Preprint
Short Note

This version is not peer-reviewed.

Filter for Submodular Partition Function: Connection to Loose Tangle

A peer-reviewed article of this preprint also exists.

Submitted:

27 May 2023

Posted:

30 May 2023

You are already at the latest version

Abstract
Loose Tangle is a concept in graph theory that has a dual relationship with branch-width which is well-known graph width parameter. Ultrafilter, a fundamental notion in mathematics, is similarly known to have a dual relationship with branch-width when extended to a connectivity system (X, f). In this compact paper, we revisit and contemplate the interplay between Loose Tangle and Filter through the lens of a submodular partition function.
Keywords: 
;  ;  ;  

1. Introduction

The investigation of graph width parameters finds extensive applications across diverse fields, such as matroid theory, lattice theory, theoretical computer science, game theory, network theory, artificial intelligence, graph theory, and discrete mathematics, as evidenced by numerous studies (for example, see [1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,22,28,29,30,31,32,33]). These graph width parameters are frequently explored in conjunction with obstruction, contributing to a robust body of research.
Loose tangle, an innovative concept initially brought forward in reference [1], occupies a central role in ascertaining whether a branch-width is at most a natural number k, where k+1 denotes the order of the tangle. The relevance and potential of loose tangles have further been explored in the context of submodular partition functions [22]. These submodular partition functions significantly broaden the understanding of various well-established tree decompositions of graphs.
Furthermore, the ultrafilter, a concept well-regarded in the mathematical arena, is acknowledged to maintain a dual relationship with branch-width when extended over a connectivity system (X, f).
In this compact and focused paper, we revisit and contemplate the interplay between Loose Tangle and Filter through the lens of a submodular partition function.

2. Preliminaries

In this section, we present the essential definitions required for this paper. Throughout the paper, we utilize a finite set (referred to as the underlying set) X, a set of partitions P, and natural numbers i, k, and p. It is important to note that a partition involves dividing the elements of a set into non-empty, distinct subsets, ensuring that each element belongs to one and only one subset.
Furthermore, in this paper, we employ the symbol α to represent collections of subsets, such as α signifying a collection A1, ..., Ak of subsets of a finite set (the underlying set) X. The collection α is deemed a partition if the sets Ai are mutually disjoint, and their union forms the underlying set X. We define the following notation: if α represents the collection A1, ..., Ak, and A is another subset, then α ∩ A denotes the collection A1 ∩ A, ..., Ak ∩ A. Similarly, we use α \ A as a related notation. Lastly, [B1, ..., Bp, α] signifies the collection obtained from α by inserting sets B1, ..., Bp into the collection. Note that this notation is adopted from reference [22].

