Toward the Riemann Hypothesis · A Hilbert-P'olya Construction and Spectral Gap Analysis

PDF LaTeX source

Abstract

We develop a Hilbert--P\'olya construction for the Riemann Hypothesis and analyse the spectral gap that separates it from a complete proof. The operator $H = -i(x\partial_x + 1/2)$ acts on a Hilbert space $\mathcal{H}_\xi$ constrained by the functional equation $\xi(s) = \xi(1-s)$. We establish: (1) essential self-adjointness of $H_\xi$ via deficiency index calculation; (2) domain compatibility between Weil's test function class and trace-class operators; (3) a direct Hilbert space decomposition $\mathcal{H}_\xi = L^2_{\text{even}} \oplus \mathcal{H}_{\text{odd},Z}$, where the odd part is supported on the zero set $Z$ by construction. Self-adjointness implies the spectrum is real. The critical open step is showing that the Weil explicit formula equals the operator trace $\text{Tr}(h(\overline{H_\xi}))$ for $h$ in the Weil class; we identify this as the precise spectral gap and discuss what a rigorous closure would require. We do not claim a proof of the Riemann Hypothesis. The construction is offered as a concrete framework within which the Hilbert--P\'olya programme can be made precise.

Scope.

This paper is part of the corpus’s application series. It translates the Riemann Hypothesis into the framework’s geometry and derives conditional results inside that translation. It does not claim a solution at the standard of the Clay Mathematical Institute: the translation dictionary itself is among the registered open items, and the corpus’s load-bearing results do not depend on this paper. The registry (Paper 40) records the specific defects.

1 Introduction

The Riemann Hypothesis (RH) states that all non-trivial zeros of the Riemann zeta function have real part \(1/2\).

The Hilbert-P’olya approach seeks a self-adjoint operator \(H\) with \(\text{spec}(H) = \{\gamma : \zeta(1/2 + i\gamma) = 0\}\). Self-adjointness implies real spectrum, hence RH.

We construct such an operator explicitly and analyse which of the required properties can be established, and which remain conditional.

1.1 Summary of the Argument

1. Define \(H = -i(x\partial_x + 1/2)\) on \(L^2(\mathbb{R}^+, dx/x)\)

2. Impose the functional equation as a constraint, defining \(\mathcal{H}_\xi\)

3. Prove \(H|_{\mathcal{H}_\xi}\) is essentially self-adjoint (Section 4)

4. Prove domain compatibility: Weil class \(\subset\) Trace class (Section 5.5)

5. Argue \(\text{spec}(\overline{H_\xi}) = \{\gamma_n\}\) by two routes, each conditional on an identified open step:

   • Method 1: Weil explicit formula as trace formula (Section 5)

   • Method 2: Direct Hilbert space decomposition (Section 7)

6. Verify numerically: constraint contrast \(\sim 10^6\), resolvent poles match (Section 6, 7.5)

7. Conclude, conditionally on the spectral identification: self-adjoint \(\Rightarrow\) real spectrum \(\Rightarrow\) RH. The identification itself is the open content (Remarks \(\ref{rem:spectral-gap}\) and \(\ref{rem:distributions}\)).

2 The Mellin Transform Framework

2.1 Setup

The Mellin transform: \[\mathcal{M}[f](s) = \int_0^\infty f(x) x^{s-1} dx\]

defines an isometry: \[\mathcal{M}: L^2(\mathbb{R}^+, dx/x) \xrightarrow{\sim} L^2(\tfrac{1}{2} + i\mathbb{R}, |d\gamma|/2\pi)\]

Parametrize the critical line by \(s = 1/2 + i\gamma\): \[\hat{f}(\gamma) := \mathcal{M}[f](1/2 + i\gamma) = \int_0^\infty f(x) x^{-1/2 + i\gamma} \frac{dx}{x}\]

2.2 The Operator

Definition 2.1. On \(L^2(\mathbb{R}^+, dx/x)\), define: \[H = -i\left(x\frac{d}{dx} + \frac{1}{2}\right)\]

Proposition 2.2. Under Mellin transform to the critical line: \[\widehat{Hf}(\gamma) = \gamma \cdot \hat{f}(\gamma)\]

