in G\left(v_i\right)$ for all $i\ge 1$ and $v_i\ne v_j$ for all $i\ne j$.

Proof. 

We proceed by contradiction. Suppose there exists an infinite cycle in T, i.e., an infinite sequence of distinct nodes $v_1,v_2,\dots \in V$ such that:

$$\begin{array}{cc}\hfill & \forall i\ge 1:v_{i+1}\in G\left(v_i\right)\hfill \\ \hfill & \forall i,j\in \mathbb{N}:(i\ne j\to v_i\ne v_j)\hfill \end{array}$$

Step 1: By the exhaustiveness property of G, for each node $v_i$ in the sequence, there exists a finite number of recursive applications of G that lead to a root node r. Formally:

$$\forall i\in \mathbb{N},\exists n_i\in \mathbb{N},\exists r\in V:\left(r\mathrm{is}\mathrm{a}\mathrm{root}\mathrm{node}\right)\wedge (v_i\in G^{n_i}\left(r\right))$$

where $G^{n_i}$ denotes the $n_i$-fold composition of G with itself.

Step 2: By the multivalued injectivity of G, each node in T has a unique parent. Therefore, for any two distinct nodes $v_i$ and $v_j$ in the sequence, their paths to the root must diverge at some point. Formally:

$$\forall i,j\in \mathbb{N}:(i\ne j\to \exists k\in \mathbb{N}:G^k\left(v_i\right)\cap G^k\left(v_j\right)=\varnothing )$$

Step 3: Consider the subsequence ${\left\{v_{n_i}\right\}}_{i=1}^{\infty }$ of nodes, where each $v_{n_i}$ is the node in the original sequence at which the path to the root is exactly $n_i$ steps long. By Step 1, this subsequence is infinite.

Step 4: By Step 2, for any two distinct nodes $v_{n_i}$ and $v_{n_j}$ in the subsequence, we have:

$$G^{min(n_i,n_j)}\left(v_{n_i}\right)\cap G^{min(n_i,n_j)}\left(v_{n_j}\right)=\varnothing$$

Step 5: We now apply the pigeonhole principle to the subsequence ${\left\{v_{n_i}\right\}}_{i=1}^{\infty }$. Let $M=\left|S\right|$, the cardinality of the state space S. Consider the first $M+1$ nodes in the subsequence: $v_{n_1},v_{n_2},\dots ,v_{n_{M+1}}$.

By the pigeonhole principle, if we have $M+1$ pigeons (nodes) and M pigeonholes (possible subsets of S), then there must be at least two pigeons (nodes) in the same pigeonhole (subset of S). In other words, there must exist two distinct nodes $v_{n_i}$ and $v_{n_j}$ with $1\le i<j\le M+1$ such that:

$$G^{min(n_i,n_j)}\left(v_{n_i}\right)=G^{min(n_i,n_j)}\left(v_{n_j}\right)$$

But this contradicts Step 4, which states that these sets should be disjoint.

Step 6: Therefore, our initial assumption must be false, and there cannot exist an infinite cycle in T. Formally:

$$\nexists v_1,v_2,\dots \in V:(\forall i\ge 1:v_{i+1}\in G\left(v_i\right))\wedge (\forall i,j\in \mathbb{N}:i\ne j\to v_i\ne v_j)$$

Thus, we have proven by contradiction that there are no infinite cycles in the Algebraic Inverse Tree T generated by the multivalued injective, surjective, and exhaustive analytic inverse function G of a Discrete Dynamical System $(S,F)$. □

Remark 5 

(Formal Proof of Algorithm Termination in IDDS). In the context of Inverse Discrete Dynamical Systems (IDDS), it is crucial to establish that the algorithms used for analysis and resolution always terminate, even in the presence of exceptional cases. This remark provides rigorous mathematical arguments demonstrating that algorithms based on IDDS principles are guaranteed to terminate and cannot enter infinite loops.

We have shown that dynamical systems satisfying the conditions of a Discrete Inverse Dynamical System (DIDS) possess an analytic inverse function that is multivalued injective, surjective, and exhaustive (Theorem 15.22). These properties are essential for the construction and analysis of the associated inverse algebraic forest.

Furthermore, we have proven that any DIDS satisfying these conditions has a unique attractor set (Theorem 15.25) and cannot contain cycles of infinite length in its associated inverse algebraic forest (Theorem 15.8). These results rely on the structural properties of the inverse algebraic forest and the characteristics of the analytic inverse function.

The proof of Theorem 15.8 utilizes the well-ordering principle of natural numbers and the exhaustiveness of the inverse function. By assuming the existence of an infinite cycle in the inverse algebraic forest and constructing a subsequence of nodes with strictly decreasing distances from the root, we arrive at a contradiction with the multivalued injectivity property of the inverse function.

Consequently, we have formally established that the inverse algebraic forest associated with a DIDS cannot contain cycles of infinite length. This implies that any trajectory in the dynamical system will converge to its unique attractor set after a finite number of iterations, ensuring the termination of algorithms based on IDDS principles.

It is important to note that these proofs hold for all elements in the domain of the dynamical system, without exceptions. The multivalued injectivity, surjectivity, and exhaustiveness properties of the analytic inverse function, together with the structure of the inverse algebraic forest, guarantee that there are no special cases that could lead to an infinite execution of the algorithm.

In summary, the theorems and proofs presented in this work provide a rigorous formal foundation for the termination of algorithms based on Inverse Discrete Dynamical Systems. The mathematical properties of IDDS and the structure of the associated inverse algebraic forests ensure the absence of infinite loops and the convergence of trajectories to unique attractor sets, even in the presence of exceptional cases. This establishes the soundness and generality of the IDDS framework for the analysis and resolution of discrete dynamical systems.

Remark 6 

(Finitude of Branches vs. Infinitude of IDDS Trees). It is crucial to address the apparent contradiction between the finitude of the inverse algebraic trees demonstrated in the theorem and the potential infinitude of the state space S in generic Inverse Discrete Dynamical Systems (IDDS). Let us clarify this point.

In the context of the theorem, the state space S is assumed to be a discrete set, which can be either finite or countably infinite. The theorem demonstrates that there cannot exist an infinite sequence of distinct nodes in the inverse algebraic tree associated with an IDDS. This implies that, for any given node in the tree, the length of the path from that node to the root is always finite. In other words, each branch of the tree has a finite length.

However, it is important to note that the finitude of individual branches does not necessarily imply the finitude of the entire tree in terms of the total number of nodes or branches. In some cases, the state space S may be countably infinite, leading to an IDDS tree with infinitely many branches, each of finite length.

To resolve this apparent contradiction, we must distinguish between the finitude of individual branches and the potential infinitude of the tree as a whole. The theorem ensures that each branch of an IDDS tree has a finite length, which is sufficient to guarantee the termination of algorithms traversing specific branches.

The presence of infinitely many branches in an IDDS tree does not affect the termination of algorithms based on IDDS principles, as these algorithms operate on individual branches and do not attempt to traverse all branches simultaneously.

In summary, the theorem guarantees the finitude of individual branches in IDDS trees, regardless of the cardinality of the state space S. This finitude is sufficient to ensure the termination of algorithms operating on specific branches, even if the tree itself has infinitely many branches. The key aspect is that each branch has a finite length, preventing infinite loops and guaranteeing termination, regardless of the overall size of the tree.

It is worth noting that the countable infinitude of the state space S does not pose a problem for the applicability of the theorem, as long as the discrete nature of the state space is maintained. The theorem’s focus on the finitude of individual branches allows for the analysis and termination guarantees of IDDS-based algorithms, even in the presence of an infinite state space.

Theorem 15.9 

(Implications of Discrete Dynamical Systems). Let $(S,F)$ be a discrete dynamical system, where S is a discrete state space and $F:S\to S$ is the evolution function. Let $G:S\to \mathcal{P}\left(S\right)$ be the inverse function of F, where $\mathcal{P}\left(S\right)$ denotes the power set of S. Then:

(1) 

$(S,F)$ being a discrete dynamical system implies that F is deterministic.

(2) 

F being deterministic implies that G is injective.

(3) 

F being surjective implies that G is surjective, which in turn implies that G is exhaustive.

Proof. 

(1)

By the definition of a discrete dynamical system, for each $s\in S$, there exists a unique $F\left(s\right)\in S$. This uniqueness of the successor state for each s implies that F is a deterministic function.

(2)

Suppose F is deterministic. Let $a,b\in S$ such that $a\ne b$. We want to show that $G\left(a\right)\cap G\left(b\right)=\varnothing$.

Assume, for contradiction, that there exists $t\in G\left(a\right)\cap G\left(b\right)$. Then, by definition of G, we have $F\left(t\right)=a$ and $F\left(t\right)=b$. But since F is a function, this implies $a=b$, contradicting the assumption that $a\ne b$.

Therefore, $G\left(a\right)\cap G\left(b\right)=\varnothing$ whenever $a\ne b$, which means G is injective.

(3)

Suppose F is surjective. Let $s\in S$. We want to show that there exists $t\in S$ such that $s\in G\left(t\right)$.

Since F is surjective, there exists $t\in S$ such that $F\left(t\right)=s$. By the definition of G, this means that $s\in G\left(t\right)$.

Therefore, G is surjective. Furthermore, if G is surjective, then for each $s\in S$, there exists a finite sequence of states leading from s to a root state under the repeated application of G, implying that G is exhaustive.

□

Theorem 15.10 

(Non-surjectivity of F implies Non-surjectivity of G). Let $(S,F)$ be a discrete dynamical system and $G:S\to \mathcal{P}\left(S\right)$ its inverse function. If G is injective but not surjective, then F is also not surjective.

Proof. 

Suppose G is injective but not surjective. This means there exists at least one state $z\in S$ such that $z\notin G\left(s\right)$ for all $s\in S$. In other words, there is no state $s\in S$ such that z is a predecessor of s under the inverse dynamics determined by G.

Now, assume for contradiction that F is surjective. Then, for every $z\in S$, there exists at least one state $s\in S$ such that $F\left(s\right)=z$. But this would imply that $s\in G\left(z\right)$, as G is the inverse function of F. However, this contradicts our initial assumption that $z\notin G\left(s\right)$ for all $s\in S$.

Therefore, our assumption that F is surjective must be false. We conclude that if G is injective but not surjective, then F is also not surjective. □

Remark 7. 

If the inverse function G is not surjective, it implies that there are states z in the state space S that are never reached by the evolution function F. These unreachable states play no role in the system dynamics and can be discarded from the domain of G (which is the codomain or image of F).

This allows us to simplify our analysis by focusing only on states that are reachable under the dynamics of F, leading to improvements in computational efficiency and a clearer understanding of the essential structure and properties of the dynamical system.

Theorem 15.11 

(Necessary and Sufficient Conditions for DIDS). Let $F:S\to S$ be a function and $G:S\to \mathcal{P}\left(S\right)$ be its inverse function. The following conditions are necessary and sufficient for $(S,F)$ to be a Discrete Inverse Dynamical System (DIDS):

(1) 

F is deterministic: $\forall s\in S,\exists !t\in S:F\left(s\right)=t$

(2) 

F is surjective: $\forall t\in S,\exists s\in S:F\left(s\right)=t$

These conditions imply:

(1) 

G is injective: $\forall a,b\in S:\left(G\right(a)=G(b)\Rightarrow a=b)$

(2) 

G is surjective: $\forall B\in \mathcal{P}\left(S\right),\exists A\in S:G\left(A\right)=B$

(3) 

G is exhaustive: $\forall s\in S,\exists n\in \mathbb{N}:G^n\left(s\right)=r$ where r is a root of G

Note: It is important to mention that the condition of multivalued injectivity, i.e., $G\left(a\right)\cap G\left(b\right)=\varnothing$ for all $a\ne b$, may not hold at certain isolated points, such as the point of contact of an attractor cycle. However, this does not invalidate the general applicability of the DIDS framework, as these violations are exceptional cases and do not represent the predominant behavior of the inverse function G.

Proof. 

⇒) Assume $(S,F)$ is a DIDS. We prove that conditions 1-2 hold, which imply conditions 3-5.

(1)

By the definition of a DIDS, F is deterministic.

(2)

By Theorem 15.23, if G is surjective, then F is surjective. Since G is surjective (condition 4), F is surjective.