2.1. Submodular Partition Functions

We will explain about submodular partition functions. The definition of a partition function and a submodular partition function of separations is provided below:
Definition 1
[21,22]. A partition function is a function that maps the set of all partitions to non-negative integers, satisfying the condition ψ([∅, α]) = ψ(α) for every partition α. In other words, inserting an empty set into a collection does not alter the value of the partition function. A partition function ψ is submodular if the following holds for every two partitions [A, α] and [B, β]:
ψ([A, α]) + ψ([B, β]) ≥ ψ([A∪ (X\B), α ∩ B]) + ψ([B∪ (X\A), β ∩ A]).
We will further assume that ψ([X]) = 0 since shifting all values of a submodular partition function by a constant does not break the property. Pk[ψ] denote the set of partitions α of X such that ψ(α) ≤ k. The submodular partition function exhibits certain characteristics. Lemma 3 and Lemma 4, in particular, are obviously valid, and we provide a proof for clarity. This allows us to establish that the submodular partition function possesses a symmetric property.
Lemma 2
[22]. Let ψ be a submodular partition function on X and [A, α] a partition. Then ψ([A, α]) ≥ ψ([A, X\A]).
Lemma 3.
Let ψ be a submodular partition function on X. Then ψ([A, X\A]) = ψ([X\A, A]).
Proof of Lemma 3:
To prove this, we can use the submodular property of the partition function as given by the inequality:
ψ([A, α]) + ψ([B, β]) ≥ ψ([A ∪ (X\B), α ∩ B]) + ψ([B ∪ (X\A), β ∩ A])
Let's consider two sets A and B, where B = X\A. We will show that ψ([A, X\A]) = ψ([X\A, A]) using the submodular property.
First, let α = X\A and β = A. Then, α ∩ B = X\A ∩ (X\A) = X\A, and β ∩ A = A ∩ A = A. Plugging these values into the inequality, we get:
ψ([A, X\A]) + ψ([X\A, A]) ≥ ψ([A ∪ (X\(X\A)), X\A]) + ψ([(X\A) ∪ (X\A), A])
Since X\(X\A) = A, we have:
ψ([A, X\A]) + ψ([X\A, A]) ≥ ψ([A, X\A]) + ψ([X\A, A])
Thus, the inequality becomes an equality, which means the submodular property holds, and we have shown that ψ([A, X\A]) = ψ([X\A, A]). This proof is completed.□
Lemma 4.
Let ψ be a submodular partition function on X. Then ψ(∅) = 0.
Proof of Lemma 4:
From the definition of a submodular partition function, we have:
ψ([∅, α]) = ψ(α) for every partition α.
Now, let's consider the partition α = X. Then we have:
ψ([∅, X]) = ψ(X)
By the assumption of submodular partition functions that ψ([X]) = 0, we get:
ψ([∅, X]) = 0
Which gives us the result that:
ψ(∅) = 0
This completes the proof of Lemma 4.□

2.2. Loose P-Tangle

The definition of Loose P-tangle for submodular partition functions is below.
Definition 5
[22]. Let ψ be a submodular partition function on X. A loose Pk[ψ]-tangle is a family T of subsets of a finite set (an underlying set) X closed under taking subsets satisfying the following three axioms.
  • (P1) ∅ ∈ T, {e} ∈ T, for all e ∈ X such that the partition [{e}, X\{e}] belongs to Pk[ψ].
  • (P2) If A1, A2, . . ., Ap ∈ T , Ci ⊆ Ai for i = 1, . . ., p, [C1, . . ., Cp, X\(C1 ∪ . . . ∪ Cp)] ∈ Pk[ψ], then C1 ∪ . . . ∪ Cp ∈ T.
  • (P3) X ∉ T.
The loose Pk[ψ]-tangle exhibits the following dual properties.
Theorem 6
[22]. Let ψ be a submodular partition function on X. There is no decomposition tree compatible with Pk[ψ] if and only if there is a loose Pk[ψ]-tangle.

2.3. Filter of partitions

This new definition holds an equivalent relationship with Loose Pk[ψ]-Tangle (see section 3). The definition of Filter for submodular partition functions is below.
Definition 7:
Let ψ be a submodular partition function on a finite set X. An Pk[ψ]-(non-principal) filter of partitions is a family F satisfying the following four axiom:
  • (F1) For all e ∈ X, if the partition [{e}, X\{e}] belongs to Pk[ψ], then {e}∉ F,
  • (F2) If A1 ∈ F, A1 ⊆ A2, [A2, X\(A2)] ∈ Pk[ψ], then A2 ∈ F,
  • (F3) If A1, A2, . . .,Ai ∈ F for i = 1, . . ., p, [X\A1, . . .,X\ Ap, X\(X\A1 . . . X\Ap)] ∈ Pk[ψ], then A1 ... Ap ∈ F,
  • (F4) ∅ ∉ F.
It's important to note that a filter is classified as principal if it encompasses a singleton.
The axioms that constitute the non-principal Pk[ψ] filter of partitions echo the conceptual underpinnings of a Sigma-filter. The Sigma-filter, acting as a selection mechanism for specific subsets within a sigma-algebra, plays a pivotal role in the exploration of measure and integration. Specifically, axiom (F3) is viewed as a counterpart to one of the axioms inherent in the Sigma-filter construct.
For reference, the definition of a Sigma-filter is provided below.
Definition 8:
Let X be a set and Σ be a sigma-algebra of subsets of X. A sigma filter on X of Σ is a collection F of subsets of X that satisfies the following properties:
  • (SF2) If A ∈ F, B ∈ Σ, A ⊆ B, then B ∈ F,
  • (SF3) If A₁, A₂, A₃, ... ∈ F, then A1 ∩ ... ∩ Ap ∈ F
  • (SF4) ∅ ∉ F.
The non-principal Pk[ψ] filter introduced in this context can be perceived as a distinctive variant of the Sigma-filter, integrating conditions of a Submodular partition function and non-principal properties into its foundational definition.
Alongside its counterpart, the Sigma-ideal, both these constructs serve as vital tools in measure theory and probability theory, with extensive research dedicated to their understanding and application. Given the abundance of research conducted in the field of sigma-algebras, it can be considered as one of the crucial areas of study (ex. [23,24,25,26,27]).

