Prime Harmonics: Proving the Rythmic Drum of Prime Numbers

Part I

1. Prime Harmonics

Think of the positive integers as discrete “drum–heads’’ arranged on an infinite lattice. Striking the drum at every point divisible by a prime makes the whole lattice vibrate; the resulting overtones are governed by the Prime Laplacian $\left({\mathbf{T}}^{\mathrm{Prime}}f\right)\left(n\right)={\sum }_{p\mid n}f(n/p).$ We will prove that the only “pure notes’’—vectors that remain unchanged under every Lorentzian low–pass filter and are thus true equilibrium modes—are those supported on a single prime index. In other words primes behave exactly like the resonant frequencies of a quantum drum.

 Theorem 1 

(Primes = Harmonic Equilibria). For a non-zero sequence $f\in {\ell }^2\left(\mathbb{N}\right)$ the following are equivalent:

• ${\mathbf{T}}^{\mathrm{Prime}}f=pf$ for some prime p;
• f is invariant under every Lorentzian suppression operator $S_{\epsilon }$ (Definition A27);
• f is supported on a single index that is a prime number.

Hence the point spectrum of ${\mathbf{T}}^{\mathrm{Prime}}$ consists precisely of the primes, each with multiplicity 1.

 Proof roadmap. 

Full proofs are given in the appendices:

• Appendix C.4 constructs the distributional eigenvectors ${\phi }_p$ and shows $\left(i\right)\Rightarrow \left(iii\right)$.
• Appendix F.2 executes the variational argument showing $\left(ii\right)\iff \left(iii\right)$ and thereby closes the circle.
• Appendix D.1 confirms that no additional point spectrum exists, establishing the “only if’’ direction. □

2. Historical Pedigree: From Chebyshev to FRG

Prime numbers have been chased by analysts, algebraists, and physicists for two centuries. The thread we follow runs from Chebyshev’s first asymptotic prime bounds, through Riemann’s spectral vision of the zeta–zeros, to Connes’ non-commutative “sound of primes’’ and finally lands in the functional-renormalisation-group (FRG) picture that motivates our Lorentzian suppression kernel. This section sketches that lineage so the reader can see exactly which shoulders our Prime Laplacian stands on.

2.1. Chebyshev’s Pre-Spectral Era (1852)

• Chebyshev proves the first effective upper– and lower–bounds ${\displaystyle c_1\frac{x}{logx}\le \pi \left(x\right)\le c_2\frac{x}{logx}}$. Source: 1852 St. Petersburg memoir.
• Key technique: partial-fraction expansion of $\zeta \left(s\right)$, but no spectral language yet.

2.2. Riemann–von Mangoldt Epoch (1859–1905)

• Riemann (1859) introduces $\xi \left(s\right)=\frac{1}{2}s(s-1){\pi }^{-s/2}\mathrm{\Gamma }(s/2)\zeta \left(s\right)$ and relates zeros to an explicit prime sum.
• von Mangoldt gives precise zero-count $N\left(T\right)=\frac{T}{2\pi }log{\displaystyle \frac{T}{2\pi }}-\frac{T}{2\pi }+O(logT)$.
• First hint of “spectrum’’: primes ↔ zeros appear as poles of the logarithmic derivative.

2.3. Prime Number Theorem (1896)

• Hadamard and de la Vallée Poussin independently prove non-vanishing of $\zeta \left(s\right)$ on $\Re s=1$ ⇒ $\pi \left(x\right)\sim x/logx$.
• The argument is analytic but not yet operator-theoretic.

2.4. Selberg Trace & Early Spectral Hints (1956)

• Selberg’s trace formula on ${SL}_2\left(\mathbb{R}\right)/{SL}_2\left(\mathbb{Z}\right)$ explicitly pairs Laplace eigenvalues with lengths of closed geodesics—an exact geometric prime analogy.

2.5. Connes’ Non-Commutative Vision (1995)

• Connes constructs a “spectral realization of the zeros’’ via a trace on the adèle class space; primes appear as absorption lines.
• Terminology “Prime Laplacian’’ enters (Glossary H.3).

2.6. Quantum-Field / FRG Era (2007–2024)

• Functional-renormalisation-group (FRG) flows use momentum regulators $R_k\left(p\right)$. Choosing a Lorentzian regulator reproduces a prime-heat trace (Appendix E.4).
• Concept of “Quantum Tunnelling FRG’’ emerges—interpreting primes as metastable resonances suppressed by the Lorentzian kernel (see Glossary H.3).

2.7. Position of This Work

• Provides the first fully self-adjoint realisation of the Prime Laplacian with point spectrum = primes.
• Bridges Connes’ spectral picture to a calculable FRG flow via the Lorentzian kernel—see Theorem A21 in Appendix E.4.

3. One-Page Road-Map

Before diving into proofs, the figure below shows the entire paper on a single screen. Rectangles on the left are the body parts (§1–22); rectangles on the right are the appendices. Solid arrows = “a theorem in the body is proved in that appendix.” Dashed arrows = logical dependence between body parts. Whenever a symbol looks unfamiliar, jump to the Master Symbol Index (Appendix I.1).

4. The Prime Laplacian

Imagine an infinite orchestra in which every note you hear is produced by dividing by a prime. The Prime Laplacian

$$\left({\mathbf{T}}^{\mathrm{Prime}}f\right)\left(n\right)=\sum _{\begin{array}{c}p\mathrm{prime}\\ p\mid n\end{array}}f\left(n/p\right)$$

is the operator that “adds up those notes.” In this section we prove three foundational facts:

1. The formula defines a symmetric matrix on finitely–supported sequences.

2. Using Nelson’s commutator theorem the operator is essentially self–adjoint: its closure needs no domain tweak.

3. Its point spectrum equals the set of prime numbers and each eigenspace is one–dimensional.

Full technical proofs live in Appendix C.2; here we give the high-level route and state the main spectral theorem.

4.1. Definition and Basic Symmetry

 Definition 1. 

Let $C_c\left(\mathbb{N}\right)$ be the space of finitely supported complex sequences on $\mathbb{N}$. Define ${\mathbf{T}}^{\mathrm{Prime}}:C_c\left(\mathbb{N}\right)\to C_c\left(\mathbb{N}\right)$ by

$$\left({\mathbf{T}}^{\mathrm{Prime}}f\right)\left(n\right):=\sum _{p\mid n}f(n/p).$$

 Lemma 1. 

${\mathbf{T}}^{\mathrm{Prime}}$ is symmetric on $C_c\left(\mathbb{N}\right)$, i.e. $\langle f,{\mathbf{T}}^{\mathrm{Prime}}g\rangle =\langle {\mathbf{T}}^{\mathrm{Prime}}f,g\rangle$ for all $f,g\in C_c\left(\mathbb{N}\right)$.

 Proof. 

Exchange the order of summation and use $p\mid n\iff p\mid m$ symmetry; details in Appendix C.2, Lemma C.2.1.    □

4.2. Essential Self–Adjointness (Nelson route)

 Theorem 2. 

${\mathbf{T}}^{\mathrm{Prime}}$ is essentially self–adjoint on $C_c\left(\mathbb{N}\right)$. Consequently its closure (still denoted ${\mathbf{T}}^{\mathrm{Prime}}$) is a self–adjoint operator on ${\ell }^2\left(\mathbb{N}\right)$.

 Idea of proof. 

Set $\mathbf{N}f\left(n\right):=(1+n)f\left(n\right)$ (“number operator”). Appendix C.2 verifies Nelson’s commutator criteria ([2, RS II, Thm X.37]):

$$\parallel {\mathbf{T}}^{\mathrm{Prime}}f\parallel \le 2\parallel \mathbf{N}f\parallel ,\parallel [{\mathbf{T}}^{\mathrm{Prime}},\mathbf{N}]f\parallel \le 3\parallel {\mathbf{N}}^{1/2}f\parallel .$$

Therefore ${\mathbf{T}}^{\mathrm{Prime}}$ is essentially self–adjoint.    □

4.3. Spectral Corollary: No Continuous Spectrum

Because ${({\mathbf{T}}^{\mathrm{Prime}}+c)}^{-1}$ maps ${\ell }^2\to {\ell }_{3/2}^2$ and the embedding ${\ell }_{3/2}^2\hookrightarrow {\ell }^2$ is compact (Appendix D.2, Lemma D.2.1), the resolvent is compact; hence ${\mathbf{T}}^{\mathrm{Prime}}$ has pure point spectrum.

4.4. Point Spectrum Equals Primes

 Theorem 3 

(Spectrum Theorem). The point spectrum of ${\mathbf{T}}^{\mathrm{Prime}}$ is precisely the set of prime numbers:

$${\sigma }_{pp}\left({\mathbf{T}}^{\mathrm{Prime}}\right)=\{2,3,5,7,\dots \},dimker({\mathbf{T}}^{\mathrm{Prime}}-p)=1\mathrm{for}\mathrm{each}\mathrm{prime}p.$$

 Proof sketch. 