(3)

By Theorem 15.22, if F is deterministic, then G is injective. Since F is deterministic (condition 1), G is injective.

(4)

By Theorem 15.23, if F is surjective, then G is surjective.

(5)

By Corollary 15.24, if F is deterministic and surjective, it is likely that G is exhaustive.

(⇐) Assume conditions 1-2 hold. We prove that $(S,F)$ is a DIDS, as conditions 1-2 imply conditions 3-5.

(1)

By condition 1, F is deterministic.

(2)

By condition 2, F is surjective.

(3)

By Theorem 15.22, if F is deterministic, then G is injective.

(4)

By Theorem 15.23, if F is surjective, then G is surjective.

(5)

By Corollary 15.24, if F is deterministic and surjective, it is likely that G is exhaustive.

Therefore, $(S,F)$ satisfies the definition of a DIDS. □

Theorem 15.12 

(Characterization of the Inverse Model). Let $(S,F)$ be a DIDS and $G:S\to \mathcal{P}\left(S\right)$ its inverse function. The inverse model $\mathcal{F}$ generated by G is an inverse forest that satisfies:

(1) 

Absence of anomalous cycles in each tree $T_i\in \mathcal{F}$:

$$\forall T_i\in \mathcal{F},\forall v_1,\dots ,v_k\in T_i:(v_1\ne v_k\to \neg ((v_1,v_2)\in E_i\wedge \dots \wedge (v_{k-1},v_k)\in E_i\wedge (v_k,v_1)\in E_i))$$

(2) 

Confluence of trajectories in each tree $T_i\in \mathcal{F}$:

$$\forall T_i\in \mathcal{F},\forall v,w\in T_i,\exists u\in T_i:(v⇝u)\wedge (w⇝u)$$

(3) 

Convergence to a unique attractor $A_i$ at the root of each tree $T_i\in \mathcal{F}$:

$$\forall T_i\in \mathcal{F},\forall v\in T_i,\exists n\in \mathbb{N}:G^n\left(v\right)\in A_i$$

if and only if F is deterministic and surjective.

Proof. 

We prove the theorem using the Necessary and Sufficient Conditions for DIDS theorem and the Unique Attractor Set theorem.

Step 1: Prove the forward implication. Assume $\mathcal{F}$ is an inverse forest satisfying properties (1)-(3). We want to show that F is deterministic and surjective.

By the Unique Attractor Set theorem, each tree $T_i\in \mathcal{F}$ converges to a unique attractor $A_i$. Let $A=\{A_1,\dots ,A_k\}$ be the set of all attractors in $\mathcal{F}$.

$$\forall T_i\in \mathcal{F},\exists !A_i\in A:\forall v\in T_i,\exists n\in \mathbb{N}:G^n\left(v\right)\in A_i$$

By the DIDS theorem, the existence of an inverse forest $\mathcal{F}$ with unique attractors implies that F is deterministic and surjective.

Step 2: Prove the backward implication. Assume F is deterministic and surjective. We want to show that the inverse model $\mathcal{F}$ generated by G satisfies properties (1)-(3).

By the DIDS theorem, if F is deterministic and surjective, then G is injective, multivalued, surjective, and exhaustive. This implies that the inverse model $\mathcal{F}$ generated by G is an inverse forest.

$$\mathcal{F}=\{T_1,\dots ,T_k\},\mathrm{where}\mathrm{each}{T}_i\mathrm{is}\mathrm{an}\mathrm{inverse}\mathrm{tree}$$

By the Unique Attractor Set theorem, each tree $T_i\in \mathcal{F}$ converges to a unique attractor $A_i$.

$$\forall T_i\in \mathcal{F},\exists !A_i\in A:\forall v\in T_i,\exists n\in \mathbb{N}:G^n\left(v\right)\in A_i$$

Therefore, $\mathcal{F}$ satisfies properties (1)-(3).

Conclusion: We have shown that the inverse model $\mathcal{F}$ generated by G is an inverse forest satisfying properties (1)-(3) if and only if F is deterministic and surjective. □

Theorem 15.13 

(Unique Inverse Forest Structure for DIDS). Let $(S,F)$ be a Discrete Dynamical System, where S is a countable state space and $F:S\to S$ is the deterministic and surjective evolution function. Let $G:S\to \mathcal{P}\left(S\right)$ be the analytic inverse of F, which is multivalued injective, surjective, and exhaustive. Let $\mathcal{F}=\{T_1,\dots ,T_k\}$ be the Inverse Algebraic Forest generated by G, where each $T_i$ is a tree.

Then, $\mathcal{F}$ is unique and each $T_i\in \mathcal{F}$ is a single connected component.

Proof. 

First, we prove that each $T_i$ is connected.

Suppose, for contradiction, that there exist two nodes $v_1,v_2\in V_i$ such that there is no sequence of edges connecting $v_1$ and $v_2$. This implies that $v_1$ and $v_2$ belong to two separate connected components, say $T_{i1}$ and $T_{i2}$, respectively.

Step 1: Exhaustiveness of G (Generalized to countable S) By the exhaustiveness property of G, for each node $v\in V_i$, there exists a finite sequence of applications of G that leads to a root node $r_i$. Formally:

$$\forall v\in V_i,\exists n\in \mathbb{N},\exists r_i\in V_i:(\mathrm{Root}\left(r_i\right)\wedge v\in G^n\left(r_i\right))$$

where $\mathrm{Root}\left(r_i\right)$ denotes that $r_i$ is a root node, and $G^n$ represents the n-fold composition of G with itself.

Let $r_{i1}$ and $r_{i2}$ be the root nodes of $T_{i1}$ and $T_{i2}$, respectively.

Step 2: Determinism and Surjectivity of F (Generalized to countable S) By the determinism of F, each node in $T_i$ has a unique child. By the surjectivity of F, each node in $T_i$, except for the root nodes, has a unique parent. Formally:

$$\forall v\in V_i\setminus \{r_{i1},r_{i2}\},\exists !u\in V_i:(u,v)\in E_i$$

Step 3: Contradiction We have shown that the existence of separate components $T_{i1}$ and $T_{i2}$ leads to a contradiction when F is deterministic and surjective, and G is exhaustive, even for a countable state space S.

Therefore, each $T_i$ must be a single connected component.

Now, we prove the uniqueness of $\mathcal{F}$ using the Path Uniqueness Theorem.

Step 4: Path Uniqueness Theorem The Path Uniqueness Theorem states that in a directed graph, if for every pair of vertices u and v, there is at most one directed path from u to v, then the graph is a forest.

In the context of our Inverse Algebraic Forest $\mathcal{F}$, this means that if for every pair of nodes $v_1,v_2\in V_i$ in each tree $T_i$, there is at most one sequence of edges from $v_1$ to $v_2$, then $\mathcal{F}$ is unique.

Step 5: Uniqueness of Paths in each $T_i$ Let $v_1,v_2\in V_i$ be any two nodes in $T_i$. Suppose there are two distinct sequences of edges from $v_1$ to $v_2$, denoted by $P_1$ and $P_2$.

Let u be the last common node of $P_1$ and $P_2$ before they diverge. Let $u_1$ and $u_2$ be the next nodes after u in $P_1$ and $P_2$, respectively.

By the determinism of F, u can have only one child. Therefore, $u_1=u_2$, contradicting the assumption that $P_1$ and $P_2$ are distinct paths.

Thus, there can be at most one path between any two nodes in each $T_i$.

Step 6: Application of Path Uniqueness Theorem By Step 5, each $T_i$ satisfies the condition of the Path Uniqueness Theorem. Therefore, $\mathcal{F}$ is unique.

Conclusion: We have shown that the Inverse Algebraic Forest $\mathcal{F}$ generated by G is unique and each tree $T_i\in \mathcal{F}$ is a single connected component, even when the state space S is countable. □

Theorem 15.14 

(Convergence to Attractors in DIDS). Let $(S,F)$ be a DIDS and $\mathcal{A}=\{A_1,\dots ,A_n\}$ be the set of attractors. Then:

(1) 

Each attractor $A_i\in \mathcal{A}$ is invariant under F: $\forall A_i\in \mathcal{A}:F\left(A_i\right)\subseteq A_i$

(2) 

Every state $s\in S$ converges to a unique attractor $A_s\in \mathcal{A}$: $\forall s\in S,\exists !A_s\in \mathcal{A}:{lim}_{n\to \infty }{F}^n\left(s\right)=A_s$

(3) 

The set of attractors $\mathcal{A}$ is globally attracting: $\forall s\in S,\exists A\in \mathcal{A}:{lim}_{n\to \infty }{F}^n\left(s\right)=A$

Proof. 

The proof leverages the structure of the inverse forest $\mathcal{F}$ and the properties of the inverse function G:

(1)

Invariance of attractors: By the definition of an attractor, $A_i$ is invariant under F.

(2)

Convergence to a unique attractor: Each $s\in S$ belongs to a unique tree $T_s$ in $\mathcal{F}$. By the Convergence Theorem, the trajectory of s converges to the attractor $A_s$ at the root of $T_s$.

(3)

Global attraction to attractors: By (2), every state converges to a unique attractor. Since $\mathcal{A}$ contains all attractors, it is globally attracting.

□

Corollary 15.1 

(Non-chaoticity of DIDS). No DIDS exhibits genuine chaotic behavior.

Proof. 

The proof follows from the existence of a well-defined inverse model with an invariant forest structure:

• Suppose a DIDS $(S,F)$ exhibits chaotic behavior.
• Then there exists sensitivity to initial conditions: $\exists ϵ>0,\forall \delta >0,\forall s\in S,\exists s^{\prime }\in S,\exists n\in \mathbb{N}:d(s,s^{\prime })<\delta \wedge d(F^n\left(s\right),F^n\left(s^{\prime }\right))>ϵ$
• However, by the Convergence to Attractors theorem, each state converges to a unique attractor determined by the inverse forest structure.
• This contradicts sensitivity to initial conditions.
• Therefore, no DIDS exhibits genuine chaotic behavior.

□

Theorem 15.15 

(Impossibility of Intrinsic Chaos in Deterministic Discrete Dynamical Systems). Intrinsic chaos, in the sense of non-periodic, non-converging trajectories, is impossible in any deterministic discrete dynamical system $(S,F)$ that satisfies the conditions for the existence of a unique inverse algebraic forest.

Proof. 

Let $(S,F)$ be a deterministic discrete dynamical system that satisfies the conditions for the existence of a unique inverse algebraic forest $\mathcal{F}=\{T_1,\dots ,T_k\}$ generated by the analytic inverse function G.

By the Impossibility of Infinite Cycles in AITs of DIDS theorem (15.8), each tree $T_i\in \mathcal{F}$ cannot contain any infinite cycles. Moreover, by the Convergence to Attractors in DIDS theorem, all trajectories in each $T_i$ converge to a unique attractor $A_i$.

Since $\mathcal{F}$ covers the entire state space S (due to the exhaustiveness of G), every trajectory in $(S,F)$ must converge to one of the attractors $A_1,\dots ,A_k$. Therefore, intrinsic chaos is impossible in $(S,F)$. □

Theorem 15.16 

(Impossibility of Intrinsic Chaos in Deterministic Discrete Dynamical Systems). Let $(S,F)$ be a deterministic discrete dynamical system (DIDS) satisfying the conditions for the existence of a unique inverse algebraic forest. Then, intrinsic chaos, in the sense of sensitivity to initial conditions, dense orbits, and topological mixing, is impossible in $(S,F)$.

Proof. 

Let $(S,F)$ be a DIDS satisfying the conditions for the existence of a unique inverse algebraic forest $\mathcal{F}=\{T_1,\dots ,T_k\}$ generated by the inverse analytic function G.