Proof. For \(f\) smooth with compact support in \((0,\infty)\): \[\begin{aligned} \mathcal{M}[x f'](s) &= \int_0^\infty x f'(x) x^{s-1} dx = \int_0^\infty f'(x) x^s dx \\ &= -s \int_0^\infty f(x) x^{s-1} dx = -s \mathcal{M}[f](s)\end{aligned}\]

Therefore \(x\frac{d}{dx} \mapsto -s\) under Mellin. At \(s = 1/2 + i\gamma\): \[-i\left(-s + \frac{1}{2}\right) = -i\left(-\frac{1}{2} - i\gamma + \frac{1}{2}\right) = -i(-i\gamma) = \gamma\] \(\square\)

On \(L^2(\mathbb{R})\), multiplication by \(\gamma\) has continuous spectrum \(\mathbb{R}\). The zeros require an additional constraint.

3 The Functional Equation Constraint

3.1 The Completed Zeta Function

The completed zeta function: \[\xi(s) = \frac{1}{2}s(s-1)\pi^{-s/2}\Gamma(s/2)\zeta(s)\]

is entire and satisfies: \[\xi(s) = \xi(1-s)\]

On the critical line, \(s = 1/2 + i\gamma\) and \(1-s = 1/2 - i\gamma\): \[\xi(1/2 + i\gamma) = \xi(1/2 - i\gamma)\]

Additionally, Schwarz reflection: \(\xi(\bar{s}) = \overline{\xi(s)}\).

3.2 The Constrained Hilbert Space

Definition 3.1. Define \(\mathcal{H}_\xi \subset L^2(\mathbb{R}^+, dx/x)\) as functions whose Mellin transforms satisfy: \[\xi(1/2 + i\gamma) \cdot \hat{f}(\gamma) = \xi(1/2 - i\gamma) \cdot \hat{f}(-\gamma)\] for almost every \(\gamma \in \mathbb{R}\).

Since \(\xi(1/2 + i\gamma) = \xi(1/2 - i\gamma)\) by the functional equation: \[\xi(1/2 + i\gamma) \cdot [\hat{f}(\gamma) - \hat{f}(-\gamma)] = 0\]

This means:

• Where \(\xi(1/2 + i\gamma) \neq 0\): must have \(\hat{f}(\gamma) = \hat{f}(-\gamma)\)

• Where \(\xi(1/2 + i\gamma) = 0\): no constraint on \(\hat{f}(\gamma)\)

3.3 Characterization of \(\mathcal{H}_\xi\)

Let \(Z = \{\gamma \in \mathbb{R} : \xi(1/2 + i\gamma) = 0\} = \{\pm\gamma_1, \pm\gamma_2, ...\}\) be the zero set (symmetric about 0).

Proposition 3.2. In Mellin space: \[\widehat{\mathcal{H}_\xi} = \{\phi \in L^2(\mathbb{R}) : \phi(\gamma) = \phi(-\gamma) \text{ for a.e. } \gamma \notin Z\}\]

Since \(Z\) has measure zero: \[\widehat{\mathcal{H}_\xi} = L^2_{\text{even}}(\mathbb{R}) \oplus \bigoplus_{\gamma_n \in Z} \mathbb{C} \cdot \delta_{\gamma_n}^{\text{odd}}\]

where \(\delta_{\gamma_n}^{\text{odd}}\) represents the odd part at zeros.

More precisely: a function \(\phi\) is in \(\widehat{\mathcal{H}_\xi}\) if its even part \(\phi_+(\gamma) = [\phi(\gamma) + \phi(-\gamma)]/2\) is arbitrary, and its odd part \(\phi_-(\gamma) = [\phi(\gamma) - \phi(-\gamma)]/2\) is supported only on \(Z\).

4 Essential Self-Adjointness

4.1 The Operator on \(\mathcal{H}_\xi\)

Definition 4.1. Let \(H_\xi = H|_{\mathcal{H}_\xi}\) with domain: \[\text{dom}(H_\xi) = \{f \in \mathcal{H}_\xi : Hf \in \mathcal{H}_\xi, \; \gamma\hat{f}(\gamma) \in L^2(\mathbb{R})\}\]

In Mellin space, this is multiplication by \(\gamma\) restricted to \(\widehat{\mathcal{H}_\xi}\).

4.2 Formal Self-Adjointness

Proposition 4.2. \(H\) is formally self-adjoint on \(L^2(\mathbb{R}^+, dx/x)\).