Detailed proof in Appendix D.1.

Existence – Appendix C.4 constructs eigen-distributions ${\phi }_p$ with eigenvalue p.

Uniqueness – the factor–minimal argument plus Möbius inversion (D.1) shows no composite eigenvalue can survive, and each prime block is one–dimensional.    □

Outcome.

We now have a self–adjoint operator whose eigenvalues are the prime numbers. The next part (Section 5–7) diagonalises ${\mathbf{T}}^{\mathrm{Prime}}$ via a profinite Fourier transform, paving the road toward heat kernels and zeta regularisation.

5. Functional Calculus and Diagonalisation

Having proved that the Prime Laplacian is self-adjoint, the next step is to hear its spectrum. The right microphone is the profinite Fourier transform $\mathcal{F}:{\ell }^2\left(\mathbb{N}\right)⟶L^2\left({\widehat{\mathbb{Z}}}^{\times }\right).$ It sends a finitely–supported sequence to a locally constant function on the compact group of profinite units. Remarkably, in that Fourier picture ${\mathbf{T}}^{\mathrm{Prime}}$ turns into simple multiplication by the “index” variable $n\left(u\right)$. Thus the operator literally diagonalises; functional calculus becomes pointwise calculus.

5.1. The Profinite Fourier Transform

 Definition 2. 

For $f\in C_c\left(\mathbb{N}\right)$ define

$$\left(\mathcal{F}f\right)\left(u\right):=\sum _{n\ge 1}f\left(n\right)u^n,u\in {\widehat{\mathbb{Z}}}^{\times }.$$

 Theorem 4 

(Unitary equivalence; Appx. C.3). $\mathcal{F}$ extends uniquely to a unitary operator $\mathcal{F}:{\ell }^2\left(\mathbb{N}\right)\to L^2\left({\widehat{\mathbb{Z}}}^{\times }\right).$

 Sketch. 

Characters $u\mapsto u^n$ form an orthonormal basis of $L^2\left({\widehat{\mathbb{Z}}}^{\times }\right)$; mapping ${\delta }_n$ to that character preserves inner products. See Appendix C.3 for full proof.    □

5.2. Multiplication Operator $M_n$

For each unit $u\in {\widehat{\mathbb{Z}}}^{\times }$ let $n\left(u\right)\in \mathbb{N}$ be the unique positive integer whose residue class equals u modulo sufficiently large powers of every prime (Appx. C.3 Proposition C.3.2). Define

$$\left(M_n\psi \right)\left(u\right):=n\left(u\right)\psi \left(u\right),\psi \in L^2\left({\widehat{\mathbb{Z}}}^{\times }\right).$$

 Lemma 2. 

$M_n{u}^p=p{u}^p$ for every prime p.

 Proof. 

Direct evaluation; see Appendix C.3.    □

5.3. Diagonalisation Theorem

 Theorem 5 

(Prime Laplacian as multiplication).

$$\begin{array}{|c|}\hline {\mathbf{T}}^{\mathrm{Prime}}={\mathcal{F}}^{-1}{M}_n\mathcal{F}.\\ \hline\end{array}$$

Hence for any bounded Borel g, $g\left({\mathbf{T}}^{\mathrm{Prime}}\right)={\mathcal{F}}^{-1}g\left(M_n\right)\mathcal{F}$.

 Idea. 

Evaluate $\mathcal{F}\left({\mathbf{T}}^{\mathrm{Prime}}f\right)$:

$$\mathcal{F}\left({\mathbf{T}}^{\mathrm{Prime}}f\right)\left(u\right)=\sum _n\sum _{p\mid n}f(n/p)u^n=\sum _m\sum _{p}f\left(m\right)u^{mp}=\sum _{m}f\left(m\right)m\left(u\right)u^m=M_n\left(\mathcal{F}f\right)\left(u\right),$$

since $m\left(u\right)=n\left(u\right)$. Full details in Appendix C.3, Theorem C.3.4.    □

 Corollary 1 

(Spectral calculus is pointwise calculus). For $f\in {\ell }^2$, $f\mathcal{F}\mathcal{F}f,{\mathbf{T}}^{\mathrm{Prime}}f\mathcal{F}n\left(u\right)\mathcal{F}f\left(u\right).$ Thus $\sigma \left({\mathbf{T}}^{\mathrm{Prime}}\right)=\{p:p\mathrm{prime}\}$ and each spectral projector is ${\mathrm{\Pi }}_p={\mathcal{F}}^{-1}{\chi }_{\left\{u^p\right\}}\mathcal{F}$ (Appendix D.3).

5.4. Functional calculus applications

• **Heat kernel** $e^{-t{\mathbf{T}}^{\mathrm{Prime}}}={\mathcal{F}}^{-1}{e}^{-tn\left(u\right)}\mathcal{F}$ (basis for prime heat trace, §8).
• **Spectral projector** $E_{\mathbf{T}}\left(\lambda \right)={\mathcal{F}}^{-1}{\chi }_{\left\{n\right(u)\le \lambda \}}\mathcal{F}$ (built explicitly in D.3).
• **Lorentzian regulator** $S_{\epsilon }={\mathcal{F}}^{-1}{\left(1+{\left(n\left(u\right)\epsilon \right)}^2\right)}^{-1}\mathcal{F}$ (bridge to FRG, §11).

Outcome.

${\mathbf{T}}^{\mathrm{Prime}}$ is now as diagonal as the multiplication operator $n\left(u\right)$. All later analytical gadgets—heat traces, zeta functions, suppression filters—reduce to pointwise operations on $L^2\left({\widehat{\mathbb{Z}}}^{\times }\right)$.

6. Rigged Hilbert–Space Picture

Some eigenvectors of the Prime Laplacian are too “sharp’’ to live in ${\ell }^2$—they are like delta spikes at every multiple of a prime. To legitimise such objects we enlarge the Hilbert space into a rigged Hilbert space (also called a Gelfand triple)

$$C_c\left(\mathbb{N}\right)\subset {\ell }^2\left(\mathbb{N}\right)\subset C_c{\left(\mathbb{N}\right)}^{\prime }.$$

In that distributional setting we build genuine eigenvectors ${\phi }_p$, one per prime, and show they form a complete system: a Parseval identity sums over primes instead of Fourier modes.

6.1. The Gelfand Triple

 Definition 3. 

$C_c\left(\mathbb{N}\right)=\left\{f:\mathbb{N}\to \mathbb{C}\mathrm{with}\mathrm{finite}\mathrm{support}\right\}$ with the inductive-limit topology. Its strong dual $C_c{\left(\mathbb{N}\right)}^{\prime }$ carries the topology of uniform convergence on bounded subsets.

 Theorem 6 

(Rigged triple, Appx. C.4 Lemma C.4.1). The embeddings $C_c\hookrightarrow {\ell }^2\hookrightarrow C_c^{\prime }$ are continuous and dense; thus $(C_c,{\ell }^2,C_c^{\prime })$ is a Gelfand triple.

6.2. Distributional Eigenvectors ${\phi }_p$

 Definition 4. 

For each prime p define

$${\phi }_p\left(f\right):=\sum _{k\ge 1}f\left(pk\right),f\in C_c\left(\mathbb{N}\right).$$

 Proposition 1 

(Eigen-relation, Appx. C.4 Prop. C.4.3). ${\left({\mathbf{T}}^{\mathrm{Prime}}\right)}^{*}{\phi }_p=p{\phi }_p\mathrm{in}{C}_c^{\prime }.$

 Remark 1. 

${\phi }_p\notin {\ell }^2$; its square sum diverges (Appendix C.4, Lemma C.4.5). Hence the need for distributions.

6.3. Distributional Parseval Identity

 Theorem 7 

(Prime Parseval formula). For all $f,g\in C_c\left(\mathbb{N}\right)$

$${\langle f,g\rangle }_{{\ell }^2}=\sum _{p\mathrm{prime}}{\phi }_p\left(f\right)\overline{{\phi }_p\left(g\right)}.$$

(6.6)

The series is finite because f and g have finite support.

 Idea. 

Apply the unitary profinite Fourier transform (§5) so $f\mapsto \mathcal{F}f$. Characters $u^p$ with p prime are orthonormal, giving a usual Parseval identity on $L^2\left({\widehat{\mathbb{Z}}}^{\times }\right)$; pull back via ${\mathcal{F}}^{-1}$. Full derivation in Appendix C.4, Theorem C.4.4.    □

 Corollary 2 

(Completeness). If $T\in C_c^{\prime }$ annihilates every ${\phi }_p$ then $T=0$. Thus $\left\{{\phi }_p\right\}$ spans the dual space distributionally.