Step 1: By the Impossibility of Infinite Cycles in AITs of DIDS theorem (15.8), each tree $T_i\in \mathcal{F}$ cannot contain any infinite cycles. Formally:

$$\begin{array}{c}\hfill \forall T_i\in \mathcal{F},\nexists (v_1,v_2,\dots )\in T_i:(v_j\ne v_k,\forall j\ne k)\wedge (\forall n\in \mathbb{N},v_{n+1}\in G\left(v_n\right))\end{array}$$

Step 2: By the Convergence to Attractors in DIDS theorem, all trajectories in each $T_i$ converge to a unique attractor $A_i$. Formally:

$$\begin{array}{c}\hfill \forall T_i\in \mathcal{F},\exists !A_i\subseteq S:\forall v\in T_i,\exists n\in \mathbb{N}:G^n\left(v\right)\in A_i\end{array}$$

Step 3: Suppose, for contradiction, that $(S,F)$ exhibits intrinsic chaotic behavior. This implies at least one of the following:

• Sensitivity to initial conditions: $\exists ϵ>0,\forall \delta >0,\forall x\in S,\exists y\in S,\exists n\in \mathbb{N}:d(x,y)<\delta \wedge d(F^n\left(x\right),F^n\left(y\right))>ϵ$
• Dense orbits: $\forall x\in S,\forall ϵ>0,\exists y\in S,\exists n\in \mathbb{N}:d(F^n\left(y\right),x)<ϵ$
• Topological mixing: $\forall U,V\subseteq S$ open, $\exists n_0\in \mathbb{N},\forall n\ge n_0:F^n\left(U\right)\cap V\ne \varnothing$

Step 4: By Steps 1 and 2, all trajectories in $(S,F)$ converge to a unique finite attractor set in a finite number of steps, contradicting the sensitivity to initial conditions, dense orbits, and topological mixing properties of chaos.

Step 5: Therefore, the assumption that $(S,F)$ exhibits intrinsic chaotic behavior must be false. □

Remark 8 

(Understanding Chaos). In the context of discrete dynamical systems, chaos is typically characterized by three main properties:

(1) 

Sensitivity to initial conditions: Arbitrarily small differences in initial states lead to exponentially diverging trajectories over time.

(2) 

Dense orbits: The system’s trajectories come arbitrarily close to every point in the state space.

(3) 

Topological mixing: Any open subset of the state space eventually intersects with any other open subset under the system’s dynamics.

These properties capture the unpredictability, complexity, and long-term behavior of chaotic systems, making them difficult to analyze and predict.

Remark 9 

(Limitations in Approaching the Termination Problem). The document "Resolving the Collatz Conjecture: A Rigorous Proof Through Inverse Discrete Dynamical Systems and Algebraic Inverse Trees" presents a solid logical-deductive system for the study of discrete dynamical systems through the Theory of Inverse Discrete Dynamical Systems (TIDDS). Theorem 15.15 (Impossibility of Intrinsic Chaos in Deterministic Discrete Dynamical Systems) establishes that, under certain conditions, all trajectories in a deterministic discrete dynamical system converge to a unique attractor set, which has relevant implications for the termination problem.

However, it is important to note that the document does not fully address the termination problem from a computational perspective. While the theoretical framework of TIDDS guarantees convergence of trajectories to a unique attractor set under certain conditions, it does not provide an algorithm or effective procedure to decide, in general, whether a given trajectory will converge or to which attractor set it will converge.

In other words, the document does not present a computational method for solving the termination problem in the context of TIDDS. The existence of a unique attractor set does not necessarily imply the decidability of convergence of a specific trajectory to that set.

Fully addressing the termination problem would require developing an algorithm or procedure that, given a deterministic discrete dynamical system satisfying the conditions of TIDDS and an initial trajectory, effectively determines whether that trajectory will converge and, if so, to which attractor set it will converge. The document does not provide such an algorithm or procedure.

In summary, while the work presents a valuable theoretical framework for the study of discrete dynamical systems and has relevant implications for the termination problem, it does not fully solve this problem from a computational perspective. Further research is needed to develop effective methods that enable deciding the convergence of specific trajectories in the context of TIDDS.

Intuition and Key Implications: The impossibility of intrinsic chaos in deterministic discrete dynamical systems satisfying the conditions for a unique inverse algebraic forest is a significant result that challenges the conventional understanding of chaos in these systems. The proof relies on two key theorems: the Impossibility of Infinite Cycles in AITs of DIDS (15.8) and the Convergence to Attractors in DIDS.

The first theorem ensures that the inverse algebraic trees (AITs) in the forest cannot contain any infinite cycles, which rules out the possibility of non-periodic trajectories. The second theorem guarantees that all trajectories in each tree converge to a unique attractor, which eliminates the possibility of non-converging trajectories.

The proof works by leveraging the properties of the analytic inverse function G and the structure of the inverse algebraic forest $\mathcal{F}$. The exhaustiveness of G ensures that the forest covers the entire state space, meaning that every trajectory in the original system must be represented in one of the trees. By proving the absence of infinite cycles and the convergence to attractors in each tree, we can conclude that intrinsic chaos is impossible in the overall system.

The key implications of this theorem are:

• It challenges the traditional view that deterministic discrete dynamical systems can exhibit intrinsic chaotic behavior.
• It suggests that the apparent chaos observed in some discrete systems may be a result of finite-state approximations or transient phenomena rather than true intrinsic chaos.
• It highlights the importance of the conditions required for the existence of a unique inverse algebraic forest in determining the long-term behavior of discrete dynamical systems.
• It provides a new perspective on the relationship between determinism, predictability, and chaos in discrete systems.

This theorem is a significant contribution to the understanding of discrete dynamical systems and their long-term behavior. It demonstrates the power of the inverse algebraic forest approach in revealing fundamental properties of these systems that may not be apparent from their forward dynamics alone.

Remark 10. 

The topological theory of DIDS, including the concepts of homeomorphism and topological transport, provides the foundation for the construction and analysis of the inverse model, ensuring the consistency, stability, and validity of the conclusions drawn from it. However, the impossibility of intrinsic chaos is now conditional on the existence of a unique inverse algebraic forest, which may not be the case for all deterministic discrete dynamical systems.

15.1. Most Remarkable Finding

The most surprising finding is that every deterministic discrete dynamical system that satisfies the conditions for the existence of a unique inverse algebraic forest is guaranteed to converge to a set of attractors, excluding the possibility of chaotic behavior. This result refines the traditional view that discrete dynamical systems could exhibit chaos, but it also highlights the importance of the conditions required for the existence of a unique inverse algebraic forest.

Figure 2. Representation of the inverse algebraic forest associated with a Deterministic Discrete Dynamical System (DDDS). Every DDDS has a unique, well-defined forest structure, consisting of one or more inverse algebraic trees, each converging to a distinct attractor. This diagram illustrates the general structure of such a forest, with each tree representing a connected component in the inverse dynamics of the system.

Figure 2. Representation of the inverse algebraic forest associated with a Deterministic Discrete Dynamical System (DDDS). Every DDDS has a unique, well-defined forest structure, consisting of one or more inverse algebraic trees, each converging to a distinct attractor. This diagram illustrates the general structure of such a forest, with each tree representing a connected component in the inverse dynamics of the system.

15.2. The Logistic Model as a DIDS

Theorem 15.17 

(Logistic Model as a DIDS). The discretized logistic model $(X_d,F_d)$, where $X_d$ is a finite subset of $[0,1]$ and $F_d:X_d\to X_d$ is the discretized version of the logistic function $F\left(x\right)=rx(1-x)$, is a Discrete Inverse Dynamical System (DIDS) for any $r\in (0,4]$.

Corollary 15.2 

(Convergence in the Discretized Logistic Model). In the discretized logistic model $(X_d,F_d)$, the apparently chaotic behavior observed for certain values of r in the continuous logistic model is not truly chaotic. According to the principles of the Theory of Inverse Discrete Dynamical Systems (TIDDS), all trajectories in the discretized model eventually converge to an attractor set, which may consist of fixed points, periodic orbits, or more complex structures.

Theorem 15.18. 

Let $F\left(x\right)=rx(1-x)$ be the logistic function with $r\ge 1$ and $x\in [0,1]$. Assume that the model reaches a fixed point and not a period. Then, the only admissible fixed point of the logistic model is $x=0$.

Proof. 

Step 1: Define the logistic function F.

$$\begin{array}{c}\hfill \forall x\in [0,1]:F\left(x\right)=rx(1-x),\mathrm{where}r\ge 1\end{array}$$

Step 2: Find the fixed points of F.

$$\begin{array}{c}\hfill \mathrm{Let}{x}^{*}\mathrm{be}\mathrm{a}\mathrm{fixed}\mathrm{point}\mathrm{of}F.\mathrm{Then}\text{:}\\ \hfill F\left(x^{*}\right)=x^{*}\\ \hfill r{x}^{*}(1-x^{*})=x^{*}\\ \hfill r{x}^{*}-r{x}^{*2}=x^{*}\\ \hfill r{x}^{*2}-(r-1)x^{*}=0\\ \hfill x^{*}(r{x}^{*}-(r-1))=0\\ \hfill \mathrm{Therefore},\mathrm{the}\mathrm{fixed}\mathrm{points}\mathrm{are}\text{:}\\ \hfill x_1^{*}=0\mathrm{and}{x}_2^{*}=1-\frac{1}{r}\end{array}$$

Step 3: Find the ancestors of each fixed point.

$$\begin{array}{c}\hfill \mathrm{For}{x}_1^{*}=0:\\ \hfill F\left(x\right)=0\Rightarrow x=0\mathrm{or}x=1\\ \hfill \mathrm{For}{x}_2^{*}=1-\frac{1}{r}:\\ \hfill F\left(x\right)=1-\frac{1}{r}\Rightarrow x=1-\frac{1}{r}\mathrm{or}x=\frac{1}{r}\end{array}$$

Step 4: Exclude inadmissible fixed points and ancestors.

$$\begin{array}{c}\hfill \mathrm{For}1<r\le 2:\frac{1}{r}\notin [0,1],\mathrm{so}{x}_2^{*}\mathrm{and}\mathrm{its}\mathrm{ancestor}\mathrm{are}\mathrm{inadmissible}.\\ \hfill \mathrm{For}2<r\le 4:x_2^{*}\mathrm{is}\mathrm{admissible},\mathrm{but}\mathrm{its}\mathrm{only}\mathrm{ancestor}\mathrm{is}\mathrm{itself},\mathrm{which}\mathrm{is}\mathrm{trivial}.\end{array}$$

Step 5: Conclude that $x=0$ is the only admissible fixed point with non-trivial ancestors.

$$\begin{array}{c}\hfill \therefore \forall x\in [0,1]:\left(F\right(x)=x\wedge x\mathrm{has}\mathrm{non}-\mathrm{trivial}\mathrm{ancestors})\Rightarrow x=0\end{array}$$

□

Theorem 15.19. 

Let $F\left(x\right)=rx(1-x)$ be the logistic function with $0<r\le 1$ and $x\in [0,1]$. Assume that the model reaches a fixed point and not a period. Then, the only admissible fixed point of the logistic model is $x=0$.

Proof. 