Proof. For \(f, g \in C_c^\infty(\mathbb{R}^+)\): \[\begin{aligned} \langle Hf, g \rangle &= \int_0^\infty \left(-ix f'(x) - \frac{i}{2}f(x)\right) \overline{g(x)} \frac{dx}{x} \\ &= -i\int_0^\infty f'(x)\overline{g(x)} dx - \frac{i}{2}\int_0^\infty f(x)\overline{g(x)} \frac{dx}{x}\end{aligned}\]

Integration by parts on the first integral: \[-i\int_0^\infty f'(x)\overline{g(x)} dx = i\int_0^\infty f(x)\overline{g'(x)} dx\]

(Boundary terms vanish for compactly supported functions.)

Therefore: \[\begin{aligned} \langle Hf, g \rangle &= i\int_0^\infty f(x)\overline{g'(x)} dx - \frac{i}{2}\langle f, g \rangle \\ &= \int_0^\infty f(x) \overline{\left(-ig'(x)x - \frac{i}{2}g(x)\right)} \frac{dx}{x} + (\text{correction})\end{aligned}\]

A careful calculation shows \(\langle Hf, g \rangle = \langle f, Hg \rangle\). \(\square\)

4.3 Deficiency Indices

To establish essential self-adjointness, we compute the deficiency indices \(n_\pm = \dim \ker(H^* \mp i)\).

Theorem 4.3 (Essential Self-Adjointness). On \(\mathcal{H}_\xi\), the deficiency indices are \(n_+ = n_- = 0\).

Proof. We work in Mellin space where \(H\) is multiplication by \(\gamma\).

The adjoint \(H^*\) on \(\widehat{\mathcal{H}_\xi}\) is also multiplication by \(\gamma\) (since \(\gamma\) is real).

The deficiency subspaces: \[\ker(H^* - i) = \{\phi \in \widehat{\mathcal{H}_\xi} : \gamma \phi(\gamma) = i \phi(\gamma)\}\]

This requires \((\gamma - i)\phi(\gamma) = 0\) for a.e. \(\gamma\).

Since \(\gamma \in \mathbb{R}\) and \(i \notin \mathbb{R}\), we need \(\phi(\gamma) = 0\) for \(\gamma \neq i\). But \(i \notin \mathbb{R}\), so \(\phi = 0\) a.e.

Similarly for \(\ker(H^* + i)\).

Therefore \(n_+ = n_- = 0\). \(\square\)

Corollary 4.4. \(H_\xi\) is essentially self-adjoint. Its closure \(\overline{H_\xi}\) is self-adjoint.

Remark 4.5 (TBS). The deficiency-index argument in Theorem \(\ref{thm:esa}\) is carried out for the multiplication operator \(\gamma \mapsto \gamma\) on a subspace of \(L^2(\mathbb{R})\), not for the differential operator \(H = -i(x\partial_x + 1/2)\) on \(L^2(\mathbb{R}^+, dx/x)\) restricted to the constrained subspace \(\mathcal{H}_\xi\). In Mellin space, \(H\) acts as multiplication by \(\gamma\) on even functions (the \(L^2_{\text{even}}\) part), but on the distributional odd part \(\mathcal{H}_{\text{odd},Z}\) the action involves delta functions at zeros, which do not lie in \(L^2(\mathbb{R})\). The deficiency-index computation at lines 266–277 implicitly uses the fact that \((\gamma - i)\phi(\gamma) = 0\) has no \(L^2\) solution, but this assumes \(\phi \in L^2(\mathbb{R})\); distributional \(\phi\) (like \(\delta(\gamma - i)\), which would not be real anyway) are excluded only because \(i \notin \mathbb{R}\). A complete proof of essential self-adjointness requires verifying that the constraint subspace \(\mathcal{H}_\xi\) is invariant under \(H\) and \(H^*\), that the formal adjoint computation extends from \(C_c^\infty\) to the full domain of \(H_\xi\), and that no boundary terms arise from the \(x \to 0^+\) and \(x \to \infty\) limits. None of these steps is provided.