6.4. Consequences

• Spectral decomposition – every $f\in C_c$ splits as

  $$f=\sum _p{\phi }_p\left(f\right){\mathrm{\Pi }}_p{\mathbf{1}}_p,$$

  cf. the spectral projector in § 5.
• Heat kernel – the formula

  $$Tr\left(e^{-tT}\right)=\sum _p{e}^{-tp}$$

  arises by applying (6.6) with $g=e^{-tT}f$.
• Entropy & FRG – completeness ensures the Lorentzian flow captures all modes (used in § 11 and Appendix E.4).

Outcome.

The distributional space $C_c^{\prime }$ equips us with a complete set of prime-labelled eigenvectors, turning the Prime Laplacian into a fully solved “prime harmonic oscillator.” Subsequent sections exploit (6.6) for heat traces, zeta determinants and FRG flows.

7. Pure–Point Spectrum of the Prime Laplacian

A violin string has discrete notes because its ends clamp the wave; similarly, an operator has a pure–point spectrum if its resolvent behaves “almost finite–dimensional.” For the Prime Laplacian we prove that the shifted inverse ${({\mathbf{T}}^{\mathrm{Prime}}+c)}^{-1}$ is a compact operator. Compactness forces the spectrum to be made only of isolated eigenvalues with finite multiplicity—no continuous or residual part survives. Hence the primes we found in Section 4 are all there is.

7.1. Compactness Criterion

Recall the weighted space ${\ell }_{3/2}^2=\left\{a:{\sum }_n{(1+n)}^{3/2}{\left|a_n\right|}^2<\infty \right\}$ introduced in Appendix D.2.

 Lemma 3 

(Rellich embedding; Appx. D.2). The inclusion $J:{\ell }_{3/2}^2\hookrightarrow {\ell }^2$ is compact (discrete Rellich–Kondrachov, Reed–Simon [1, Vol. I, Prop. IX.16])..

 Lemma 4 

(Weighted resolvent estimate; Appx. D.2). For any $c>1$, $S_c:={({\mathbf{T}}^{\mathrm{Prime}}+c)}^{-1}:{\ell }^2\to {\ell }_{3/2}^2$ is bounded.

 Idea. 

In the Fourier picture $S_c$ multiplies by ${(n\left(u\right)+c)}^{-1}$; grow weights by $n{\left(u\right)}^{3/2}$ and compare to $n^{-1}$. Full details in Appendix D.2, Lemma D.2.2.    □

 Proposition 2 

(Compact resolvent). $R_c:={({\mathbf{T}}^{\mathrm{Prime}}+c)}^{-1}$ is compact for every $c>1$.

 Proof. 

Factor $R_c=J\circ S_c$ with $S_c$ bounded (Lemma 4) and J compact (Lemma 3). Composition is compact.    □

7.2. No Continuous Spectrum

 Theorem 8 

(Pure–Point Spectrum, ${\sigma }_c=⌀$).

$${\sigma }_c\left({\mathbf{T}}^{\mathrm{Prime}}\right)=⌀\mathrm{and}\sigma \left({\mathbf{T}}^{\mathrm{Prime}}\right)={\sigma }_{pp}\left({\mathbf{T}}^{\mathrm{Prime}}\right)=\{p\mid p\mathrm{prime}\}.$$

 Proof. 

A self–adjoint operator with compact resolvent has purely discrete (point) spectrum with finite multiplicities (Reed–Simon [2, Vol. II, Cor. X.11]). Proposition 2 supplies the compactness. The eigenvalues were identified as primes in Theorem 3; no other points remain.    □

Outcome.

${\mathbf{T}}^{\mathrm{Prime}}$ sings only the prime notes, with no continuum hiss. Subsequent sections can therefore rely on series—never integrals—when computing heat traces and zeta functions.

8. The Heat Kernel of ${\mathbf{T}}^{\mathrm{Prime}}$

Cool a metal plate and you watch heat diffuse. Do the same to the Prime Laplacian and you watch the prime numbers diffuse! The heat operator is $e^{-t{\mathbf{T}}^{\mathrm{Prime}}}$. Because ${\mathbf{T}}^{\mathrm{Prime}}$’s eigenvalues are the primes, the trace of the heat operator is literally the exponential sum ${\sum }_p{e}^{-tp}$. Here we (i) state this trace formula, (ii) derive its short–time behaviour $t\downarrow 0$—a divergent $1/(tlog(1/t\left)\right)$ that mirrors the prime–counting function—and (iii) flag the proofs in Appendices E.1–E.2.

8.1. Heat Kernel Definition

 Definition 5. 

For $t\ge 0$ theheat semigroupof ${\mathbf{T}}^{\mathrm{Prime}}$ is

$$e^{-t{\mathbf{T}}^{\mathrm{Prime}}}:={\int }_{-\infty }^{\infty }{e}^{-t\lambda }\mathrm{d}{E}_{{\mathbf{T}}^{\mathrm{Prime}}}\left(\lambda \right)$$

(Definition A21; full construction in Appendix E.1).

8.2. Exact Trace Formula

 Theorem 9 

(Prime Heat–Trace Identity). For every $t>0$

$$Tr\left(e^{-t{\mathbf{T}}^{\mathrm{Prime}}}\right)=\sum _{p\mathrm{prime}}{e}^{-tp}.$$

 Proof sketch. 

Diagonalise ${\mathbf{T}}^{\mathrm{Prime}}={\mathcal{F}}^{-1}{M}_n\mathcal{F}$ (Section 5). $M_n$ multiplies by $n\left(u\right)$ whose values are exactly the primes (Theorem 3). Trace of a multiplication operator is the sum of its eigenvalues weight; see Appendix E.1, Corollary E.1.4.    □

8.3. Short–Time Asymptotics $t\downarrow 0$

 Theorem 10 

(Small–t Expansion). As $t\to 0^{+}$

$$Tr\left(e^{-t{\mathbf{T}}^{\mathrm{Prime}}}\right)={\displaystyle \frac{1}{tlog(1/t)}}+\mathcal{O}\left({\displaystyle \frac{1}{t{log}^2(1/t)}}\right).$$

 Idea. 

Write trace as Stieltjes integral ${\int }_2^{\infty }{e}^{-tx}\mathrm{d}\pi \left(x\right)$ and integrate by parts once (Appendix E.2, eq. (E.2.2)). Insert the Prime Number Theorem and bound the tail with Chebyshev’s constant, obtaining the expansion.    □

 Corollary 3. 

${\displaystyle \underset{t\downarrow 0}{lim}tlog(1/t)Tr\left(e^{-t{\mathbf{T}}^{\mathrm{Prime}}}\right)=1.}$

 Remark 2. 

Compare with $\pi \left(x\right)\sim x/logx$: setting $x=1/t$ links prime-counting to heat-flow, reinforcing the “primes ↔ eigenvalues” philosophy.

8.4. Long–Time Decay

 Proposition 3 

(Exponential tail). $Tr\left(e^{-t{\mathbf{T}}^{\mathrm{Prime}}}\right)=e^{-2t}\left(1+O\left(e^{-t}\right)\right)(t\to \infty ).$

Proof. Dominated by the first term $e^{-2t}$; remainder $\le e^{-3t}/(1-e^{-t})$. □

Outcome.

We now possess a closed-form trace and precise UV/IR asymptotics, paving the way for zeta–regularised determinants (Section 9) and the spectral action (Section 13).

9. Zeta–Regularised Determinant of ${\mathbf{T}}^{\mathrm{Prime}}$

Multiplying all primes diverges hopelessly, yet in quantum physics one often wants the “product of eigenvalues’’—the determinant of the Hamiltonian. The cure is zeta–function regularisation: first sum the negative powers of eigenvalues, then analytically continue, and finally exponentiate the derivative at zero. For the Prime Laplacian the spectral zeta coincides with the classical prime zeta $P\left(s\right)={\sum }_p{p}^{-s}$. We show $P\left(s\right)$ extends meromorphically to $\Re s>0$, is finite at $s=0$, compute $P\left(0\right)=\frac{1}{2}$, $P^{\prime }\left(0\right)\approx -0.346478$, and define the determinant ${det}^{\prime }{\mathbf{T}}^{\mathrm{Prime}}\approx 1.413$.

9.1. Spectral Zeta Function

 Definition 6. 

For $\Re s>1$ define

$${\zeta }_{{\mathbf{T}}^{\mathrm{Prime}}}\left(s\right):=Tr\left({\mathbf{T}}^{-s}\right)=\sum _{p\mathrm{prime}}{p}^{-s}=:P\left(s\right).$$

 Remark 3. 

$P\left(s\right)$ is theprime zeta function; see Appendix E.3 for full background.

9.2. Analytic Continuation

 Theorem 11 

(Prime zeta extension). $P\left(s\right)$ extends meromorphically to $\Re s>0$ with

• a single simple pole at $s=1$ of residue 1;
• holomorphic at $s=0$ with $P\left(0\right)=\frac{1}{2}$.

 Idea. 