Step 1: Define the logistic function F.

$$\begin{array}{c}\hfill \forall x\in [0,1]:F\left(x\right)=rx(1-x),\mathrm{where}0<r\le 1\end{array}$$

Step 2: Find the fixed points of F.

$$\begin{array}{c}\hfill \mathrm{Let}{x}^{*}\mathrm{be}\mathrm{a}\mathrm{fixed}\mathrm{point}\mathrm{of}F.\mathrm{Then}\text{:}\\ \hfill F\left(x^{*}\right)=x^{*}\\ \hfill r{x}^{*}(1-x^{*})=x^{*}\\ \hfill r{x}^{*}-r{x}^{*2}=x^{*}\\ \hfill r{x}^{*2}-(r-1)x^{*}=0\\ \hfill x^{*}(r{x}^{*}-(r-1))=0\\ \hfill \mathrm{Therefore},\mathrm{the}\mathrm{fixed}\mathrm{points}\mathrm{are}\text{:}\\ \hfill x_1^{*}=0\mathrm{and}{x}_2^{*}=1-\frac{1}{r}\end{array}$$

Step 3: Verify if the fixed points are within the interval $[0,1]$.

$$\begin{array}{c}\hfill \mathrm{For}{x}_1^{*}=0:\\ \hfill 0\in [0,1],\mathrm{so}{x}_1^{*}\mathrm{is}\mathrm{admissible}.\\ \hfill \mathrm{For}{x}_2^{*}=1-\frac{1}{r}:\\ \hfill \mathrm{If}0<r<1,\mathrm{then}1-\frac{1}{r}>1,\mathrm{so}{x}_2^{*}\notin [0,1]\mathrm{and}\mathrm{is}\mathrm{not}\mathrm{admissible}.\\ \hfill \mathrm{If}r=1,\mathrm{then}1-\frac{1}{r}=0,\mathrm{which}\mathrm{coincides}\mathrm{with}{x}_1^{*}\mathrm{and}\mathrm{is}\mathrm{admissible}.\end{array}$$

Step 4: Find the ancestors of each admissible fixed point.

$$\begin{array}{c}\hfill \mathrm{For}{x}_1^{*}=0:\\ \hfill F\left(x\right)=0\Rightarrow x=0\mathrm{or}x=1\\ \hfill \mathrm{Therefore},\mathrm{the}\mathrm{ancestors}\mathrm{of}{x}_1^{*}\mathrm{are}0\mathrm{and}1.\end{array}$$

Step 5: Exclude inadmissible ancestors.

$$\begin{array}{c}\hfill \mathrm{The}\mathrm{ancestor}x=1\mathrm{of}{x}_1^{*}\mathrm{is}\mathrm{admissible},\mathrm{as}1\in [0,1].\end{array}$$

Step 6: Conclude that $x=0$ is the only admissible fixed point with non-trivial ancestors for $0<r\le 1$.

$$\begin{array}{c}\hfill \therefore \forall x\in [0,1]:\left(F\right(x)=x\wedge x\mathrm{has}\mathrm{non}-\mathrm{trivial}\mathrm{ancestors})\Rightarrow x=0\end{array}$$

□

Theorem 15.20 

(Uncountable State Space Theorem). Let $(S,F)$ be a discrete dynamical system, where $S=[a,b]\subset \mathbb{R}$ is an uncountably infinite state space and $F:S\to S$ is a deterministic and surjective evolution function. Let $G:S\to \mathcal{P}\left(S\right)$ be the inverse function of F, where $\mathcal{P}\left(S\right)$ denotes the power set of S. Then, the inverse algebraic forest associated with $(S,F)$ has an uncountable number of trees, each converging to a unique attractor set.

Proof. 

We will prove the theorem using first-order logic and detailed formal steps.

Step 1: Define the uncountable state space S.

$$\begin{array}{cc}\hfill S& =[a,b]\subset \mathbb{R}\hfill \\ \hfill \left|S\right|& =2^{{\aleph }_0}>{\aleph }_0\hfill \end{array}$$

Step 2: State the properties of the evolution function F.

$$\begin{array}{cc}\hfill \forall s\in S,\exists !t\in S& :F\left(s\right)=t\left(\mathrm{Determinism}\right)\hfill \\ \hfill \forall t\in S,\exists s\in S& :F\left(s\right)=t\left(\mathrm{Surjectivity}\right)\hfill \end{array}$$

Step 3: Define the inverse function G.

$$\forall s\in S:G\left(s\right)=\{t\in S:F(t)=s\}$$

Step 4: Prove that G is multivalued injective, surjective, and exhaustive.

$$\begin{array}{cc}\hfill \forall a,b\in S& :(a\ne b\to G(a)\cap G(b)=\varnothing )(\mathrm{Multivalued}\mathrm{injectivity})\hfill \\ \hfill \forall B\in \mathcal{P}\left(S\right),\exists A\in S& :G\left(A\right)=B\left(\mathrm{Surjectivity}\right)\hfill \\ \hfill \forall s\in S,\exists n\in \mathbb{N}& :G^n\left(s\right)=r\left(\mathrm{Exhaustiveness}\right)\hfill \end{array}$$

where r is a root of the inverse algebraic forest.

Step 5: Construct the inverse algebraic forest $\mathcal{F}$.

$$\mathcal{F}=\{T_s:s\in S\}$$

where $T_s$ is the inverse algebraic tree rooted at the attractor set containing s.

Step 6: Prove that each tree $T_s$ in $\mathcal{F}$ converges to a unique attractor set.

$$\begin{array}{cc}\hfill \forall s\in S,\exists !A_s\subset S& :A_s\mathrm{is}\mathrm{an}\mathrm{attractor}\mathrm{set}\hfill \\ \hfill \forall t\in T_s,\exists n\in \mathbb{N}& :G^n\left(t\right)\in A_s\hfill \end{array}$$

Step 7: Conclude that the inverse algebraic forest $\mathcal{F}$ has an uncountable number of trees, each converging to a unique attractor set.

$$\left|\mathcal{F}\right|=\left|S\right|=2^{{\aleph }_0}>{\aleph }_0$$

□

Theorem 15.21 

(Discretization Theorem for the Logistic Model). Let $F_r:[0,1]\to [0,1]$ be the continuous logistic function defined as $F_r\left(x\right)=rx(1-x)$, where $r\in [0,4]$ is a parameter. Let $S_d\subset [0,1]$ be a finite discretization of the state space, and let $F_d:S_d\to S_d$ be the discretized version of $F_r$. Then, for each fixed value of r, the number of attractors in the discretized model $(S_d,F_d)$ is finite, while the continuous model $([0,1],F_r)$ has a single attractor (fixed point, limit cycle, or chaotic attractor).

Proof. 

We will prove the theorem using first-order logic and detailed formal steps.

Step 1: Define the continuous logistic function $F_r$.

$$\begin{array}{cc}\hfill \forall x\in [0,1],\forall r\in [0,4]& :F_r\left(x\right)=rx(1-x)\hfill \\ \hfill \forall x\in [0,1],\forall r\in [0,4]& :F_r\left(x\right)\in [0,1]\left(\mathrm{Closure}\mathrm{property}\right)\hfill \end{array}$$

Step 2: Define the discretized state space $S_d$ and the discretized logistic function $F_d$.

$$\begin{array}{cc}\hfill S_d& =\{x_1,\cdots ,x_n\}\subset [0,1],\left|S_d\right|=n<\infty \hfill \\ \hfill \forall x\in S_d,\forall r\in [0,4]& :F_d\left(x\right)=arg\underset{y\in S_d}{min}|F_r\left(x\right)-y|\hfill \\ \hfill \forall x\in S_d,\forall r\in [0,4]& :F_d\left(x\right)\in S_d\left(\mathrm{Closure}\mathrm{property}\right)\hfill \end{array}$$

Step 3: Prove that for each fixed r, the continuous model $([0,1],F_r)$ has a single attractor.

$$\begin{array}{cc}\hfill \forall r\in [0,1]& :\exists !x^{*}=0:F_r\left(x^{*}\right)=x^{*}\left(\mathrm{Fixed}\mathrm{point}\right)\hfill \\ \hfill \forall r\in (1,3]& :\exists !x^{*}=1-\frac{1}{r}:F_r\left(x^{*}\right)=x^{*}\left(\mathrm{Fixed}\mathrm{point}\right)\hfill \\ \hfill \forall r\in (3,3.57]& :\exists !\{x_1^{*},x_2^{*}\}:F_r\left(x_1^{*}\right)=x_2^{*}\wedge F_r\left(x_2^{*}\right)=x_1^{*}\left(\mathrm{Limit}\mathrm{cycle}\right)\hfill \\ \hfill \forall r\in (3.57,4]& :\exists !A_r\subset [0,1]:A_r\mathrm{is}\mathrm{a}\mathrm{chaotic}\mathrm{attractor}\hfill \end{array}$$

where $A_r$ is the unique attractor for each r.

Step 4: Prove that for each fixed r, the discretized model $(S_d,F_d)$ has a finite number of attractors.

$$\begin{array}{cc}\hfill \forall r\in [0,4],\exists k\in \mathbb{N},\exists A_1,\cdots ,A_k\subset S_d& :A_1,\cdots ,A_k\mathrm{are}\mathrm{attractors}\hfill \\ \hfill \forall x\in S_d,\exists i\in \{1,\cdots ,k\},\exists n\in \mathbb{N}& :F_d^n\left(x\right)\in A_i\hfill \end{array}$$

where k is finite since $S_d$ is finite.

To prove this, we use the following reasoning:

$$\begin{array}{cc}\hfill |S_d|=n<\infty & \Rightarrow |\mathcal{P}\left(S_d\right)|=2^n<\infty \hfill \\ \hfill \forall r\in [0,4],\forall A\subset S_d& :(A\mathrm{is}\mathrm{an}\mathrm{attractor}\Rightarrow A\in \mathcal{P}\left(S_d\right))\hfill \\ \hfill \therefore \forall r\in [0,4]& :|\{A\subset S_d:A\mathrm{is}\mathrm{an}\mathrm{attractor}\}|\le 2^n<\infty \hfill \end{array}$$

Step 5: Conclude that for each fixed r, the number of attractors in the discretized model is finite, while the continuous model has a single attractor.

Therefore, we have proven that the discretization of the continuous logistic model results in a finite number of attractors for each fixed parameter value r, while the continuous model exhibits a single attractor (fixed point, limit cycle, or chaotic attractor) for each r. □

Remark 11 

(Reconciling the Logistic Model with TIDDS). The logistic model, a well-known example of a continuous dynamical system exhibiting complex behaviors, presents challenges when reconciled with the Theory of Inverse Discrete Dynamical Systems (TIDDS). The Discretization Theorem for the Logistic Model highlights the fundamental differences between the discretized and continuous versions of the model, particularly in terms of the number and nature of attractors.

The Uncountable State Space Theorem states that when the state space S is an uncountably infinite set, such as the interval $[a,b]$, the discretized model $(S,F_d)$ has an uncountably infinite number of attractors. This result stands in stark contrast to the continuous logistic model $([0,1],F_r)$, which has a single attractor (fixed point, limit cycle, or chaotic attractor) for each fixed parameter value r.

The implications of this contrast are as follows:

(1) 

Disparity in Attractor Count:The uncountably infinite number of attractors in the discretized model with $S=[a,b]$ differs significantly from the single attractor in the continuous logistic model. This disparity suggests that the discretization process may introduce artifacts and complexities that are not inherent to the original continuous system.

(2) 

Limitations of Discretization:The presence of uncountably infinite attractors in the discretized model with $S=[a,b]$ highlights the limitations of discretization in capturing the true dynamics of the continuous logistic model. The discretization process may alter the system’s properties and introduce behaviors that are not representative of the original continuous system.

(3) 

Extension of TIDDS:To effectively reconcile the logistic model with TIDDS, it is necessary to extend the theory to accommodate the uncountably infinite state space and the unique properties of continuous dynamical systems. This extension would involve developing a more general formulation of TIDDS that can handle both countable and uncountable state spaces and account for the differences in attractor count and nature between discretized and continuous models.

(4) 

Reinterpretation of Discretization Results:The Uncountable State Space Theorem and the Discretization Theorem for the Logistic Model suggest that the results obtained from the discretized model with $S=[a,b]$, such as the presence of uncountably infinite attractors, may not directly translate to the continuous logistic model. The interpretation of these results should be done cautiously, considering the limitations of discretization and the fundamental differences between the discretized and continuous versions of the model.

In conclusion, the contrast between the uncountably infinite attractors in the discretized model with $S=[a,b]$ and the single attractor in the continuous logistic model highlights the challenges in reconciling the logistic model with TIDDS. The Uncountable State Space Theorem and the Discretization Theorem for the Logistic Model underscore the need for extending TIDDS to accommodate continuous state spaces and account for the differences between discretized and continuous models. Further research in this direction will contribute to a more comprehensive understanding of the logistic model and its relationship to the theory of inverse discrete dynamical systems, while also providing insights into the limitations and interpretation of discretization results.

15.3. Proof of the Collatz Conjecture

Definition 15.5. 

Let $C:\mathbb{N}\to \mathbb{N}$ be the Collatz function defined as:

$$C\left(n\right)=\left\{\begin{array}{cc}\frac{n}{2}\hfill & \mathit{if}n\mathit{is}\mathit{even}\hfill \\ 3n+1\hfill & \mathit{if}n\mathit{is}\mathit{odd}\hfill \end{array}\right.$$

Definition 15.6 

(Inverse Collatz Function). Let $\mathbb{N}$ be the set of natural numbers. The multivalued inverse function of Collatz $C^{-1}:\mathbb{N}\to \mathcal{P}\left(\mathbb{N}\right)$ is defined for every $n\in \mathbb{N}$ as:

$$C^{-1}\left(n\right)=\left\{\begin{array}{cc}\left\{2n\right\}\hfill & \mathit{if}n\not\equiv 4(mod6)\hfill \\ \{2n,\frac{n-1}{3}\}\hfill & \mathit{if}n\equiv 4(mod6)\hfill \end{array}\right.$$

where $\mathcal{P}\left(\mathbb{N}\right)$ denotes the power set of $\mathbb{N}$.