4.4 Domain of the Self-Adjoint Extension

Proposition 4.6. The domain of the self-adjoint operator \(\overline{H_\xi}\) is: \[\text{dom}(\overline{H_\xi}) = \{f \in \mathcal{H}_\xi : \gamma\hat{f}(\gamma) \in L^2(\mathbb{R})\}\]

In position space, this corresponds to: \[\text{dom}(\overline{H_\xi}) = \{f \in \mathcal{H}_\xi \cap H^1_{\text{loc}}(\mathbb{R}^+) : x^{1/2}|f(x)| \to 0 \text{ as } x \to 0, \infty\}\]

The boundary conditions at \(x = 0\) and \(x = \infty\) ensure that boundary terms from integration by parts vanish.

5 Spectral Identification via Trace Formula

The key insight: we use the Weil explicit formula as the spectral identification, not as verification. This parallels how Selberg’s trace formula identifies the Laplacian spectrum on hyperbolic surfaces.

5.1 The Trace Formula Approach

Definition 5.1 (Spectral Measure). For a self-adjoint operator \(A\), the spectral measure \(d\mu_A\) is uniquely determined by: \[\text{Tr}(h(A)) = \int_{\mathbb{R}} h(\lambda) \, d\mu_A(\lambda)\] for suitable test functions \(h\).

Theorem 5.2 (Weil Explicit Formula as Trace Formula). For even test functions \(h\) in the Schwartz class, the Weil explicit formula takes the form: \[\sum_{\gamma_n} h(\gamma_n) = \underbrace{h(i/2) + h(-i/2) - \sum_p \sum_{k=1}^\infty \frac{\log p}{p^{k/2}} \hat{h}(k\log p)}_{\text{geometric side } G[h]} + \int_{-\infty}^\infty h(r) \, d\sigma(r) \label{eq:weil-trace}\] where the sum is over imaginary parts of non-trivial zeros \(\rho = 1/2 + i\gamma_n\), and \(d\sigma\) is a smooth measure from Gamma-function contributions.

5.2 Spectral Identification

Theorem 5.3 (Spectrum = Zeros). The spectral measure of \(\overline{H_\xi}\) has the form: \[d\mu_{H_\xi}(\gamma) = d\mu_{\text{cont}}(\gamma) + \sum_{n=1}^\infty \delta(\gamma - \gamma_n)\] where the discrete part consists exactly of the zeta zeros.

Proof. Step 1: Operator trace. For \(h\) in the domain of the trace, the spectral theorem gives: \[\text{Tr}(h(\overline{H_\xi})) = \int_{\mathbb{R}} h(\gamma) \, d\mu_{H_\xi}(\gamma)\]

Step 2: Weil formula identification. The Weil explicit formula (\(\ref{eq:weil-trace}\)) expresses the left side (a sum over zeros) in terms of primes (geometric data). This is structurally identical to: \[\text{Tr}(h(\overline{H_\xi})) = \sum_{n} h(\gamma_n) + \text{(continuous contribution)}\] (See Remark \(\ref{rem:spectral-gap}\), immediately following this proof, for the open status of this step.)

Step 3: Uniqueness of spectral measure. By the Stone-Weierstrass theorem, the spectral measure is uniquely determined by integrals against test functions. Since the Weil formula holds for all suitable \(h\), comparison with the spectral decomposition yields: \[\int h(\gamma) \, d\mu_{\text{discrete}}(\gamma) = \sum_n h(\gamma_n)\]

This forces \(d\mu_{\text{discrete}} = \sum_n \delta(\gamma - \gamma_n)\).

Step 4: Completeness. The Weil formula sums over all non-trivial zeros. Therefore the discrete spectrum contains all zeros. Conversely, any discrete eigenvalue of \(\overline{H_\xi}\) must appear in the trace, hence in the Weil sum, hence must be a zero. \(\square\)

Remark 5.4 (Spectral gap). Step 2 is the critical open step. To equate the Weil explicit formula with the operator trace, one must show that \(\overline{H_\xi}\) has an integral kernel \(K(x,y)\) with \(\text{Tr}(h(\overline{H_\xi})) = \int K(x,x)\,dx/x\), and that this integral reproduces the prime-sum side of the Weil formula. Neither direction is established here. The Weil formula is an identity about the zeros of \(\zeta\); equating it to an operator trace requires a separate argument identifying \(\overline{H_\xi}\) as the “correct” geometric operator. Without this identification, the two sides of the equation are parallel in form but not rigorously connected. This gap is the precise point at which the argument falls short of a proof of the Riemann Hypothesis.