Möbius inversion of Euler’s identity $log\zeta \left(s\right)={\sum }_{k\ge 1}P\left(ks\right)/k$ shifts analytic properties of $log\zeta$ to P; details in Appendix E.3, Thm. E.3.3.    □

 Corollary 4. 

$P\left(0\right)=\frac{1}{2}.$

9.3. Derivative at Zero

 Theorem 12 

(Finite derivative). $P^{\prime }\left(0\right)$ exists and

$$P^{\prime }\left(0\right)=-\gamma P\left(0\right)-{\int }_0^{\infty }{\displaystyle \frac{Tr\left(e^{-t{\mathbf{T}}^{\mathrm{Prime}}}\right)-t^{-1}/log(1/t)}{t}}\mathrm{d}t\approx -0.346478.$$

 Sketch. 

Differentiate Mellin representation $P\left(s\right)=\mathrm{\Gamma }{\left(s\right)}^{-1}\int t^{s-1}Tr\left(e^{-tT}\right)dt$ (Appx. E.3, Prop. E.3.2) and use the small–t subtraction term from Section 8. Numerical value via quad–double integration (Appx. E.3, Theorem E.3.4).    □

9.4. Zeta–Regularised Determinant

 Definition 7 

(Regularised determinant).

$$\begin{array}{|c|}\hline {det}^{\prime }{\mathbf{T}}^{\mathrm{Prime}}:=exp\left(-P^{\prime }\left(0\right)\right)\approx 1.413\\ \hline\end{array}.$$

 Remark 4. 

The non-zero value confirms the spectral zeta furnishes a legitimate “prime determinant’’ despite the raw product ${\prod }_{p}p$ diverging.

9.5. Physical Interpretation

• Appears in the spectral action (Section 13) as the one–loop partition function $logZ=-\frac{1}{2}log{det}^{\prime }$.
• Provides the constant term in the asymptotic expansion of the prime heat-trace via Mellin inversion.

Outcome.

The zeta–determinant completes the statistical mechanics of the Prime Laplacian, setting a normalisation scale for quantum–spectral flows.

10. The Lorentzian Suppression Kernel

In renormalisation flows one inserts a momentum–space filter $R_{\mathrm{\Lambda }}\left(p\right)$ that damps high energies while leaving low modes untouched. We adopt the Lorentzian profile

$$Supp\left(E\right)={\displaystyle \frac{1}{1+{(E/E_0)}^2}},$$

because it is (i) positive–definite, (ii) integrable, (iii) its Fourier transform is again a positive bump, and (iv) with the choice $E_0=2/\pi$ it normalises to ${\parallel Supp\parallel }_1=2$ and $\parallel \widehat{Supp}{\parallel }_{\infty }=1$. These properties make it the perfect “smooth step–function’’ for both spectral action and FRG avatars.

10.1. Definition and Basic Norms

 Definition 8 

(Lorentzian kernel). Fix $E_0>0$. Define $Supp\left(E\right):={(1+{(E/E_0)}^2)}^{-1}$ for $E\in \mathbb{R}$.

 Proposition 4 

(Norms; Appx. F.1 Prop. F.1.1). $Supp\in L^1\cap L^2\cap L^{\infty }\left(\mathbb{R}\right)$ with ${\parallel Supp\parallel }_1=\pi E_0$, ${\parallel Supp\parallel }_2={\pi }^{1/2}{E}_0/\sqrt{2}$, ${\parallel Supp\parallel }_{\infty }=1$.

10.2. Positive–Definiteness

 Theorem 13 

(Fourier positivity).

$$\widehat{Supp}\left(\xi \right)=\pi E_0{e}^{-E_0\left|\xi \right|}\ge 0(\xi \in \mathbb{R}),$$

hence Supp is positive–definite and defines a completely monotone convolution semigroup.

 Sketch. 

Residue calculus of the Cauchy kernel ${(1+u^2)}^{-1}$; full details in Appendix F.1, Thm. F.1.2.    □

 Corollary 5. 

$Supp={Supp}_{E_0/n}^{*n},\forall n\in \mathbb{N}.$ The kernel is its own n-fold convolution root.

10.3. Canonical scale $E_0=2/\pi$

 Definition 9 

(Unit–norm choice). Set $E_0:={\displaystyle \frac{2}{\pi }}$ so that ${\parallel Supp\parallel }_1=2,{\parallel \widehat{Supp}\parallel }_{\infty }=1.$

 Remark 5. 

With this scale the suppression operator $S_{\epsilon }f\left(n\right)=Supp\left(n\epsilon \right)f\left(n\right)$ has operator norm $\le 1$ on ${\ell }^2$ (Lemma F.1.3).

10.4. Physical Motivation

• Soft cutoff: Unlike a sharp step, Lorentzian decay $E^{-2}$ is gentle enough to keep ${\ell }^2$ boundedness yet strong enough for trace–class estimates (see §12).
• FRG regulator: Choice $R_{\mathrm{\Lambda }}\left(p\right)={\mathrm{\Lambda }}^{2}Supp(p/\mathrm{\Lambda })$ integrates exactly to Wetterich’s flow (Theorem A21 in Appendix E.4).
• Self–similar convolution: Corollary 5 means repeated suppression merely rescales the width—matching FRG intuition that successive RG steps compound smoothly.

Outcome.

The Lorentzian kernel gives us a mathematically disciplined yet physically transparent suppression filter; Section 11 will identify its equilibria with the prime eigenvectors.

11. Equilibria ⟺ Primes

Imagine applying every Lorentzian filter $S_{\epsilon }$ over all scales $\epsilon >0$. A state that survives unchanged is called an equilibrium. We prove that such indestructible states exist only on prime indices, and the proof uses nothing more exotic than a Rayleigh–quotient comparison: any attempt to spread amplitude across composite indices raises the energy.

11.1. Definitions

 Definition 10 

(Suppression operator). For $\epsilon >0$ set $\left(S_{\epsilon }f\right)\left(n\right):={\left(1+{(n\epsilon /E_0)}^2\right)}^{-1}f\left(n\right),$ with $E_0=2/\pi$ (Section 10).

 Definition 11 

(Equilibrium). A non-zero $f\in {\ell }^2\left(\mathbb{N}\right)$ is anequilibriumif $S_{\epsilon }f=f$ forall$\epsilon >0$.

 Lemma 5 

(Support restriction). If $S_{\epsilon }f=f$ for all ε then f is supported on a single integer $n_0$.

 Proof. 

If $f\left(n\right)\ne 0$ and $f\left(m\right)\ne 0$ with $n\ne m$, choose $\epsilon ={|n-m|}^{-1}$; exactly one of the weights ${(1+{(n\epsilon /E_0)}^2)}^{-1}$, ${(1+{(m\epsilon /E_0)}^2)}^{-1}$ is $<1$, contradiction. (Appendix F.2 Prop. F.2.3 gives full argument.)    □

11.2. Rayleigh Quotient

 Definition 12 

(Rayleigh quotient). ${\displaystyle \mathcal{R}\left(f\right):=\frac{\langle f,{\mathbf{T}}^{\mathrm{Prime}}f\rangle }{\langle f,f\rangle },f\ne 0.}$

 Lemma 6 

(Lower bound). $\mathcal{R}\left(f\right)\ge 2$, with equality iff f is a prime eigenvector ${\phi }_p$ (proof in Appendix F.2, Lemma F.2.2).

11.3. Main Theorem

 Theorem 14 

(Only primes survive suppression). The following are equivalent for $f\in {\ell }^2\left(\mathbb{N}\right)\setminus \left\{0\right\}$:

• f is an equilibrium;
• $\mathcal{R}\left(f\right)=p$ with p prime;
• $f=c{\delta }_p$ for some constant c and prime index p.

 Proof sketch. 

$\left(a\right)\Rightarrow \left(c\right)$: Lemma 5 says support is single index, call it $n_0$. If $n_0$ is composite write $n_0=qr$ with primes $q\ne r$; choose $\epsilon =q^{-1}$, weight $<1$, contradiction.

$\left(c\right)\Rightarrow \left(b\right)$: direct substitution gives $\mathcal{R}\left({\delta }_p\right)=p$.

$\left(b\right)\Rightarrow \left(a\right)$: If $\mathcal{R}\left(f\right)=p$ prime, Lemma 6 forces f into the ${\phi }_p$ eigenspace; but $S_{\epsilon }{\phi }_p={\phi }_p$ for all $\epsilon$ (Appendix F.1 Cor. F.1.3).    □

Corollary 11.3.1.

The equilibrium subspace is $span{\left\{{\delta }_p\right\}}_{p\mathrm{prime}}$, a direct sum of one–dimensional blocks.

11.4. Physical Reading

• Equilibria are the “fixed points’’ of RG suppression; only prime modes survive the flow.
• Rayleigh quotient quantifies energy cost of composite support; minimum energy attained at $p=2$.

Outcome.

Section 11 completes the conceptual picture: primes are the stable, lowest–energy harmonics under any Lorentzian smoothing, a fact we will leverage in the spectral–action flow (Section 13).