Note:It is important to mention that multivalued injectivity is not satisfied at the point of contact $pc=1$ for the Collatz function, as $G\left(1\right)\cap G\left(2\right)=\left\{4\right\}\ne \varnothing$. However, this violation is an exceptional case and does not invalidate the overall applicability of the inverse Collatz function in the context of the DIDS framework.

Theorem 15.22. 

Let $F:S\to S$ be a function and $G:S\to \mathcal{P}\left(S\right)$ be its inverse function. Then:

$\left(F\mathrm{is}\mathrm{deterministic}\right)\iff \left(G\mathrm{is}\mathrm{multivalued}\mathrm{injective}\mathrm{except}\mathrm{at}\mathrm{isolated}\mathrm{points}\right)$

Proof. 

Let’s define the terms using first-order logic:

Step 1: Define determinism of F.

$$\forall s\in S,\exists !t\in S:F\left(s\right)=t$$

Step 2: Define multivalued injectivity of G except at isolated points.

$$\exists I\subseteq S,\left|I\right|<\infty :\left(\forall a,b\in S\setminus I:(a\ne b\to G(a)\cap G(b)=\varnothing )\right)\wedge$$

$$\left(\forall x\in I,\exists y,z\in I:G\left(x\right)\cap G\left(y\right)\ne \varnothing \wedge G\left(x\right)\cap G\left(z\right)\ne \varnothing \right)$$

$(\Rightarrow )$ Suppose F is deterministic. We will prove that G is multivalued injective except at isolated points.

Let $I=\{x\in S:\exists y,z\in S,y\ne z,G(x)\cap G(y)\ne \varnothing \wedge G(x)\cap G(z)\ne \varnothing \}$. By the definition of G, I must be a finite set.

For any $a,b\in S\setminus I$ with $a\ne b$, we can show that $G\left(a\right)\cap G\left(b\right)=\varnothing$ using the same argument as in the original proof.

For any $x\in I$, there must exist $y,z\in S$ with $y\ne z$ such that $G\left(x\right)\cap G\left(y\right)\ne \varnothing$ and $G\left(x\right)\cap G\left(z\right)\ne \varnothing$ by the definition of I.

Therefore, G is multivalued injective except at the isolated points in I.

$(\Leftarrow )$ Suppose G is multivalued injective except at isolated points. We will prove that F is deterministic.

Let $s\in S$. Assume, for contradiction, that there exist $t_1,t_2\in S$ with $t_1\ne t_2$ such that $F\left(s\right)=t_1$ and $F\left(s\right)=t_2$. Then, $s\in G\left(t_1\right)$ and $s\in G\left(t_2\right)$, which implies $G\left(t_1\right)\cap G\left(t_2\right)\ne \varnothing$. However, this contradicts the multivalued injectivity of G except at isolated points, since $t_1$ and $t_2$ cannot be isolated points. Therefore, there cannot exist two distinct $t_1,t_2\in S$ such that $F\left(s\right)=t_1$ and $F\left(s\right)=t_2$, proving that F is deterministic.

$$\mathrm{Thus},\mathrm{we}\mathrm{have}\mathrm{shown}\mathrm{that}\left(F\mathrm{is}\mathrm{deterministic}\right)\iff$$

$$\left(G\mathrm{is}\mathrm{multivalued}\mathrm{injective}\mathrm{except}\mathrm{at}\mathrm{isolated}\mathrm{points}\right).$$

□

Theorem 15.23. 

Let $F:S\to S$ be a function and $G:S\to \mathcal{P}\left(S\right)$ be its inverse function. Then:

$\left(F\mathit{is}\mathit{surjective}\right)\Rightarrow \left(G\mathit{is}\mathit{surjective}\right)\Rightarrow \left(G\mathit{is}\mathit{exhaustive}\right)$

Proof. 

Let’s define the terms using first-order logic:

Step 1: Define surjectivity of F.

$$\forall t\in S,\exists s\in S:F\left(s\right)=t$$

Step 2: Define surjectivity of G.

$$\forall B\in \mathcal{P}\left(S\right),\exists A\in S:G\left(A\right)=B$$

Step 3: Define exhaustiveness of G.

$$\forall s\in S,\exists n\in \mathbb{N}:s\in G^n\left(F\left(s\right)\right)$$

where $G^n$ denotes the n-fold composition of G with itself.

$(\Rightarrow )$ Suppose F is surjective. We will prove that G is surjective.

Let $B\in \mathcal{P}\left(S\right)$. By the surjectivity of F, for each $t\in B$, there exists $s\in S$ such that $F\left(s\right)=t$. Let $A=\{s\in S:F(s)\in B\}$. Then,

$$\begin{array}{cc}\hfill G\left(A\right)& =\{t\in S:t\in G(s)\mathrm{for}\mathrm{some}s\in A\}\hfill \\ \hfill & =\{t\in S:F(t)\in B\}\hfill \\ \hfill & =B\hfill \end{array}$$

Thus, G is surjective.

$(\Rightarrow )$ Suppose G is surjective. We will prove that G is exhaustive.

Let $s\in S$. Since G is surjective, there exists $A\in S$ such that $G\left(A\right)=\left\{s\right\}$. This implies that $s\in G\left(A\right)$, which means $s\in G^1\left(F\left(A\right)\right)$. Therefore, G is exhaustive.

Thus, we have shown that $\left(F\mathrm{is}\mathrm{surjective}\right)\Rightarrow \left(G\mathrm{is}\mathrm{surjective}\right)\Rightarrow \left(G\mathrm{is}\mathrm{exhaustive}\right)$. □

Theorem 15.24. 

Corollary:Let $F:S\to S$ be a function and $G:S\to \mathcal{P}\left(S\right)$ be its inverse function. If F is deterministic and surjective, then G is exhaustive.

Proof. 

Step 1: Define determinism of F.

$$\forall s\in S,\exists !t\in S:F\left(s\right)=t$$

Step 2: Define surjectivity of F.

$$\forall t\in S,\exists s\in S:F\left(s\right)=t$$

Step 3: Define exhaustiveness of G.

$$\forall s\in S,\exists n\in \mathbb{N}:s\in G^n\left(F\left(s\right)\right)$$

where $G^n$ denotes the n-fold composition of G with itself.

Assume that F is deterministic and surjective.

Step 4: Prove that for any $s\in S$, there exists a finite sequence of applications of G that leads to s. Let $s\in S$. Since F is surjective, there exists $t\in S$ such that $F\left(t\right)=s$. Since F is deterministic, there exists a unique sequence $(t_0,t_1,\cdots ,t_n)$ such that $t_0=t$ and $F\left(t_i\right)=t_{i+1}$ for all $0\le i<n$, and $t_n=s$.

By the definition of G, we have:

$$s=t_n\in G\left(t_{n-1}\right)$$

$$t_{n-1}\in G\left(t_{n-2}\right)$$

$$\cdots$$

$$t_1\in G\left(t_0\right)$$

Therefore, $s\in G^n\left(t\right)$, which implies that $s\in G^n\left(F\left(s\right)\right)$.

Step 5: Conclude that G is exhaustive. Since Step 4 holds for all $s\in S$, we have proven that:

$$\forall s\in S,\exists n\in \mathbb{N}:s\in G^n\left(F\left(s\right)\right)$$

Therefore, if F is deterministic and surjective, then G is exhaustive. □

Theorem 15.25 

(Collatz System as a DIDS). $(\mathbb{N},C)$ is a Discrete Inverse Dynamical System (DIDS) with inverse function $C^{-1}$.

Proof. 

We have already proved that C is deterministic and surjective. By the necessary and sufficient conditions for a function F being deterministic and surjective, it follows that its inverse function G is multivalued injective, surjective, and exhaustive.

In the context of the Collatz system:

(1)

C is deterministic: For each $n\in \mathbb{N}$, $C\left(n\right)$ is uniquely defined based on the value of n modulo 2.

(2)

C is surjective: For each $m\in \mathbb{N}$, there exists an $n\in \mathbb{N}$ such that $C\left(n\right)=m$, based on the value of m modulo 2 and 6.

Therefore, by the necessary and sufficient conditions, the inverse function $C^{-1}$ satisfies:

(1)

$C^{-1}$ is multivalued injective: For any $m,n\in \mathbb{N}$, if $m\ne n$, then $C^{-1}\left(m\right)\cap C^{-1}\left(n\right)=\varnothing$.

(2)

$C^{-1}$ is surjective: For each $n\in \mathbb{N}$, there exists an $m\in \mathbb{N}$ such that $n\in C^{-1}\left(m\right)$.

(3)

$C^{-1}$ is exhaustive: For each $n\in \mathbb{N}$, there exists a $k\in \mathbb{N}$ such that the k-fold composition of $C^{-1}$ applied to n contains the minimum element of the attractor cycle.

Thus, $(\mathbb{N},C)$ is a DIDS with inverse function $C^{-1}$. □

Theorem 15.26 

(Convergence of Attraction Points in the Generalized Collatz Conjecture). Let $C:\mathbb{N}\to \mathbb{N}$ be the Collatz function defined as:

$$C\left(x\right)=\left\{\begin{array}{cc}\frac{x}{2}\hfill & \mathit{if}x\equiv 0(mod2)\hfill \\ 3x+1\hfill & \mathit{if}x\equiv 1(mod2)\hfill \end{array}\right.$$

Then, all possible attraction points in the Collatz Conjecture converge to one of the two attraction cycles: $(1,4,2)$ with the point of contact being 1, or $\left(0\right)$ with the point of contact being 0.

Proof. 

Step 1: Define the set of possible attraction points A as:

$$A=\{x\in \mathbb{N}:x\equiv 0,1,2,3,4,5(mod6\left)\right\}$$

Step 2: For each $x\in A$, apply the Collatz function C iteratively until a value repeats, forming a cycle.

Step 3: Verify the convergence of each attraction point:

• For $x=0$: $C\left(0\right)=0$, forming the trivial cycle $\left(0\right)$ of length 1.
• For $x=1$: $C\left(1\right)=4$, $C\left(4\right)=2$, $C\left(2\right)=1$, forming the cycle $(1,4,2)$ of length 3.
• For $x=2$: $C\left(2\right)=1$, $C\left(1\right)=4$, $C\left(4\right)=2$, forming the cycle $(1,4,2)$ of length 3.
• For $x=3$: $C\left(3\right)=10$, $C\left(10\right)=5$, $C\left(5\right)=16$, $C\left(16\right)=8$, $C\left(8\right)=4$, $C\left(4\right)=2$, $C\left(2\right)=1$, converging to the cycle $(1,4,2)$.
• For $x=4$: $C\left(4\right)=2$, $C\left(2\right)=1$, $C\left(1\right)=4$, forming the cycle $(1,4,2)$ of length 3.
• For $x=5$: $C\left(5\right)=16$, $C\left(16\right)=8$, $C\left(8\right)=4$, $C\left(4\right)=2$, $C\left(2\right)=1$, converging to the cycle $(1,4,2)$.

Step 4: Conclude that all possible attraction points $x\in A$ converge to one of the two cycles: $(1,4,2)$ with the point of contact being 1, or $\left(0\right)$ with the point of contact being 0.

Formally, we can express this convergence using first-order logic:

$$\forall x\in A,\exists n\in \mathbb{N}:C^n\left(x\right)\in \{1,4,2\}\mathrm{or}{C}^n\left(x\right)=0$$

where $C^n$ denotes the n-fold composition of C with itself. □

Theorem 15.27 

(Sufficiency of Modulo 6 Representatives). Let $C:\mathbb{N}\to \mathbb{N}$ be the Collatz function defined as:

$$C\left(n\right)=\left\{\begin{array}{cc}\frac{n}{2}\hfill & \mathit{if}n\equiv 0(mod2)\hfill \\ 3n+1\hfill & \mathit{if}n\equiv 1(mod2)\hfill \end{array}\right.$$

To determine all possible attracting cycles in the Collatz Conjecture, it is sufficient to consider the minimum values of each equivalence class modulo 6, i.e., the set $\{0,1,2,3,4,5\}$.

Proof. 

We will proceed by cases, showing that for each equivalence class modulo 6, all values converge to an attracting cycle initiated by its minimum representative.