3. Cryptomorphism between Loose tangle of partitions and Filter of partitions

In this section, we demonstrate the cryptomorphism between Loose P-tangle of partitions and Filter of partitions. The main result of this paper is presented below.
Theorem 9.
Let ψ be a submodular partition function on a finite set X. T is a loose Pk[ψ]-tangle iff F = {A | X\A ∈ T } is a Pk[ψ]-(non-principal) filter.
Proof of Theorem 9:
We'll prove the theorem in two steps:
-
First, we'll show that if T is a loose Pk[ψ]-tangle, then F = {A | X\A ∈ T } is a Pk[ψ]-(non-principal) filter.
-
Secondly, we'll show that if F is a Pk[ψ]-( (non-principal) filter, then T = {A | X\A ∈ F} is a loose Pk[ψ]-tangle.
Part 1:
Assume that T is a loose Pk[ψ]-tangle. We'll show that F = {A | X\A ∈ T } is a Pk[ψ]-(non-principal) filter.
Let's show axiom (F1). If [{e}, X\{e}] belongs to Pk[ψ], then by (P1) in the definition of T, we have {e} ∈ T. Hence, X\{e} is in F.
Now, let's show axiom (F2).Suppose A1 ∈ F, A1 ⊆ A2, and [A2, X\A2] ∈ Pk[ψ]. Since X\A1 ∈ T and X\A2 ⊆ X\A1, by the closure of T under taking subsets and [X\A2, A2] ∈ Pk[ψ], we have X\A1 ∈ T. Hence A1 = X\(X\A2) ∈ F.
Let's show axiom (F3). Suppose that A1, A2, ...,Ai ∈ F for i = 1, ..., p, and [X\A1, . . .,X\ Ap, X\(X\A1 ∪ . . . ∪ X\Ap)] ∈ Pk[ψ]. By definition of F, X\Ai ∈ T. Thus, by axiom (P2) in the definition of T and [X\A1, . . .,X\ Ap, X\(X\A1 ∪ . . . ∪ X\Ap)] ∈ Pk[ψ], (X\A1) ∪ ... ∪ (X\Ap) ∈ T. Therefore, by definition of F, X\((X\A1) ∪ ... ∪ (X\Ap)) = A1 ∩ ... ∩ Ap is in F.
Finally, let's show axiom (F4). By axiom (P3) in the definition of T, we have X ∉ T. Therefore, X\X = ∅ is in F.
Thus, if T is a loose Pk[ψ]-tangle, then F = {A | X\A ∈ T } is a Pk[ψ]- (non-principal) filter.
Part 2:
Now, assume that F is a Pk[ψ]-(non-principal) filter. We'll show that T = {A | X\A ∈ F} is a loose Pk[ψ]-tangle.
Let's show axiom (P1). If [{e}, X\{e}] belongs to Pk[ψ], then by axiom (F1) in the definition of F, we have {e} ∉ F. Hence, X\{e} ∉ T.
Let's show axiom (P2). Suppose that A1, A2, ..., Ap belong to T, Ci ⊆ Ai for i = 1, ..., p, and [C1, ..., Cp, X\(C1 ∪ ... ∪ Cp)] belongs to Pk[ψ]. By definition of T, we have X\Ai ∈ F and X\Ai ⊆ X\Ci. Thus, by axiom (F2) in the definition of F and Lemma 2, X\C1, X\C2, …, X\Cp is in F. By axiom (F3) in the definition of F, X\C1 ∩ ... ∩ X\Cp = C1 ∪ ... ∪ Cp is in T. So axiom (P2) holds.
Finally, let's show (P3). By (F4) in the definition of F, we have ∅ ∉ F. Therefore, X\∅ = X ∉ T.
Thus, if F is a Pk[ψ]- (non-principal) filter, then T = {A | X\A ∈ F} is a loose Pk[ψ]-tangle.
Hence, based on parts 1 and 2, the theorem is proven, thus concluding the proof.