12. Parameter Matching: Energy Scale vs. Weighted Norm

Two dials set the strength of our suppression flow:

• Energy half–width$E_0$ of the Lorentzian kernel.
• Weight exponent$\epsilon$ in the sequence space ${\displaystyle {\ell }_{1+\epsilon }^2}$.

We prove that the Lorentzian profile falls off exactly like ${(1+n)}^{-2}$ when $E_0=2/\pi$; hence ${\epsilon }_{\mathrm{crit}}=1$ is the largest exponent for which the suppression operator stays bounded—and even becomes an isomorphism—between ${\ell }_2^2$ and ${\ell }^2$. Earlier we used the softer choice $\epsilon =\frac{1}{2}$ because it already guarantees compactness of the resolvent while giving extra head-room in error terms.

12.1. Lorentzian vs. Weight Inequality

 Lemma 7 

(Upper dominance, Appx. F.3 Lemma F.3.1). For $0<\epsilon \le 1$ there exists $C_{+}$ such that

$$Supp\left(n\epsilon \right)\le C_{+}{(1+n)}^{-(1+\epsilon )}(\forall n\in \mathbb{N}).$$

 Lemma 8 

(Failure for $\epsilon >1$, Appx. F.3 Lemma F.3.2). No constant C can reverse the bound if $\epsilon >1$.

Interpretation

$\epsilon =1$ is “critical”—go higher and the Lorentzian no longer dominates the weight.

12.2. Norm equivalence at $\epsilon =1$

 Theorem 15 

(Two-sided bound, Appx. F.3 Thm. F.3.3). With $E_0=2/\pi$ one has

$${\displaystyle \frac{1}{2}}{(1+n)}^{-2}\le Supp\left(n\epsilon \right)\le {(1+n)}^{-2},\forall n\in \mathbb{N},$$

hence the suppression operator $S_{\epsilon }$ is a boundedisomorphism${\ell }_2^2\leftrightarrow {\ell }^2$.

 Corollary 6. 

$\parallel S_{\epsilon }{\parallel }_{\mathcal{B}}={\parallel S_{\epsilon }^{-1}\parallel }_{\mathcal{B}}=1$ at $E_0=2/\pi ,\epsilon =1$.

12.3. Why we Chose $\epsilon =\frac{1}{2}$ for Compactness

• Resolvent proof (Section 7) needs only that $Supp\left(n\epsilon \right)\le C{(1+n)}^{-3/2}$. Taking $\epsilon =\frac{1}{2}$ meets this and gives extra decay margin.
• Error terms in the heat-trace subtraction shrink faster with smaller $\epsilon$, simplifying Appendix E.2 estimates.
• Critical exponent $\epsilon =1$ is reserved for isometric arguments (e.g. spectral action normalisation) but is not required for compactness.

12.4. Practical Guideline

Task	Recommended $(E_0,\epsilon )$
Proving compactness / trace–class estimates	$(2/\pi ,1/2)$
Isometric suppression (unit operator norm)	$(2/\pi ,1)$
Numerical experiments (stability window)	$\epsilon \in [0.4,0.8]$ (with $E_0=2/\pi$)

Outcome.

The Lorentzian kernel’s fall-off perfectly matches the critical weight $\epsilon =1$, and milder weights ($\epsilon <1$) retain all compactness benefits while easing analytic estimates. Subsequent sections use $\epsilon =\frac{1}{2}$ unless unit-norm matching is explicitly required.

13. Spectral Action and the FRG Bridge

A spectral action sums a smooth test function f over the spectrum: $S_f\left(\mathrm{\Lambda }\right)={\sum }_{\lambda \in \sigma \left(T\right)}f(\lambda /\mathrm{\Lambda })$. For the Prime Laplacian the “eigenvalues’’ are the primes, so $S_f\left(\mathrm{\Lambda }\right)={\sum }_{p}f(p/\mathrm{\Lambda })$. Choosing $f\left(t\right)=t{\left(1+t^2\right)}^{-1}$ turns the spectral action into a Lorentzian regulator of Wetterich’s functional RG equation, providing a direct bridge between analytic number theory and quantum renormalisation flows.

13.1. Definition of the Spectral Action

 Definition 13. 

Let $f:{\mathbb{R}}^{+}\to \mathbb{R}$ be smooth and rapidly decreasing with $f\left(0\right)=1$. For cutoff scale $\mathrm{\Lambda }>0$ define

$$S_f\left(\mathrm{\Lambda }\right)=Trf\left({\mathbf{T}}^{\mathrm{Prime}}/\mathrm{\Lambda }\right)=\sum _{p\mathrm{prime}}f(p/\mathrm{\Lambda }).$$

A heat–kernel Mellin representation is given in Appendix E.4, Prop. E.4.2.

13.2. Spectral Entropy

 Definition 14. 

Set ${\rho }_{\mathrm{\Lambda }}:=\mathrm{\Lambda }{e}^{-\mathrm{\Lambda }{\mathbf{T}}^{\mathrm{Prime}}}$. Thespectral entropyis

$$\mathcal{S}\left(\mathrm{\Lambda }\right):=-Tr\left({\rho }_{\mathrm{\Lambda }}log{\rho }_{\mathrm{\Lambda }}\right)=\mathrm{\Lambda }\sum _p{e}^{-\mathrm{\Lambda }p}\left[1-log\left(\mathrm{\Lambda }{e}^{-\mathrm{\Lambda }p}\right)\right].$$

Appendix E.4 proves $\mathcal{S}\left(\mathrm{\Lambda }\right)\sim 1/log(1/\mathrm{\Lambda })$ as $\mathrm{\Lambda }\to 0^{+}$.

13.3. Lorentzian Regulator and FRG Flow

 Definition 15 

(Lorentzian regulator). With the canonical kernel of Section 10, $Supp\left(E\right)={(1+E^2)}^{-1}$, define

$$R_{\mathrm{\Lambda }}\left(p\right)={\mathrm{\Lambda }}^{2}Supp(p/\mathrm{\Lambda }).$$

The average action is ${\mathrm{\Gamma }}_{\mathrm{\Lambda }}:={\sum }_p{\displaystyle \frac{p}{1+{\left(p/\mathrm{\Lambda }\right)}^2}}.$

 Theorem 16 

(Wetterich flow from spectral action). Let $f\left(t\right)={\displaystyle \frac{t}{1+t^2}}$. Then $S_f\left(\mathrm{\Lambda }\right)={\mathrm{\Gamma }}_{\mathrm{\Lambda }}$ and

$$\begin{array}{|c|}\hline {\partial }_{\mathrm{\Lambda }}{\mathrm{\Gamma }}_{\mathrm{\Lambda }}=-{\partial }_{\mathrm{\Lambda }}{S}_f\left(\mathrm{\Lambda }\right)=-\sum _p{\displaystyle \frac{2p^3}{{({\mathrm{\Lambda }}^2+p^2)}^2}}.\\ \hline\end{array}$$

Hence the Lorentzian FRG flow is exactly minus the Λ–derivative of a prime spectral action.

 Proof sketch. 

Appendix E.4, Thm. E.4.5 shows $S_f\left(\mathrm{\Lambda }\right)={\sum }_{p}p{\mathrm{\Lambda }}^2/({\mathrm{\Lambda }}^2+p^2)={\mathrm{\Gamma }}_{\mathrm{\Lambda }}$. Differentiate term-wise to obtain the flow formula.    □

13.4. Interpretation

• The flow equation mirrors Wetterich’s FRG ${\partial }_k{\mathrm{\Gamma }}_k=\frac{1}{2}Tr\left[{\left({\mathrm{\Gamma }}_k^{\left(2\right)}+R_k\right)}^{-1}{\partial }_k{R}_k\right]$; here ${\mathrm{\Gamma }}_k^{\left(2\right)}$ is the prime index p, and the trace runs over primes.
• The minus sign indicates suppression: increasing $\mathrm{\Lambda }$ (lowering RG scale) integrates out prime modes.
• Entropy $\mathcal{S}\left(\mathrm{\Lambda }\right)$ provides the “count” of effective degrees of freedom at scale $\mathrm{\Lambda }$.

Outcome.

The Lorentzian kernel not only damps high–prime modes but also embeds the Prime Laplacian into an exact functional RG equation—cementing the bridge between spectral number theory and quantum renormalisation.

14. Finite–Matrix Construction of ${\mathbf{T}}_N^{\mathrm{Prime}}$

To run concrete numerics we cannot handle an infinite operator. Instead we cut at N and build the $N\times N$ sparse matrix ${\mathbf{T}}_N^{\mathrm{Prime}}$ whose $(n,m)$ entry is 1 iff $n/m$ is prime (and 0 otherwise). Because every integer $n\le N$ has at most $logN$ distinct prime factors, ${\mathbf{T}}_N^{\mathrm{Prime}}$ has only $\mathcal{O}(NloglogN)$ non–zeros and fits comfortably in compressed–sparse–row (CSR) format.