Case 1:$n\equiv 0(mod6)$

Let $n=6k$ for some $k\in \mathbb{N}$. Then:

$$\begin{array}{cc}\hfill C\left(n\right)& =C\left(6k\right)\hfill \\ \hfill & =3k\hfill \\ \hfill & \equiv 0(mod6)\hfill \end{array}$$

Therefore, all values in this class converge to the trivial attractor $\left\{0\right\}$.

Case 2:$n\equiv 1(mod6)$

Let $n=6k+1$ for some $k\in \mathbb{N}$. Then:

$$\begin{array}{cc}\hfill C\left(n\right)& =C(6k+1)\hfill \\ \hfill & =3(6k+1)+1\hfill \\ \hfill & =18k+4\hfill \\ \hfill & \equiv 4(mod6)\hfill \end{array}$$

Next, the sequence will continue as:

$$\begin{array}{cc}\hfill C(18k+4)& =9k+2\hfill \\ \hfill & \equiv 2(mod6)\hfill \\ \hfill C(9k+2)& =3(9k+2)+1\hfill \\ \hfill & =27k+7\hfill \\ \hfill & \equiv 1(mod6)\hfill \end{array}$$

Thus, all values in this class converge to the cycle $\{1,4,2\}$.

The cases for $n\equiv 2,3,4,5(mod6)$ can be demonstrated similarly, showing convergence to $\{1,4,2\}$.

In conclusion, to find all possible attracting cycles, it is sufficient to consider the minimum representatives of the equivalence classes modulo 6, $\{0,1,2,3,4,5\}$, as all other values in each class will converge to the attractors found from these representatives. □ □

Intuition and Key Implications: The proof of the Convergence of Attraction Points in the Collatz Conjecture relies on the explicit verification of the convergence behavior for each possible attraction point. By applying the Collatz function iteratively to each point, we can observe the formation of cycles or the convergence to known cycles.

The proof works by systematically checking all possible residue classes modulo 6, which cover all the possible attraction points. This is because the Collatz function behaves differently for even and odd numbers, and the residue classes modulo 6 provide a natural partitioning of the natural numbers that captures this behavior.

The key implications of this theorem are:

• It demonstrates that the Collatz Conjecture holds for all possible attraction points, not just for specific initial values.
• It reveals the existence of two distinct attraction cycles: the trivial cycle $\left(0\right)$ and the non-trivial cycle $(1,4,2)$.
• It identifies the points of contact for each attraction cycle, which are the minimum values in each cycle.
• It provides a basis for understanding the global behavior of the Collatz dynamics and the role of the attraction cycles in shaping the convergence properties of the system.

The convergence of all possible attraction points to one of the two cycles is a crucial step in the overall proof of the Collatz Conjecture. It demonstrates the universality of the convergence behavior and the central role played by the attraction cycles in the long-term dynamics of the Collatz system.

Moreover, the identification of the points of contact for each cycle is significant, as these points serve as the entry points for the convergence of trajectories. Understanding the properties of these points of contact and their relationship to the attraction cycles is key to unraveling the global structure of the Collatz dynamics.

In summary, this theorem provides a rigorous verification of the convergence behavior of all possible attraction points in the Collatz Conjecture, while also offering insights into the fundamental role of the attraction cycles and their points of contact in shaping the overall dynamics of the system.

Theorem 15.28 

(Uniqueness of the Collatz Attractor). The Collatz dynamical system $(S,C)$, where $S=\mathbb{N}$ and $C:S\to S$ is the Collatz function, has a unique attractor set consisting of two disjoint cycles: $\{1,4,2\}$ and $\left\{0\right\}$.

Proof. 

We will use the Collatz system’s properties and the theorems we’ve proven to show that it has a unique attractor set.

Step 1: Apply the unique inverse algebraic forest theorem.

• By the theorem, since $(S,C)$ is a DIDS and $C^{-1}$ satisfies the necessary conditions, the inverse model of the Collatz system can be represented by a unique inverse algebraic forest $\mathcal{F}=\{T_1,T_2\}$, where $T_1$ is rooted at the attractor $\{1,4,2\}$ and $T_2$ is rooted at the attractor $\left\{0\right\}$.

Step 2: Conclude that the Collatz system has a unique attractor set.

• By the corollary on the uniqueness of attractors in DIDS (15.1), since the Collatz system has a unique inverse algebraic forest, it must have a unique attractor set $A=\left\{\right\{1,4,2\},\{0\left\}\right\}$.

Therefore, we have formally demonstrated that the Collatz dynamical system $(S,C)$ has a unique attractor set consisting of two disjoint cycles: $\{1,4,2\}$ and $\left\{0\right\}$. □

Theorem 15.29 

(Points of Contact of the Attractor Sets in the Collatz System). In the Collatz dynamic system $(\mathbb{N},C)$, the attractor sets are the cycles $\{1,4,2\}$ and $\left\{0\right\}$, with points of contact 1 and 0, respectively.

Proof. 

First, we have already shown in the previous theorem that $\{1,4,2\}$ and $\left\{0\right\}$ are the attractor cycles under the Collatz function C.

Now, we will show that 1 and 0 are the points of contact for their respective cycles.

For the cycle $\{1,4,2\}$: Suppose, for contradiction, that there exists a natural number $n<1$ in the attractor cycle. Then, there exists $k\in \mathbb{N}$ such that $C^k\left(n\right)=1$. However, for any $n<1$, we have $C\left(n\right)>n$ (since n would be negative or zero), which contradicts the assumption that n is in the attractor cycle.

Therefore, 1 is the smallest element in the attractor cycle $\{1,4,2\}$ and, hence, is the point of contact.

For the cycle $\left\{0\right\}$: The cycle $\left\{0\right\}$ consists of a single element, which is the fixed point 0. By definition, 0 is the point of contact for this cycle.

In conclusion, the attractor sets of the Collatz system are the cycles $\{1,4,2\}$ and $\left\{0\right\}$, with points of contact 1 and 0, respectively. □

Theorem 15.30 

(Uniqueness of Point of Contact in Violating Multivalued Injectivity). Let $(S,F)$ be a discrete dynamical system and $G:S\to \mathcal{P}\left(S\right)$ be its inverse function. If $\{x_1,\dots ,x_n\}$ is an attractor cycle with point of contact $pc$, then $pc$ is the unique point in the cycle that violates multivalued injectivity. That is, for all $x_i,x_j\in \{x_1,\dots ,x_n\}\setminus \left\{pc\right\}$ with $i\ne j$:

$$G\left(x_i\right)\cap G\left(x_j\right)=\varnothing$$

Proof. 

Let $pc$ be the point of contact in an attractor cycle $A=\{x_1,\dots ,x_n\}$ of the Collatz system. Suppose, for contradiction, that there exists another point $x_i\in A$ with $x_i\ne pc$ that also violates multivalued injectivity.

By the definition of the point of contact, we have $G\left(x_i\right)\cap G\left(pc\right)=\left\{pc\right\}$ for all $x_i\ne pc$, where G is the inverse Collatz function.

However, if $x_i$ also violates multivalued injectivity, then there must exist a point $y\in G\left(x_i\right)\cap G\left(x_j\right)$ for some $x_j\in A$ with $x_j\ne x_i$. But this would imply that y is a predecessor of both $x_i$ and $x_j$ under G, contradicting the uniqueness of predecessors in an attractor cycle.

Therefore, the assumption that there exists another point $x_i\ne pc$ violating multivalued injectivity must be false, and $pc$ is indeed the unique such point in the attractor cycle. □

Theorem 15.31 

(Collatz Conjecture). For all $n\in \mathbb{N}$, the Collatz sequence starting at n eventually reaches one of the two attractor cycles: $\{1,4,2\}$ at the point of contact 1, or $\left\{0\right\}$ at the point of contact 0.

Proof. 

Let $(\mathbb{N},C)$ be the Collatz dynamical system and $C^{-1}$ its analytic inverse.

(1)

By the Collatz System as a DIDS theorem, $(\mathbb{N},C)$ is a DIDS.(15.25)

(2)

By the properties of DIDS, $(\mathbb{N},C)$ has no non-trivial cycles other than the attractor cycles, and all sequences converge to an attractor set.

(3)

The attractor sets of the Collatz system are the cycles $\{1,4,2\}$ and $\left\{0\right\}$, with points of contact 1 and 0, respectively.

(4)

The basin of attraction of the attractor set $\left\{\right\{1,4,2\},\{0\left\}\right\}$ is $\mathbb{N}$, due to the exhaustiveness of $C^{-1}$.

Therefore, for all $n\in \mathbb{N}$, the Collatz sequence starting at n converges to one of the two attractor cycles: $\{1,4,2\}$ at the point of contact 1, or $\left\{0\right\}$ at the point of contact 0. □

15.4. A Generalization of the Collatz Conjecture

Definition 15.7. 

Let $C_G:\mathbb{N}\to \mathbb{N}$ be the “Generalized Collatz Function” defined as follows:

$$C_G(x;a,b)=\left\{\begin{array}{cc}\frac{x}{a}\hfill & \mathit{if}x\equiv 0(moda),\hfill \\ bx+m\hfill & \mathit{otherwise}.\hfill \end{array}\right.$$

where a, b are arbitrary positive integer parameters.

Conjecture 1 

(Generalized Collatz Conjecture). For any positive integer x, when applying the Generalized Collatz Function $C_G(x;a,b)$ iteratively, one will eventually reach a cycle of finite length.

Definition 15.8. 

Let $C_G^{-1}:\mathbb{N}\to \mathcal{P}\left(\mathbb{N}\right)$ be the inverse function of $C_G$ defined as:

$$C_G^{-1}\left(x\right)=\left\{\begin{array}{cc}\left\{ax\right\}\hfill & \mathit{if}x\not\equiv (b+m)(modab),\hfill \\ \{ax,\frac{x-m}{b}\}\hfill & \mathit{if}x\equiv (b+m)(modab).\hfill \end{array}\right.$$

Theorem 15.32. 

The Generalized Collatz function $C_G:\mathbb{N}\to \mathbb{N}$ is deterministic and surjective.

Proof. 

First, we define the Generalized Collatz function $C_G$ using first-order logic:

$$\forall n\in \mathbb{N}:C_G\left(n\right)=\left\{\begin{array}{cc}\frac{n}{a}\hfill & \mathrm{if}\exists k\in \mathbb{N}:n=ak\hfill \\ bn+m\hfill & \mathrm{otherwise}\hfill \end{array}\right.$$

Step 1: Prove that $C_G$ is deterministic.

$$\begin{array}{cc}\hfill & \forall n\in \mathbb{N},\exists !m\in \mathbb{N}:C_G\left(n\right)=m\hfill \\ \hfill & \equiv \forall n\in \mathbb{N},(\exists !m\in \mathbb{N}:((\exists k\in \mathbb{N}:n=ak)\wedge m=\frac{n}{a})\hfill \\ \hfill & \vee (\neg (\exists k\in \mathbb{N}:n=ak)\wedge m=bn+m))\hfill \\ \hfill & \equiv \forall n\in \mathbb{N},(((\exists k\in \mathbb{N}:n=ak)\wedge \exists !m\in \mathbb{N}:m=\frac{n}{a})\hfill \\ \hfill & \vee (\neg (\exists k\in \mathbb{N}:n=ak)\wedge \exists !m\in \mathbb{N}:m=bn+m))\hfill \\ \hfill & \equiv \mathrm{true}\hfill \end{array}$$

Thus, $C_G$ is deterministic.