4. Future tasks

We will consider about single ideal [19] and linear tangle [20] using Submodular Partition Function. Also we will discuss about ultrafilter [7,11], tangle [10,32] using Submodular Partition Function.

Acknowledgements

I humbly express my sincere gratitude to all those who have extended their invaluable support, enabling me to successfully accomplish this paper.

References

  1. Oum, Sang-il, and Paul Seymour. "Testing branch-width.". Journal of Combinatorial Theory, Series B 2007, 97, 385–393. [CrossRef]
  2. HICKS, Illya V.; BRIMKOV, Boris. Tangle bases: Revisited. Networks 2021, 77, 161–172. [CrossRef]
  3. Fujita, T. and Yamazaki, K. Equivalence between Linear Tangle and Single Ideal. Open Journal of Discrete Mathematics 2019, 9, 7–10. [CrossRef]
  4. Yamazaki, Koichi. "Tangle and maximal ideal." WALCOM: Algorithms and Computation: 11th International Conference and Workshops, WALCOM 2017. Proceedings 11. Hsinchu, Taiwan, 29–31 March 2017; Springer International Publishing, 29 March 2017. [CrossRef]
  5. Yamazaki, Koichi. "Inapproximability of rank, clique, boolean, and maximum induced matching-widths under small set expansion hypothesis.". Algorithms 2018, 11, 173. [CrossRef]
  6. Fedor V Fomin and Dimitrios M Thilikos. On the monotonicity of games generated by symmetric submodular functions. Discrete Applied Mathematics 2003, 131, 323–335. [CrossRef]
  7. Fujita, Takaaki, and Koichi Yamazaki. "Tangle and Ultrafilter: Game Theoretical Interpretation.". Graphs and Combinatorics 2020, 36, 319–330. [CrossRef]
  8. P. Seymour and R. Thomas. Graph searching and a min-max theorem for tree-width. Journal of Combinatorial Theory, Series B 1993, 58, 22–23. [CrossRef]
  9. Isolde Adler. Games for width parameters and monotonicity. arXiv 2009, arXiv:0906.3857. [CrossRef]
  10. Jim Geelen, Bert Gerards, Neil Robertson, and Geoff Whittle. Obstructions to branch-decomposition of matroids. Journal of Combinatorial Theory, Series B 2006, 96, 560–570. [CrossRef]
  11. Fujita, Takaaki. "Reconsideration of Tangle and Ultrafilter using Separation and Partition.". arXiv 2023, arXiv:2305.04306. [CrossRef]
  12. Paul, Christophe, Evangelos Protopapas, and Dimitrios M. Thilikos. "Graph Parameters, Universal Obstructions, and WQO.". arXiv 2023, arXiv:2304.03688. [CrossRef]
  13. Reed, Bruce A. "Tree width and tangles: A new connectivity measure and some applications.". Surveys in combinatorics 1997, 87–162.
  14. KURKOFKA, Jan. Ends and tangles, stars and combs, minors and the Farey graph. PhD Thesis, Staats-und Universitätsbibliothek Hamburg Carl von Ossietzky, 2020.
  15. Yamazaki, Koichi, et al. "Isomorphism for graphs of bounded distance width.". Algorithmica 1999, 24, 105–127. [CrossRef]
  16. Fujita, Takaaki. "Revisiting Linear Width: Rethinking the Relationship Between Single Ideal and Linear Obstacle.". arXiv 2023, arXiv:2305.04740. [CrossRef]
  17. DIESTEL, Reinhard; OUM, Sang-il. Tangle-tree duality: in graphs, matroids and beyond. Combinatorica 2019, 39, 879–910. [CrossRef]