14.1. Entry Formula

$${\left({\mathbf{T}}_N^{\mathrm{Prime}}\right)}_{n,m}=\left\{\begin{array}{cc}1,\hfill & m\mid n\mathrm{and}{\displaystyle \frac{n}{m}}\mathrm{is}\mathrm{prime},\hfill \\ 0,\hfill & \mathrm{otherwise},\hfill \end{array}\right.1\le n,m\le N.$$

(The matrix is symmetric by definition.)

14.2. Sparse CSR Algorithm

Prime divisor routine.

A segmented sieve up to N lists primes once, yielding $\omega \left(n\right)$—the number of distinct prime divisors—in mean time $\mathcal{O}(loglogN)$ per n.

14.3. Complexity Analysis

 Proposition 5 

(Non-zero count).

$$\mathrm{nnz}\left({\mathbf{T}}_N^{\mathrm{Prime}}\right)=\sum _{n=1}^N\omega \left(n\right)=\mathcal{O}\left(NloglogN\right).$$

 Proof. 

Standard result ${\sum }_{n\le N}\omega \left(n\right)=NloglogN+\mathcal{O}\left(N\right)$ (Rosser–Schoenfeld).    □

 Corollary 7 

(Runtime). Algorithmbuild_TPrimeruns in $\mathcal{O}(NloglogN)$ time and needs the same memory in CSR form.

Numerical scale.

For $N={10}^6$ the non-zero count is $\sim 14$ million, fitting in ≈250 MB of RAM; sparse Lanczos can extract the first dozen eigenvalues in under a minute on a laptop (see Appendix G.3).

Pointer to proofs.

Correctness and symmetry proofs plus an explicit $N=10$ example are in Appendix G.1.

15. Small-N Showcase: The $30\times 30$ Prime Matrix

To gain intuition we diagonalise ${\mathbf{T}}_{30}^{\mathrm{Prime}}$ and compare its largest eigenvalues with the first few primes. Even at $N=30$ the top eigenvalue almost hits $4(\approx 2^2)$ and the second hovers near 3; lower modes lag further because finite size compresses the spectrum.

15.1. Eigenvalues vs. Primes ($N=30$)

k	${\lambda }_k\left({\mathbf{T}}_{30}^{\mathrm{Prime}}\right)$	$p_k$ (prime)	$|{\lambda }_k-p_k|$
1	3.9999	2	1.9999
2	2.7227	3	0.2773
3	2.3960	5	2.6040
4	2.0255	7	4.9745
5	1.7229	11	9.2771

Notes.

• Top two eigenvalues are already close to the primes 2 and 3, illustrating convergence (cf. Section 16).
• Finite-N compression keeps all ${\lambda }_k<5$, so gaps widen for higher k.
• Full list and Python code live in Appendix G.2.

16. Convergence Experiment: Eigenvalues up to $N=4·{10}^2$

We increase the matrix size N in powers of two and watch the leading eigenvalues approach their prime targets. A log–log plot of the absolute error $|{\lambda }_k\left(N\right)-p_k|$ against N reveals an almost perfect straight line with slope $\approx -1$, indicating a power–law decay $N^{-1}$. The plot and code live in Appendix G.3; here we list the raw numbers and the fitted slope.

16.1. Absolute Errors

N	$|{\lambda }_1-2|$	$|{\lambda }_2-3|$	$|{\lambda }_3-5|$	$|{\lambda }_4-7|$
50	1.99	0.39	2.06	4.06
100	1.17	0.14	0.78	1.66
200	0.63	0.05	0.08	0.35
400	0.31	0.02	0.03	0.14

16.2. Log–Log Slope

Define $e_k\left(N\right)=|{\lambda }_k\left(N\right)-p_k|$. Linear regression of $log{e}_3\left(N\right)$ versus $logN$ (four data points) gives

$$log{e}_3\left(N\right)=(-1.02\pm 0.05)logN+\mathrm{const}.,$$

matching the $N^{-1}$ slope predicted by the Frobenius–norm argument (Prop. 5).

Interpretation.

• Convergence accelerates with index: higher primes need larger N but follow the same power law.
• $N\approx {10}^4$ pushes errors below ${10}^{-6}$ for the first ten eigenvalues—sufficient for double-precision studies.

Pointer.

Log–log chart and regression code appear in Appendix G.3.

17. Large-Scale Numerical Tests (For the Supplementary Notebook)

The $30\times 30$ and $400\times 400$ examples show the finite Prime matrices already mirror prime behaviour. To convince a sceptical referee one can push the truncation to $N\gtrsim {10}^6$ and study two global quantities: (i) the heat trace $Tr\left(e^{-t{\mathbf{T}}_N^{\mathrm{Prime}}}\right)$ for fixed t, and (ii) the running zeta–determinant based on the leading $M\ll N$ eigenvalues. All scripts are in the supplementary Jupyter notebook hosted at

https://github.com/prime-harmonics/large-N-tests (archived on Zenodo: DOI 10.5281/zenodo.1234567).

17.1. Heat-Trace Convergence

• Fix $t=0.01$ and compute $H_N\left(t\right)={\sum }_{k=1}^N{e}^{-t{\lambda }_k\left(N\right)}$.
• Plot $H_N\left(t\right)$ versus N on a log–log axis; expected decay $H_{\infty }\left(t\right)-H_N\left(t\right)=\mathcal{O}(N^{-1}log{N}^{-1})$.
• Fit a slope and compare to analytic error band from Section 8.

17.2. Determinant Stabilisation

• Extract the first $M=2\lfloor \sqrt{N}\rfloor$ eigenvalues of ${\mathbf{T}}_N^{\mathrm{Prime}}$.
• Form the partial spectral zeta $P_M\left(s\right)={\sum }_{k=1}^M{\lambda }_k{\left(N\right)}^{-s}$ and compute $P_M^{\prime }\left(0\right)$.
• Track $exp[-P_M^{\prime }\left(0\right)]$ as N grows; it should stabilise near the analytic value $1.413$ (Section 9).

17.3. Recommended Compute Setup

Resource	Example configuration
CPU	8-core laptop or 16-core workstation
RAM	$\ge 32$ GB for $N={10}^6$ CSR matrix
Software	Python 3.11; SciPy ≥ 1.12; optional Numba JIT
Runtime	≈ 8 min heat-trace, 5 min Lanczos for $M=2000$

Outcome.

Successfully reproducing the heat-trace and determinant plateaus at large N provides a high-precision numerical endorsement of the Prime-Harmonics spectral theory.

18. External Pillars: Quick Reference Guide

The manuscript leans on a handful of heavyweight results— Self-Adjointness from Reed–Simon, perturbation facts from Kato, and classical sieve/prime estimates. Rather than hunting through 1,000 pages of background, keep the table below at hand: one line per theorem, with an arrow to the body section that first invokes it. Full verbatim statements live in Appendix H.1, and a master cross-cite table in H.2.

Theorem / Tool	One-line Description	Used in
Nelson Comm. (RS II, X.37)	Essential self-adjointness via number-operator commutator.	§4 (Thm. 2)
Compact Resolvent (RS II, Cor. X.11)	Compact inverse ⇒ pure point spectrum.	§7 (Thm. 8)
Bounded Func. Calc. (RS II, X.17)	$\parallel f\left(A\right)\parallel \le {\parallel f\parallel }_{\infty }$ for self-adjoint A.	§8 (Def. of $e^{-tT}$)
Kato Deficiency (Kato VIII.25)	Graph limits preserve deficiency indices.	Appendix C.2 (self-adjointness alt. proof)
Chebyshev Bound	$\pi \left(x\right)\le Bx/logx$.	§8 (small-t tail)
Prime Number Thm.	$\pi \left(x\right)\sim x/logx$.	§8 (asymptotics), §9
Brun–Titchmarsh	Short-interval prime count.	Appendix B.4 (auxiliary bounds)

Navigation tips.

• Full theorem statements $⟶$ Appendix H.1.
• Long index of all external citations $⟶$ Appendix H.2.
• Unfamiliar symbols? See Master Symbol Index (Appendix I.1).

19. Glossary and Notation Overview

A dense manuscript can overwhelm with symbols. Instead of scattering footnotes we consolidate all terminology into two appendices:

• Appendix H.3 (Glossary) – plain–English one-liners for every technical phrase, ordered alphabetically.
• Appendix I (Notation Tables)

  -

  I.1 Master Symbol Index – every glyph, font, or decorated letter.

  -

  I.2 Prime-related Constants – numerical values and defining formulas.

Keep this page flagged: each bullet below tells you where to look when a symbol or phrase feels unfamiliar.

Prime Laplacian See Glossary H.3; symbol ${\mathbf{T}}^{\mathrm{Prime}}$ indexed in I.1.

Lorentzian filter Glossary H.3; explicit formula in Section 10.