Step 2: Prove that $C_G$ is surjective.

$$\begin{array}{cc}\hfill & \forall m\in \mathbb{N},\exists n\in \mathbb{N}:C_G\left(n\right)=m\hfill \\ \hfill & \equiv \forall m\in \mathbb{N},(\exists n\in \mathbb{N}:((\exists k\in \mathbb{N}:n=ak)\wedge m=\frac{n}{a})\hfill \\ \hfill & \vee (\neg (\exists k\in \mathbb{N}:n=ak)\wedge m=bn+m))\hfill \\ \hfill & \equiv \forall m\in \mathbb{N},(\exists n\in \mathbb{N}:n=am)\hfill \\ \hfill & \vee (\exists n\in \mathbb{N}:m=bn+m\wedge \neg (\exists k\in \mathbb{N}:n=ak\left)\right)\hfill \\ \hfill & \equiv \forall m\in \mathbb{N},(\exists n\in \mathbb{N}:n=am)\hfill \\ \hfill & \vee (\exists n\in \mathbb{N}:m-m=bn\wedge \neg (\exists k\in \mathbb{N}:n=ak\left)\right)\hfill \\ \hfill & \equiv \forall m\in \mathbb{N},(\exists n\in \mathbb{N}:n=am)\hfill \\ \hfill & \vee (\exists n\in \mathbb{N}:m=m\wedge \neg (\exists k\in \mathbb{N}:n=ak\left)\right)\hfill \\ \hfill & \equiv \forall m\in \mathbb{N},(\exists n\in \mathbb{N}:n=am)\hfill \\ \hfill & \vee (\exists n\in \mathbb{N}:\neg (\exists k\in \mathbb{N}:n=ak\left)\right)\hfill \\ \hfill & \equiv \forall m\in \mathbb{N},(\exists n\in \mathbb{N}:n=am)\hfill \\ \hfill & \vee (\exists n\in \mathbb{N}:n\neg \equiv 0(moda\left)\right)\hfill \\ \hfill & \equiv \mathrm{true}\hfill \end{array}$$

Thus, $C_G$ is surjective.

In conclusion, as $C_G$ is both deterministic and surjective, the theorem is proved. □

Theorem 15.33 

(Generalized Collatz System as a DIDS). $(\mathbb{N},C_G)$ is a Discrete Inverse Dynamical System (DIDS) with inverse function $C_G^{-1}$.

Proof. 

Since $C_G$ is deterministic and surjective, by the necessary and sufficient conditions for a function F being deterministic and surjective (15.11), it follows that its inverse function G is multivalued injective, surjective, and exhaustive. Therefore, $(\mathbb{N},C_G)$ is a DIDS with inverse function $C_G^{-1}$. □

Theorem 15.34 

(Convergence of Attraction Points in the Generalized Collatz Conjecture). Let $C_G:\mathbb{N}\to \mathbb{N}$ be the Generalized Collatz function defined as:

$$C_G(x;a,b,m)=\left\{\begin{array}{cc}\frac{x}{a}\hfill & \mathit{if}x\equiv 0(moda)\hfill \\ bx+m\hfill & \mathit{otherwise}\hfill \end{array}\right.$$

where $a,b,m\in \mathbb{N}$ are arbitrary positive integer parameters.

Then, for any valid choice of parameters $a,b,m$, all possible attraction points in the Generalized Collatz Conjecture converge to a unique attractor set, with the points of contact being the minimum values in each cycle of the attractor set.

Proof. 

Step 1: Define the set of possible attraction points A as:

$$A=\{x\in \mathbb{N}:x\equiv r(moda),r\in \{0,1,\dots ,a-1\left\}\right\}$$

Step 2: Partition A into two subsets:

$$A_0=\{x\in A:x\equiv 0(moda)\}$$

$$A_1=\{x\in A:x\neg \equiv 0(moda)\}$$

Step 3: For each $x\in A_0$, the only preimage under $C_G$ is $y=ax$. Applying $C_G$ to y yields:

$$C_G\left(y\right)=C_G\left(ax\right)=\frac{ax}{a}=x$$

Thus, each $x\in A_0$ converges to itself, forming a trivial cycle of length 1.

Step 4: For each $x\in A_1$, apply the Generalized Collatz function $C_G$ iteratively until a value repeats, forming a cycle.

Step 5: Verify that all $x\in A_1$ converge to the same unique attractor set:

Suppose $x_1,x_2\in A_1$ converge to different attractor sets. Then, there exist $n_1,n_2\in \mathbb{N}$ such that:

$$C_G^{n_1}\left(x_1\right)=x_1\mathrm{and}{C}_G^{n_2}\left(x_2\right)=x_2$$

where $C_G^n$ denotes the n-fold composition of $C_G$ with itself.

However, since $C_G$ is a function, each $x\in A_1$ has a unique sequence of iterates under $C_G$. Thus, $x_1$ and $x_2$ must eventually converge to the same attractor set, contradicting the assumption that they converge to different attractor sets.

Step 6: Let $\{x_{min,1},\dots ,x_{min,k}\}$ be the set of minimum values in each cycle of the unique attractor set. Then, $\{x_{min,1},\dots ,x_{min,k}\}$ are the points of contact for the attractor set.

Formally, we can express the convergence of all $x\in A_1$ to the unique attractor set with points of contact $\{x_{min,1},\dots ,x_{min,k}\}$ using first-order logic:

$$\forall x\in A_1,\exists n\in \mathbb{N}:C_G^n\left(x\right)\in \{x_{min,1},\dots ,x_{min,k}\}$$

□

Remark 12. 

The set of minimum values $\{x_{min,1},\dots ,x_{min,k}\}$ in the unique attractor set of the Generalized Collatz Conjecture depends on the specific values of the parameters $a,b,m$. It can be calculated by finding fixed points or cycles through the iterative application of $C_G$.

Theorem 15.35 

(Generalized Collatz Conjecture). For all $n\in \mathbb{N}$, the Generalized Collatz sequence starting at n eventually reaches the unique attractor set containing the points of contact $\{x_{min,1},\dots ,x_{min,k}\}$.

Proof. 

The proof follows from the properties of DIDS:

Step 1: By the Generalized Collatz System as a DIDS theorem, $(\mathbb{N},C_G)$ is a DIDS (Section 15.4).

Step 2: By the properties of DIDS, $(\mathbb{N},C_G)$ has no non-trivial cycles other than the cycles in the unique attractor set, and all sequences converge to the attractor set.

Step 3: The attractor set of the Generalized Collatz system is unique, and the points of contact are the minimum values $\{x_{min,1},\dots ,x_{min,k}\}$ in each cycle of the attractor set, which can be proven by analyzing the behavior of $C_G$.

Step 4: The basin of attraction of the unique attractor set is $\mathbb{N}$, due to the exhaustiveness of $C_G^{-1}$.

Therefore, for all $n\in \mathbb{N}$, the Generalized Collatz sequence starting at n converges to the unique attractor set containing the points of contact $\{x_{min,1},\dots ,x_{min,k}\}$. □

Theorem 15.36 

(Termination of Algorithms in IDDS). Let $(S,F)$ be a discrete dynamical system satisfying the conditions of a Discrete Inverse Dynamical System (DIDS), where S is the state space and $F:S\to S$ is the evolution function. Let $G:S\to \mathcal{P}\left(S\right)$ be the analytic inverse of F, which is multivalued injective, surjective, and exhaustive. Let $T=(V,E)$ be the inverse algebraic forest generated by G. Then, any algorithm based on the IDDS principles applied to $(S,F)$ will always terminate.

Proof. 

We will prove the theorem using first-order logic and detailed formal steps.

Step 1: Define the properties of the analytic inverse function G.

$$\begin{array}{cc}\hfill \forall s_1,s_2\in S:(s_1\ne s_2\to G\left(s_1\right)\cap G\left(s_2\right)=\varnothing )& (\mathrm{Multivalued}\mathrm{injectivity})\hfill \\ \hfill \forall s\in S,\exists t\in S:s\in G\left(t\right)& \left(\mathrm{Surjectivity}\right)\hfill \\ \hfill \forall s\in S,\exists n\in \mathbb{N}:G^n\left(s\right)=r& \left(\mathrm{Exhaustiveness}\right)\hfill \end{array}$$

where r is a root of the inverse algebraic forest T.

Step 2: Assume, for contradiction, that there exists an infinite sequence of distinct nodes $v_1,v_2,\dots \in V$ in T such that:

$$\forall i\ge 1:v_{i+1}\in G\left(v_i\right)$$

Step 3: By the exhaustiveness property of G, for each node $v_i$ in the sequence, there exists a finite number of recursive applications of G that lead to a root node r. Formally:

$$\forall i\in \mathbb{N},\exists n_i\in \mathbb{N},\exists r\in V:(r\mathrm{is}\mathrm{a}\mathrm{root}\mathrm{node}\wedge v_i\in G^{n_i}\left(r\right))$$

Step 4: By the multivalued injectivity of G, each node in T has a unique parent. Therefore, for any two distinct nodes $v_i$ and $v_j$ in the sequence, their paths to the root must diverge at some point. Formally:

$$\forall i,j\in \mathbb{N}:(i\ne j\to \exists k\in \mathbb{N}:G^k\left(v_i\right)\cap G^k\left(v_j\right)=\varnothing )$$

Step 5: Consider the subsequence ${\left\{v_{n_i}\right\}}_{i=1}^{\infty }$ of nodes, where each $v_{n_i}$ is the node in the original sequence at which the path to the root is exactly $n_i$ steps long. By Step 3, this subsequence is infinite.

Step 6: By Step 4, for any two distinct nodes $v_{n_i}$ and $v_{n_j}$ in the subsequence, we have:

$$G^{min(n_i,n_j)}\left(v_{n_i}\right)\cap G^{min(n_i,n_j)}\left(v_{n_j}\right)=\varnothing$$

Step 7: Apply the pigeonhole principle to the subsequence ${\left\{v_{n_i}\right\}}_{i=1}^{\infty }$. Let $M=\left|S\right|$, the cardinality of the state space S. Consider the first $M+1$ nodes in the subsequence: $v_{n_1},v_{n_2},\dots ,v_{n_{M+1}}$. There must exist two distinct nodes $v_{n_i}$ and $v_{n_j}$ with $1\le i<j\le M+1$ such that:

$$G^{min(n_i,n_j)}\left(v_{n_i}\right)=G^{min(n_i,n_j)}\left(v_{n_j}\right)$$

But this contradicts Step 6, which states that these sets should be disjoint.

Step 8: Therefore, the assumption in Step 2 must be false, and there cannot exist an infinite sequence of distinct nodes in T. Formally:

$$\nexists v_1,v_2,\dots \in V:(\forall i\ge 1:v_{i+1}\in G\left(v_i\right)\wedge \forall i,j\in \mathbb{N}:i\ne j\to v_i\ne v_j)$$

Step 9: Consequently, any algorithm based on the IDDS principles applied to $(S,F)$ will always terminate, as it cannot generate an infinite sequence of distinct nodes in the inverse algebraic forest T.

□

Construction of the Inverse Forest: The inverse forest $\mathcal{F}$ associated with the Generalized Collatz system $(\mathbb{N},C_G)$ is constructed using the inverse function $C_G^{-1}$. The construction process is as follows:

(1)

Identify the unique attractor set $A=\{A_1,\dots ,A_m\}$ of the Generalized Collatz system by analyzing the behavior of $C_G$. Each $A_i$ is a cycle or a fixed point.

(2)

For each $A_i\in A$, choose a point of contact $x_{min,i}$, which is the minimum value in the cycle or the fixed point itself.

(3)

Create a root node for each point of contact $x_{min,i}$, and label it as the root of a tree $T_i$.

(4)

For each root node $x_{min,i}$, apply the inverse function $C_G^{-1}$ to generate its children nodes. These children nodes represent the preimages of $x_{min,i}$ under $C_G$.

(5)

Recursively apply $C_G^{-1}$ to each newly generated node to create its children, and continue this process indefinitely. This step constructs the branches of each tree $T_i$.

(6)

The resulting collection of trees $\mathcal{F}=\{T_1,\dots ,T_m\}$ forms the inverse forest associated with the Generalized Collatz system.

The inverse forest $\mathcal{F}$ encodes all the possible preimages and trajectories that lead to the attractor set A under the Generalized Collatz function $C_G$. Each tree $T_i$ in the forest represents the basin of attraction of the corresponding attractor $A_i$.

Convergence Guarantee: The convergence of all Generalized Collatz sequences to the unique attractor set is guaranteed by the properties of DIDS and the structure of the inverse forest $\mathcal{F}$:

• The exhaustiveness of $C_G^{-1}$ ensures that every natural number $n\in \mathbb{N}$ appears as a node in one of the trees $T_i$ of the inverse forest $\mathcal{F}$. This means that every Generalized Collatz sequence is represented in the inverse forest.
• The absence of non-trivial cycles outside the attractor set, which is a property of DIDS, guarantees that every path in each tree $T_i$ eventually leads to the corresponding point of contact $x_{min,i}$ at the root of the tree. This implies that every Generalized Collatz sequence must converge to one of the attractors in the unique attractor set.
• The uniqueness of the attractor set, which is proven by analyzing the behavior of $C_G$, ensures that there are no other possible limit points or cycles outside the attractor set. This means that the convergence of Generalized Collatz sequences is limited to the unique attractor set only.