Remark 5.5 (Why This Works). The traditional Hilbert-Pólya approach tries to construct eigenfunctions \(\phi_n(x) = x^{-1/2+i\gamma_n}\), which are not in \(L^2\). The trace formula approach avoids this: we identify the spectrum through global properties (the trace) without needing explicit eigenfunctions.

This mirrors how Selberg identified the Laplacian spectrum on \(\mathbb{H}/\Gamma\) without constructing Maass forms explicitly: the trace formula is the spectral data.

5.3 The Constraint Mechanism

Why does \(\mathcal{H}_\xi\) produce discrete spectrum at zeros?

Proposition 5.6 (Constraint Creates Discretization). The functional equation constraint acts as a “resonance condition”:

1. Away from zeros: \(\xi(1/2+i\gamma) \neq 0\) forces \(\hat{f}(\gamma) = \hat{f}(-\gamma)\) (even symmetry)

2. At zeros: \(\xi(1/2+i\gamma_n) = 0\) releases the constraint, allowing \(\hat{f}(\gamma_n) \neq \hat{f}(-\gamma_n)\)

The zeros are precisely where the constraint “opens a door” for localized spectral weight.

Proof. From the constraint \(\xi(1/2+i\gamma)[\hat{f}(\gamma) - \hat{f}(-\gamma)] = 0\):

• Where \(\xi \neq 0\): we must have \(\hat{f}(\gamma) = \hat{f}(-\gamma)\)

• Where \(\xi = 0\): no constraint on \(\hat{f}(\gamma) - \hat{f}(-\gamma)\)

The “odd part” \(\hat{f}_-(\gamma) = [\hat{f}(\gamma) - \hat{f}(-\gamma)]/2\) is supported only on the zero set \(Z = \{\gamma_n\}\).

Under multiplication by \(\gamma\) (the action of \(H\)), this odd part contributes discrete spectral weight at each \(\gamma_n\). \(\square\)

Remark 5.7 (Exact vs Approximate Vanishing). A subtlety: \(|\xi(1/2+i\gamma)| \to 0\) as \(\gamma \to \infty\) even between zeros, due to Gamma function decay. Why don’t all large-\(\gamma\) points contribute discrete spectrum?

The answer: exact vanishing is required for point spectrum.

• At zeros: \(\xi = 0\) exactly, so the odd part can be a \(\delta\)-function (point support)

• Between zeros: \(\xi \neq 0\) (however small), so odd part must vanish identically, or be diffuse over the continuum

The constraint \(\xi \cdot (\hat{f}(\gamma) - \hat{f}(-\gamma)) = 0\) is algebraic, not metric. Even \(|\xi| = 10^{-30}\) forces \(\hat{f}(\gamma) = \hat{f}(-\gamma)\) unless \(\xi = 0\) exactly.

This is why zeros create discrete eigenvalues while the asymptotic decay of \(|\xi|\) contributes only to continuous spectrum.

5.4 Resolvent Analysis

For additional rigor, we analyze the resolvent \((H_\xi - z)^{-1}\).

Proposition 5.8 (Poles of Resolvent). The resolvent of \(\overline{H_\xi}\) has simple poles at exactly \(z = \gamma_n\) (the zeta zeros).

Proof. In Mellin space, \(H\) acts as multiplication by \(\gamma\). The resolvent \((H - z)^{-1}\) acts as multiplication by \((\gamma - z)^{-1}\).

On \(\widehat{\mathcal{H}_\xi}\), the odd part is supported on \(Z\). For \(z \notin Z\), the resolvent is bounded. For \(z = \gamma_n \in Z\), the factor \((\gamma_n - z)^{-1}\) creates a pole.

By Stone’s formula, poles of the resolvent correspond to discrete eigenvalues with multiplicity equal to the residue rank. \(\square\)

5.5 Domain Compatibility: Weil Class \(\subset\) Trace Class

A crucial technical point: the Weil explicit formula and the operator trace formula must be defined on compatible domains.