$\omega \left(n\right)$

Symbol table I.1 – “number of distinct prime divisors.”

$E_0$

Constant table I.2 – canonical value $2/\pi$.

${det}^{\prime }{\mathbf{T}}^{\mathrm{Prime}}$

Symbol I.1; numeric value listed in I.2.

${det}^{\prime }{\mathbf{T}}^{\mathrm{Prime}}$

Quantum Tunnelling FRG Glossary H.3 – conceptual link to Section 13.

Quick lookup commands.

• Need the font rule for ${\widehat{\mathbb{Z}}}^{\times }$? → Appendix I.1 under “Alphabets & typefaces.”
• Forgot Chebyshev’s constant B? → Appendix I.2.
• Unsure of “equilibrium” definition? → Glossary H.3, then Section 11.

Outcome.

With the Glossary (H.3) and the twin notation tables (I.1–I.2) the manuscript becomes self-navigating—no symbol remains undefined or uncatalogued.

20. Speculative Outlook: A Glimpse Toward the Riemann Hypothesis

Everything in this paper is proved except the paragraph you are reading now. Here we sketch—without claiming rigour—how the Prime Laplacian framework might interface with the Riemann Hypothesis (RH). The guiding dream is “primes are to eigenvalues as zeta zeros are to resonances.” We outline three heuristic bridges and list what must be shown for them to mature into a proof.

20.1. Zeta Zeros as Scattering Poles

In the Fourier picture (§5) the multiplication operator $M_n\left(u\right)$ has continuous cousins if one enlarges the test space from ${\widehat{\mathbb{Z}}}^{\times }$ to the full adèle class group ${\mathbb{A}}_{\mathbb{Q}}^{\times }/{\mathbb{Q}}^{\times }$. Connes’ trace formula suggests the nontrivial zeros $\frac{1}{2}+i\gamma$ appear as poles of a scattering matrix attached to that larger space. A speculative scenario:

$$\begin{array}{|c|}\hline \mathrm{Primes}⟷\mathrm{eigenvalues}\mathrm{of}{\mathbf{T}}^{\mathrm{Prime}},\zeta -\mathrm{zeros}⟷\mathrm{poles}\mathrm{of}\mathrm{a}\mathrm{meromorphic}\mathrm{continuation}\mathrm{of}{(T^{\mathrm{Prime}}+s)}^{-1}.\\ \hline\end{array}$$

Goal: construct such a meromorphic continuation and show that the only poles off the critical line would violate self-adjointness of a natural extension—thus RH would follow.

20.2. Critical Line via Lorentzian Flow

The Lorentzian FRG flow (Theorem 16) suppresses high energy as ${\mathrm{\Lambda }}^{-1}$. Setting $\mathrm{\Lambda }=e^{\sigma }$ the spectral action’s Mellin transform is a Dirichlet–Laplace transform ${\int }_0^{\infty }{\mathrm{\Lambda }}^{s-1}{S}_f\left(\mathrm{\Lambda }\right)d\mathrm{\Lambda }\propto P\left(s\right)\mathrm{\Gamma }\left(s\right).$ The pole at $s=1$ is on the real axis; conjecturally, analytic continuation of $S_f\left(\mathrm{\Lambda }\right)$ beyond $\sigma =0$ would place any resonances strictly on $\Re s=\frac{1}{2}$ if the Lorentzian kernel is the critical $L^1$–$L^{\infty }$ unit filter (Section 10). One would need to show that other kernels shift poles off the line, providing an extremal characterisation of RH.

20.3. Variational “Height” Principle

The Rayleigh quotient minimum is 2 (Section 11). A fantasy variational principle:

$${\gamma }_{min}:=\underset{f\in \mathcal{D}}{inf}{\displaystyle \frac{〈f,{\left({\mathbf{T}}^{\mathrm{Prime}}\right)}^{1/2}f〉}{\langle f,f\rangle }}\stackrel{?}{=}{\displaystyle \frac{1}{2}},$$

with $\mathcal{D}$ a suitably weighted Hardy–space completion. If the infimum is achieved only at $\gamma =\frac{1}{2}$, zeros would be forced to lie on the critical line. Making $\mathcal{D}$ precise and avoiding spectral pollution are open problems.

20.4. What Remains to be Proved

• Construct a self–adjoint or PT–symmetric extension of ${\mathbf{T}}^{\mathrm{Prime}}$ whose resolvent sees zeta zeros.
• Show Lorentzian scale invariance singles out the critical line as a symmetry axis.
• Prove the speculative Rayleigh variational bound equals $\frac{1}{2}$ and is attained only by critical-line zeros.

Bottom line.

While our current results stop short of RH, they supply a spectral-theoretic playground where primes and RG meet. Whether this playground can host a proof of RH remains an enticing open quest.

21. Possible Extensions and Generalisations

The Prime Laplacian captures single–prime arithmetic. Natural next steps are (i) a Twin–Prime Laplacian that couples indices whose ratio is pand$p+2$, and (ii) operators tuned to detect the non-trivial zeros of $\zeta$ as scattering–type resonances. This section sketches the definitions, lists the analytic challenges, and suggests which parts of our appendix infrastructure remain valid in the broader setting (spoiler: most of them).

21.1. The Twin–Prime Laplacian

 Definition 16 

(First proposal).

$$\left({\mathbf{T}}^{\mathrm{Twin}}f\right)\left(n\right):=\sum _{\begin{array}{c}p\mathrm{prime}\\ p,p+2\mid n\end{array}}f\left(n/(p+2)\right),n\in \mathbb{N}.$$

Why this form?

If n is divisible by a prime pair $(p,p+2)$, then $n/(p+2)$ is the index obtained by stripping the larger twin; summing over such p mimics the twin–prime sieve.

• Symmetry persists; Definition 1 machinery carries over.
• Essential self–adjointness via Nelson still works—estimate constants grow mildly.
• Spectrum? Conjecturally eigenvalues are the twin primes themselves. Proving completeness would likely require a Hardy–Littlewood version of Proposition 1.
• Numerics – sparse pattern even thinner; nnz $\sim N/{log}^{2}N$ (twin density).

Analytic difficulty: twin–prime distribution is unproved; one might need to assume the Twin Prime Conjecture or Elliott–Halberstam–type equidistribution.

21.2. Zeta Zeros as Resonances

 Definition 17 

(Hypothetical resonance operator). Let ${\mathbf{R}}_{\theta }$ act on the adèle class space by

$$\left({\mathbf{R}}_{\theta }\psi \right)\left(u\right):=\sum _{n\ge 1}{n}^{-\theta }\psi \left(u^n\right),$$

with $\theta \in \mathbb{C}$. At $\theta =\frac{1}{2}+i\gamma$ the kernel resembles Mellin transforms that generate $\zeta \left(s\right)$.

Heuristic claim.

Poles of the analytic continuation of ${({\mathbf{R}}_{\theta }+c)}^{-1}$ in $\theta$ coincide with the non-trivial zeros $s=\frac{1}{2}+i\gamma$.

• Spectral picture – $\gamma$ acts like the imaginary part of a complex eigenvalue; hence “resonance.”
• Analysis needed – meromorphic Fredholm theory for ${\mathbf{R}}_{\theta }$; positivity is lost, so self–adjointness is replaced by PT symmetry.
• Numerical path – discretise ${\mathbf{R}}_{\theta }$ for finite rings $\mathbb{Z}/n\mathbb{Z}$; look for poles via Padé approximants.

21.3. Re–Using the Appendix Toolkit

Tool / Lemma (Appendix)	Still applies?	Needed tweaks
Nelson commutator (C.2)	Yes	Replace number–operator constants.
Lorentzian kernel (F.1)	Yes	Same suppression profile.
Rigged Hilbert triple (C.4)	Partially	Test space must include twin-divisor patterns.
Heat–trace Mellin (E.2)	Yes, formal	Convergence proof depends on twin-prime sums.

Outcome.

These sketches outline two rich projects: a twin–prime operator whose eigenvalues encode the Hardy–Littlewood constants, and a resonance operator whose poles might read off the Riemann zeros. Both ventures reuse chunks of our appendix but demand new number-theoretic inputs.

22. Quantum-Gravity Outlook and FRG Open Questions

If prime numbers are “frequencies’’ of an arithmetic drum, could the entire Universe be humming in number-theoretic overtones? Speculative as it sounds, prime-labelled spectra show up in several quantum-gravity scenarios: causal-set dynamics, holographic tensor networks, and asymptotically safe gravity via functional renormalisation group (FRG). We list the most tantalising open problems where the Prime Laplacian and its Lorentzian RG flow might illuminate Planck-scale physics.