The combination of these properties, which are derived from the DIDS structure and the inverse forest construction, provides a strong guarantee of convergence for all Generalized Collatz sequences to the unique attractor set containing the points of contact $\{x_{min,1},\dots ,x_{min,k}\}$.

In summary, the inverse forest $\mathcal{F}$ serves as a comprehensive model of the Generalized Collatz system, capturing all the possible trajectories and their convergence behavior. The properties of DIDS, such as the absence of non-trivial cycles and the exhaustiveness of the inverse function, ensure that the inverse forest provides a faithful representation of the system’s dynamics. By analyzing the structure of the inverse forest and the properties of the attractor set, we can derive a rigorous proof of the Generalized Collatz Conjecture and establish the universal convergence of all sequences to the unique attractor set.

15.5. Resolution of the Collatz Conjecture in Its Entirety

It is crucial to emphasize that the Theory of Inverse Discrete Dynamical Systems (TIDDS) resolves the Collatz Conjecture in its entirety, not merely for specific cases such as the $3x+1$ problem. This comprehensive resolution is achieved by leveraging two powerful theorems established within the TIDDS framework: the Unique Attractor Set Theorem and the Impossibility of Infinite-Length Attractor Theorem (15.8).

The Unique Attractor Set Theorem, as demonstrated in Section 16.3, proves that the Collatz dynamical system $(S,C)$, where $S=\mathbb{N}$ and $C:S\to S$ is the Collatz function, possesses a single, globally attracting set consisting of two disjoint cycles. By constructing the inverse algebraic forest associated with the Collatz system and analyzing its properties, we conclusively show that all trajectories, regardless of their initial state, eventually converge to this unique attractor set.

Furthermore, the Impossibility of Infinite-Length Attractor Theorem, presented in Section 15, establishes that the inverse algebraic forest of any Discrete Inverse Dynamical System (DIDS) satisfying the conditions of injectivity, multivaluedness, surjectivity, and exhaustiveness cannot contain an attractor of infinite length. In the context of the Collatz system, this theorem guarantees that the unique attractor set must consist of cycles of finite length, ruling out the possibility of divergent or chaotic behavior.

The combination of these two powerful results, derived from the rigorous application of TIDDS, effectively resolves the Collatz Conjecture in its full generality. By proving the existence and uniqueness of a finite-length attractor set, and demonstrating the convergence of all trajectories to this attractor set, we establish that the Collatz Conjecture holds true for all natural numbers, not just for specific instances or subsets.

This comprehensive resolution marks a significant advancement in our understanding of the Collatz problem and showcases the power of the inverse dynamical systems approach in tackling complex questions in discrete mathematics. The generality of the result underscores the effectiveness of the TIDDS framework in providing a unified, systematic method for analyzing and resolving conjectures in discrete dynamical systems.

16. Potential Applications of TIDDS

Some of the theoretical conclusions of TIDDS that have significant practical implications are:

(1)

Existence and uniqueness of the inverse model: TIDDS demonstrates that for each deterministic discrete dynamical system, there exists a unique, well-defined algebraic inverse model. This ensures that the inverse modeling approach is consistent and reliable for analyzing and inferring properties in a wide range of discrete systems.

(2)

Topological transport of properties: TIDDS establishes that topological properties demonstrated in the algebraic inverse model are effectively and validly transferred to the original dynamical system through homeomorphisms. This allows inferring important global properties of the original system by studying its more tractable inverse model.

(3)

Guaranteed convergence to attractor sets: TIDDS proves that all deterministic discrete dynamical systems converge to a set of attractors, which may include fixed points and periodic orbits, but exclude the possibility of genuine chaotic behavior. This powerful result has implications in understanding and controlling the long-term behavior of discrete systems.

(4)

Impossibility of infinite cycles: TIDDS shows that in the algebraic inverse forests of discrete inverse dynamical systems (DIDS), infinite cycles cannot exist. This implies that all trajectories eventually converge to an attractor set after a finite number of iterations, which is fundamental for ensuring the termination and convergence of discrete algorithms and processes.

(5)

Invariant structure of the inverse model: TIDDS demonstrates that the algebraic inverse model of a deterministic discrete dynamical system has an invariant forest structure that completely captures the dynamics of the original system. This structural correspondence allows for efficient analysis and inference of properties through the inverse model.

These theoretical conclusions have direct practical applications in areas such as:

• Analysis and control of complex systems
• Formal verification and optimization of algorithms
• Design of controllers and decision-making systems
• Data analysis and machine learning

By ensuring the existence and uniqueness of inverse models, allowing for property transport, proving convergence to attractor sets, ruling out infinite cycles, and establishing an invariant structure, TIDDS provides a robust theoretical framework for effectively addressing a wide range of practical problems involving discrete dynamical systems. These theoretical conclusions underpin the applicability and practical utility of the inverse modeling and analysis techniques developed in TIDDS.

Based on the theoretical conclusions that can be derived from TIDDS, several practical applications of value can be identified, and ways to implement them:

(1)

Analysis and control of complex systems: TIDDS allows modeling and studying complex discrete systems through their inverse algebraic models. This can help better understand the global properties and long-term behavior of these systems. For example, complex networks (such as social, economic, or biological networks) could be analyzed by constructing their inverse forest and studying properties such as cycles, convergence, etc. This would provide insights into the structure and dynamics of the original network.

(2)

Optimization of algorithms: TIDDS techniques could be used to analyze the complexity and termination of recursive algorithms by modeling them as discrete dynamical systems. By studying the inverse forest of an algorithm, bottlenecks could be identified, redundant steps optimized, and convergence formally proven. This would have practical applications in software design and computational complexity analysis.

(3)

Formal software verification: TIDDS methods would allow formally verifying properties of programs and algorithms by representing them as discrete systems. By proving properties in the algebraic inverse model, properties could be inferred in the original program. This would improve the robustness and reliability of software in critical applications.

(4)

Control system design: TIDDS provides a framework for designing controllers for discrete dynamical systems. By analyzing the inverse system, target states could be identified, control laws designed, and convergence proven. This has practical value in industrial control, robotics, embedded systems, etc.

(5)

Prediction and decision-making: TIDDS models can be used for prediction and decision support in complex discrete systems. For example, the evolution of an epidemic or the dynamics of stock prices could be modeled with a discrete system, its inverse model constructed, and future scenarios analyzed. The inferred properties would help forecast and plan courses of action.

(6)

Data analysis and machine learning: TIDDS ideas could be applied to analyze large discrete datasets and train machine learning models. The data would be modeled as trajectories of a dynamical system, and properties of the inverse model (such as attractor sets) would allow discovering patterns and relationships. This would be used for clustering, classification, prediction, etc.

In summary, TIDDS theory opens up numerous possibilities for practical application in areas such as complex systems analysis, algorithm optimization, software verification, automatic control, prediction, data analysis, and machine learning. The key is to model the problem as a discrete dynamical system, construct its inverse algebraic model, and transfer inferred properties back to the original problem. This allows for a novel and powerful approach to tackle many real-world computational and analytical challenges.

17. Conclusion and Future Directions

In this groundbreaking work, we have introduced the Theory of Inverse Discrete Dynamical Systems (TIDDS), a novel framework for modeling and analyzing discrete dynamical systems through inverse algebraic models. The central theorems on homeomorphic invariance and topological transport, rigorously proven, validate the transfer of cardinal attributes between dynamical representations, opening up new avenues for studying global properties of complex systems.

The successful application of TIDDS to provide an alternative proof of the Collatz Conjecture not only demonstrates the theory’s capability to tackle open problems but also highlights its potential for addressing a wide range of challenges in discrete dynamics. By constructing an associated inverse model and leveraging analytical property transfers within the inverted forest structure, we have shown how TIDDS can unravel previously inaccessible insights.

Moreover, our work has led to a remarkable discovery: every deterministic discrete dynamical system (DIDS) that satisfies the conditions of injectivity, multivaluedness, surjectivity, and exhaustiveness for its inverse function necessarily converges to a unique attractor set. This groundbreaking result establishes a universal principle of guaranteed convergence and non-chaoticity for a broad class of discrete systems, shedding new light on the fundamental nature of determinism and predictability in discrete dynamics.

The implications of these findings are far-reaching, both for the specific problem of the Collatz Conjecture and for the broader field of discrete dynamical systems. For the Collatz Conjecture, our proof not only resolves the long-standing question of convergence but also reveals the existence of a single, unique attractor set governing the system’s behavior. This insight provides a deeper understanding of the problem’s structure and dynamics, paving the way for potential generalizations and extensions.

More broadly, the universal principle of guaranteed convergence to a unique attractor set for DIDS satisfying certain conditions has the potential to revolutionize our understanding of discrete dynamical systems across various domains. This result suggests that a wide range of complex systems, from biological networks to social dynamics, may exhibit more predictable and stable long-term behavior than previously thought, provided they satisfy the necessary conditions on their inverse functions.

Furthermore, the theoretical foundations of TIDDS, particularly the concepts of homeomorphic invariance, topological transport, and the unique inverse algebraic forest, contribute to the fundamental understanding of discrete dynamical systems. These ideas provide a powerful framework for analyzing the relationships between different representations of a system and the transfer of properties between them, enabling the discovery of hidden structures and behaviors that may have been previously overlooked.

The work also opens up several exciting avenues for future research. One immediate direction is to explore the applicability of the unique attractor set principle to other classes of discrete dynamical systems and to investigate the necessary and sufficient conditions for its validity. This could lead to the development of new classification schemes for discrete systems based on their convergence properties and the characteristics of their inverse functions.

Another promising direction is the further development of computational methods for constructing and analyzing inverse algebraic forests for large-scale discrete systems. This could involve the design of efficient algorithms for building inverse models, the development of heuristics for identifying attractors and convergence properties, and the exploration of parallel and distributed computing techniques for handling systems with high computational complexity.

There is also immense potential for applying TIDDS and the unique attractor set principle to real-world problems across various fields. In biology, for example, these ideas could be used to study the robustness and stability of gene regulatory networks, to identify critical control points in cellular processes, and to develop new strategies for disease diagnosis and treatment. In social sciences, the unique attractor set principle could provide new insights into the emergence of collective behaviors, the dynamics of opinion formation, and the stability of social institutions.

In conclusion, the Theory of Inverse Discrete Dynamical Systems and the discovery of the unique attractor set principle for DIDS satisfying certain conditions represent a significant leap forward in our understanding and analysis of discrete dynamical systems. By providing a rigorous framework for inverse modeling, demonstrating its power through the resolution of the Collatz Conjecture, and establishing a universal principle of guaranteed convergence, this work lays the foundation for a new paradigm in discrete dynamics that could have transformative implications across mathematics, science, and engineering.

The key contributions of this work are:

(1)

A rigorous mathematical framework for inverse modeling of discrete dynamical systems, establishing the theoretical foundations and key properties of inverse algebraic forests.

(2)

The demonstration of powerful theorems on homeomorphic invariance and topological transport, validating the transfer of cardinal attributes between equivalent dynamical representations.

(3)

A groundbreaking application in providing an alternative proof of the Collatz Conjecture, through the construction of an associated inverse model and the analytical transfer of properties within the inverted forest structure.

(4)

The discovery of a universal principle of guaranteed convergence to a unique attractor set for deterministic discrete dynamical systems satisfying certain conditions on their inverse functions, excluding the possibility of chaos and establishing a new paradigm for understanding determinism and predictability in discrete dynamics.

(5)

The opening of new avenues for research and the inspiration of further applications of TIDDS and the unique attractor set principle to a wide range of problems in mathematics, computer science, biology, social sciences, and beyond.

In summary, this work not only resolves the Collatz Conjecture and its generalizations but also establishes a powerful framework for inverse modeling of discrete dynamical systems and uncovers a universal principle of guaranteed convergence to a unique attractor set for a broad class of systems. These groundbreaking results have the potential to reshape our understanding of discrete dynamics, unveil hidden structures and behaviors in complex systems, and drive transformative advances across multiple fields of inquiry.

Figure 4. High level sketch of the Theory.

Figure 4. High level sketch of the Theory.