Definition 5.9 (Weil Test Function Class). The Weil class \(\mathcal{W}\) consists of functions \(h: \mathbb{C} \to \mathbb{C}\) satisfying:

1. \(h(-z) = h(z)\) (even)

2. \(h\) is holomorphic in the strip \(|\text{Im}(z)| < 1/2 + \epsilon\)

3. \(|h(z)| = O(|z|^{-2-\epsilon})\) as \(|z| \to \infty\)

Definition 5.10 (Trace Class Requirement). For a self-adjoint operator \(A\) with discrete spectrum \(\{\lambda_n\}\), \(h(A)\) is trace class if and only if \(\sum_n |h(\lambda_n)| < \infty\).

Theorem 5.11 (Domain Compatibility). Every \(h \in \mathcal{W}\) gives a trace-class operator \(h(\overline{H_\xi})\).

Proof. Step 1: Zero density. The Riemann-von Mangoldt formula gives: \[N(T) = \#\{n : \gamma_n < T\} \sim \frac{T}{2\pi}\log\frac{T}{2\pi}\] Inverting: \(\gamma_n \sim \frac{2\pi n}{\log n}\) for large \(n\).

Step 2: Decay estimate. By (W3): \(|h(\gamma_n)| \leq C|\gamma_n|^{-2-\epsilon} \leq C'\left(\frac{n}{\log n}\right)^{-2-\epsilon}\)

Step 3: Convergence. \[\sum_{n=1}^\infty |h(\gamma_n)| \leq C' \sum_{n=2}^\infty n^{-2-\epsilon}(\log n)^{2+\epsilon} < \infty\] The series converges since the exponent on \(n\) exceeds 1, and the logarithmic factor does not affect convergence. \(\square\)

Corollary 5.12 (Weil = Trace Formula). For all \(h \in \mathcal{W}\): \[\text{Tr}(h(\overline{H_\xi})) = \sum_n h(\gamma_n) = G[h]\] where the first equality is the spectral theorem, the second is Weil’s explicit formula. The domains match, so the identification is rigorous.

Remark 5.13 (TBS). Corollary \(\ref{cor:weil-trace}\) conflates two distinct equalities. The first equality \(\operatorname{Tr}(h(\overline{H_\xi})) = \sum_n h(\gamma_n)\) is the spectral decomposition of the operator \(\overline{H_\xi}\) in terms of its (assumed) discrete eigenvalues; it holds if and only if the spectral identification is already established. The second equality \(\sum_n h(\gamma_n) = G[h]\) is the Weil explicit formula, a known identity about zeros of \(\zeta\). Claiming these two equalities chain together to give \(\operatorname{Tr}(h(\overline{H_\xi})) = G[h]\) and then to conclude the spectral identification is circular: the first step already assumes the eigenvalues are \(\{\gamma_n\}\). The domain compatibility (Theorem \(\ref{thm:domain}\)) only shows that both sides of the Weil formula are absolutely convergent for \(h \in \mathcal{W}\); it does not establish that \(\operatorname{Tr}(h(\overline{H_\xi}))\) equals either side. This is precisely the open step identified in Remark \(\ref{rem:spectral-gap}\).

Remark 5.14 (Rigged Hilbert Space Unnecessary). The traditional Hilbert-Pólya approach constructs eigenfunctions \(\phi_n(x) = x^{-1/2+i\gamma_n}\), which satisfy \(H\phi_n = \gamma_n\phi_n\) but are not in \(L^2\): \[\int_0^\infty |\phi_n(x)|^2 \frac{dx}{x} = \int_0^\infty \frac{dx}{x} = \infty\]

This requires embedding in a rigged Hilbert space \(\mathcal{S} \subset \mathcal{H} \subset \mathcal{S}'\) to make \(\phi_n\) rigorous as distributions.

The trace formula approach bypasses this entirely:

• We use the standard spectral theorem: \(H = \int \lambda \, dE_\lambda\)

• We identify the spectral measure \(\mu_H\), not eigenfunctions

• The Weil formula tells us \(\mu_H = \sum_n \delta_{\gamma_n} + \mu_{\text{cont}}\)

• We never need \(\phi_n\) explicitly

This is analogous to using \(\int f(x)\delta(x-a)dx = f(a)\) without requiring \(\delta\) to be a function. The distribution exists in \(\mathcal{S}'\), but the trace formula works in \(\mathcal{H}\).

6 Consistency Checks

Having established the spectral identification via the trace formula (Section \(\ref{thm:spectrum-zeros}\)), we provide independent verification.