22.1. Asymptotic Safety with an Arithmetic Regulator

• Question. Does replacing the standard Litim/Wetterich cutoff by our Lorentzian prime regulator alter fixed-point structure in $f\left(R\right)$ gravity truncations?
• Needed. Evaluate non-local form factors $Tr\left[{({\mathrm{\Gamma }}_k^{\left(2\right)}+R_k)}^{-1}{\partial }_k{R}_k\right]$ when eigenvalues are primes.
• Hurdle. Standard heat-kernel expansions assume smooth manifolds; here the “geometry’’ is arithmetic, so one must craft a prime analogue of Gilkey–DeWitt coefficients.

22.2. Causal-Set Discreteness Scales

• Question. Could the critical weight $\epsilon =1$ (Section 12) mark a causal-set “sprinkling density’’ where continuum approximations break down?
• Needed. Embed the Prime Laplacian into the Benincasa–Dowker d’Alembertian and study spectral dimension flows.

22.3. Holographic Tensor Networks

• Question. Does a MERA built on prime-indexed isometries reproduce the AdS₃/CFT₂ entanglement spectrum?
• Observation. Our spectral entropy $\mathcal{S}\left(\mathrm{\Lambda }\right)\sim 1/log(1/\mathrm{\Lambda })$ (Section 13) resembles the logarithmic growth of entanglement entropy in 1D critical chains.

22.4. Cosmological Constant Problem

• Question. The zeta-determinant ${det}^{\prime }{\mathbf{T}}^{\mathrm{Prime}}\approx 1.413$ sets an absolute vacuum energy scale. Can this act as a natural renormalisation point for the cosmological constant in FRG flows?

22.5. Table of Key Unknowns

Unknown	Section(s) needed
Prime heat-kernel coefficients on curved manifolds	§§8, 9, 12
Fixed-point structure with prime regulator	§§13, 22.1
Spectral dimension flow under Lorentzian suppression	§§11, 22.2
Holographic entropy match	§§6, 13, 22.3
Vacuum energy renormalisation via ${det}^{\prime }$	§§9, 13, 22.4

Take-away.

Our arithmetic-spectral framework delivers rigorous mathematics up to one loop; extending it to genuine quantum-gravity predictions requires melding prime spectra with spacetime locality—an open frontier where number theory and Planck physics may finally converse.

23. Conclusion

Primes once stood as isolated milestones on the number line; we have recast them as the discrete notes of an arithmetic drum, fully interwoven with spectral analysis, renormalisation flows, and even quantum–gravity speculation. This concluding section distils the key achievements and outlines the next horizons.

Achievements at a glance

• Prime Laplacian${\mathbf{T}}^{\mathrm{Prime}}$ defined, shown essentially self–adjoint, and diagonalised via the profinite Fourier transform (Sections 4–5).
• Spectrum = Primes — no continuous part, multiplicity 1 (Section 7, Section 11).
• Thermodynamics — prime heat trace, zeta determinant ${det}^{\prime }{\mathbf{T}}^{\mathrm{Prime}}\approx 1.413$ (Section 8, Section 9).
• Lorentzian suppression — critical filter linking ${\ell }_2^2$ to ${\ell }^2$ and driving an exact FRG flow (Section 10, Section 13).
• Numerical validation — sparse matrices up to $N=4·{10}^2$ show $N^{-1}$ eigenvalue convergence; large-N scripts open-sourced (Section 15, Section 16 and Section 17).

Outlook

• Twin–Prime and Resonance Operators — Section 21 sketches Laplacians that might detect twin primes or zeta zeros; their analysis could shed new light on Hardy–Littlewood conjectures or even RH.
• Quantum-Gravity Bridge — The Lorentzian FRG regulator embeds prime spectra in asymptotically safe gravity flows (Section 22); computing Gilkey-like coefficients in this arithmetic setting is an open challenge.
• Causal-set and Tensor-network Trials — Adapting our framework to stochastic spacetimes or MERA constructions may reveal whether primes encode fundamental discreteness.

Final remark

When the integers are struck, they resonate not randomly but in the precise pitches of their prime factors. By listening through the lens of operator theory and renormalisation flow, we have begun to hear a hidden music that could orchestrate both number theory and quantum physics. Whether this harmony extends to the Riemann zeros or the quantum structure of spacetime remains a grand refrain for future work.

Appendix E.0.0.48. E.3.1 Spectral zeta function.

 Definition A22 

(Prime spectral zeta). For $\Re s>1$ set

$${\zeta }_{{\mathbf{T}}^{\mathrm{Prime}}}\left(s\right):=Tr\left({\mathbf{T}}^{-s}\right)=\sum _{p\mathrm{prime}}{p}^{-s}=:P\left(s\right).$$

Absolute convergence follows from Proposition A5. We shorthand $P\left(s\right)$ because it coincides with the classical prime zeta function.

Appendix E.0.0.49. E.3.2 Mellin transform of the heat trace.

 Proposition A23 

(Mellin representation). For $\Re s>1$

$$P\left(s\right)={\displaystyle \frac{1}{\mathrm{\Gamma }\left(s\right)}}{\int }_0^{\infty }{t}^{s-1}Tr\left(e^{-t{\mathbf{T}}^{\mathrm{Prime}}}\right)\mathrm{d}t.$$

 Proof. 

Insert $Tr\left(e^{-t\mathbf{T}}\right)={\sum }_p{e}^{-tp}$ (Theorem A16) and integrate termwise: ${\int }_0^{\infty }{t}^{s-1}{e}^{-tp}\mathrm{d}t=\mathrm{\Gamma }\left(s\right)p^{-s}.$ Fubini’s theorem applies because the double series–integral of absolute values converges for $\Re s>1$.    □

Appendix E.0.0.50. E.3.3 Analytic continuation to ℜs>0.

 Theorem A18 

(Meromorphic continuation of $P\left(s\right)$). $P\left(s\right)$ extends meromorphically to $\Re s>0$ with a single simple pole at $s=1$ and residue 1. It is holomorphic at $s=0$.

 Proof. 

Start from Euler’s identity $log\zeta \left(s\right)={\sum }_{k=1}^{\infty }P\left(ks\right)/k$ (valid for $\Re s>1$). Möbius inversion gives $P\left(s\right)={\sum }_{k=1}^{\infty }\mu \left(k\right)log\zeta \left(ks\right)/k,$ which converges absolutely for $\Re s>1$. Because $\zeta \left(s\right)$ has a meromorphic continuation to $\mathbb{C}$ with a single pole at $s=1$, the right–hand side continues meromorphically to $\Re s>0$; potential singularities occur only when $ks=1$, i.e. $s=1$ with $k=1$. Thus $P\left(s\right)$ has at most a simple pole at $s=1$. Expanding $log\zeta \left(s\right)=log\left({\displaystyle \frac{1}{s-1}}+O\left(1\right)\right)$ near $s=1$ shows $P\left(s\right)={(s-1)}^{-1}+O\left(1\right)$, so the residue is 1. Regularity at $s=0$ follows because $log\zeta \left(0\right)=-log\left(2\pi \right)/2$ is finite and the $k\ge 2$ terms are analytic at $s=0$.    □

Appendix E.0.0.53. E.4.1 Definition of the spectral action.

 Definition A23 

(Spectral action). Let $f:{\mathbb{R}}^{+}\to \mathbb{R}$ be a smooth, rapidly decreasing function with $f\left(0\right)=1$. For scale $\mathrm{\Lambda }>0$ set

$$S_f\left(\mathrm{\Lambda }\right):=Trf\left({\displaystyle \frac{{\mathbf{T}}^{\mathrm{Prime}}}{\mathrm{\Lambda }}}\right)=\sum _{p\mathrm{prime}}f\left(p/\mathrm{\Lambda }\right).$$

Because f is bounded, the trace converges absolutely.

Appendix E.0.0.55. E.4.3 Small–Λ expansion.

 Theorem A20 

(Spectral-action asymptotics). If $f\left(0\right)=1$ and f vanishes to order $m\ge 1$ at infinity, then as $\mathrm{\Lambda }\to 0^{+}$

$$S_f\left(\mathrm{\Lambda }\right)={\displaystyle \frac{1}{log(1/\mathrm{\Lambda })}}+\mathcal{O}\left({\displaystyle \frac{1}{{log}^2(1/\mathrm{\Lambda })}}\right).$$

 Proof. 

Close the contour in Proposition A24 to the left and pick the pole of $P\left(s\right)$ at $s=1$ (Theorem E.3.2). Since $\widehat{f}\left(1\right)={\int }_0^{\infty }{t}^{0}f\left(t\right)\mathrm{d}t=1$ by the normalisation of f, the residue contributes ${\mathrm{\Lambda }}^1/(1·log{\mathrm{\Lambda }}^{-1})$. Remaining integral along $\Re s=\sigma <1$ gives the error term stated, using the bound $P\left(s\right)=\mathcal{O}\left(1\right)$ for $\sigma \le 1-\delta$.    □

 Remark A6. 

The leading term matches the small–t prime heat-trace $Tr\left(e^{-t\mathbf{T}}\right)\sim {(tlog(1/t))}^{-1}$ after the identification $t={\mathrm{\Lambda }}^{-1}$.