  18. Fomin, Fedor V., and Tuukka Korhonen. "Fast fpt-approximation of branchwidth.". In Proceedings of the 54th Annual ACM SIGACT Symposium on Theory of Computing; 2022.
  19. Takaaki Fujita and Koichi Yamazaki. Linear-width and single ideal. The 20th Anniversary of the Japan Conference on Discrete and Computational Geometry, Graphs, and Games; 2017; pp. 110–111.
  20. Daniel Bienstock. Graph searching, path-width, tree-width and related problems (a survey). Reliability of Computer and Communication Networks , Vol.DIMACS. Series in Discrete Mathematics and Theoretical Computer Science. 1989; pp. 33–50.
  21. Amini, Omid, et al. "Submodular partition functions.". Discrete Mathematics 2009, 309, 6000–6008. [CrossRef]
  22. Škoda, Petr. "Computability of width of submodular partition functions.". In Combinatorial Algorithms; Springer-Verlag: Berlin, Heidelberg, 2009; Vol. 5874, pp. 450–459.
  23. Halton, J. H. Sigma-algebra theorems. 2008. [Google Scholar] [CrossRef]
  24. Kubrusly, C. S. Measure on a σ-Algebra. Essentials of Measure Theory. 2015, 23–39. [Google Scholar]
  25. Gentili, Stefano. "Monotone Classes and σ-Algebras." Measure, Integration and a Primer on Probability Theory: Volume 1. Springer International Publishing: Cham, 2020; pp. 131–145.
  26. Ohba, Sachio. "Topological-group-valued measures.". Yokohama Math. J 1974, 22, 101–104.
  27. Calin, Ovidiu, and Ovidiu Calin. "Information Representation.". Deep Learning Architectures: A Mathematical Approach 2020, 317–349.
  28. Fujita, T. Alternative Proof of Linear Tangle and Linear Obstacle: An Equivalence Result. Asian Research Journal of Mathematics 2023, 19, 61–66. [Google Scholar] [CrossRef]
  29. Fomin, Fedor V., Petr Golovach, and Dimitrios M. Thilikos. "Contraction obstructions for treewidth.". Fomin, Fedor V., Petr Golovach, and Dimitrios M. Thilikos. "Contraction obstructions for treewidth." 2011, 101, 302–314.
  30. Giannopoulou, Archontia C., et al. "Cutwidth: obstructions and algorithmic aspects.". Algorithmica 2019, 81, 557–588. [CrossRef]
  31. Yamazaki, Koichi, et al. "Isomorphism for graphs of bounded distance width.". Algorithmica 1999, 24, 105–127. [CrossRef]
  32. Robertson, Neil, and Paul D. Seymour. "Graph minors. X. Obstructions to tree-decomposition.". Journal of Combinatorial Theory, Series B 1991, 52, 153–190. [CrossRef]
  33. Hicks, Illya V., Arie MCA Koster, and Elif Kolotoğlu. "Branch and tree decomposition techniques for discrete optimization." Emerging Theory, Methods, and Applications. INFORMS. 2005, 1–29.
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.
Copyright: This open access article is published under a Creative Commons CC BY 4.0 license, which permit the free download, distribution, and reuse, provided that the author and preprint are cited in any reuse.
Prerpints.org logo

Preprints.org is a free preprint server supported by MDPI in Basel, Switzerland.

Subscribe

Disclaimer

Terms of Use

Privacy Policy

Privacy Settings

© 2025 MDPI (Basel, Switzerland) unless otherwise stated