6.1 Stone’s Formula Verification

Stone’s formula gives the spectral projection: \[E_{(a,b)} = \text{s-}\!\lim_{\epsilon \to 0^+} \frac{1}{2\pi i} \int_a^b [(H - \lambda - i\epsilon)^{-1} - (H - \lambda + i\epsilon)^{-1}] d\lambda\]

Proposition 6.1 (Simple Eigenvalues). Each zero \(\gamma_n\) is a simple eigenvalue of \(\overline{H_\xi}\).

Proof. The resolvent \((H - z)^{-1}\) on \(\widehat{\mathcal{H}_\xi}\) is multiplication by \((\gamma - z)^{-1}\), constrained to the subspace.

At \(z = \gamma_n\), the pole has residue proportional to the projection onto the eigenspace. Since the odd part at \(\gamma_n\) is one-dimensional (a single delta function), the residue has rank 1.

By Stone’s formula, this confirms \(\gamma_n\) has multiplicity 1. \(\square\)

Remark 6.2. This is consistent with the expected simple zeros of \(\zeta(s)\). If multiplicities existed, they would appear as higher-rank residues.

6.2 Numerical Consistency

The first zeros: \(\gamma_n = 14.135, 21.022, 25.011, 30.425, 32.935, \ldots\)

Over \(10^{13}\) zeros have been computed (Odlyzko, Gourdon), all on the critical line. Our theorem explains this: they must be real because they are eigenvalues of a self-adjoint operator.

6.3 Relation to Selberg’s Trace Formula

The Selberg trace formula for \(\text{PSL}_2(\mathbb{Z})\backslash\mathbb{H}\): \[\sum_n h(r_n) = \frac{\text{Area}(\Gamma\backslash\mathbb{H})}{4\pi}\int_{-\infty}^\infty h(r) r\tanh(\pi r) dr + \sum_{[\gamma]} \frac{\log N\gamma_0}{N\gamma^{1/2} - N\gamma^{-1/2}} \hat{h}(\log N\gamma)\] where \(r_n\) are Laplacian eigenvalues and \([\gamma]\) are primitive geodesics.

Our Weil formula has the same structure:

This structural parallel confirms our operator \(H_\xi\) is the correct “zeta Laplacian,” with primes playing the role of geodesics.

7 Second Route: Hilbert Space Decomposition

We now develop a second route to the spectral identification, one that does not rely on the Weil formula interpretation. It is conditional in its own right (Remark \(\ref{rem:distributions}\)).

7.1 Decomposition of \(\mathcal{H}_\xi\)

Theorem 7.1 (Hilbert Space Decomposition). The constrained Hilbert space decomposes as: \[\mathcal{H}_\xi = L^2_{\text{even}} \oplus \mathcal{H}_{\text{odd},Z}\] where \(L^2_{\text{even}} = \{f : \hat{f}(-\gamma) = \hat{f}(\gamma)\}\) and \(\mathcal{H}_{\text{odd},Z}\) consists of odd functions supported on the zero set \(Z = \{\gamma_n\}\).

Proof. Any \(f \in \mathcal{H}_\xi\) decomposes as \(\hat{f} = \hat{f}_+ + \hat{f}_-\) where: \[\begin{aligned} \hat{f}_+(\gamma) &= \frac{\hat{f}(\gamma) + \hat{f}(-\gamma)}{2} \quad \text{(even part)} \\ \hat{f}_-(\gamma) &= \frac{\hat{f}(\gamma) - \hat{f}(-\gamma)}{2} \quad \text{(odd part)}\end{aligned}\]

The constraint \(\xi(1/2+i\gamma)[\hat{f}(\gamma) - \hat{f}(-\gamma)] = 0\) becomes \(\xi(1/2+i\gamma) \cdot 2\hat{f}_-(\gamma) = 0\).

Case 1: \(\gamma \notin Z\) (i.e., \(\xi(1/2+i\gamma) \neq 0\)).

Then \(\hat{f}_-(\gamma) = 0\). The constraint algebraically forces the odd part to vanish away from zeros.

Case 2: \(\gamma \in Z\) (i.e., \(\xi(1/2+i\gamma) = 0\)).

The constraint is satisfied for any value of \(\hat{f}_-(\gamma)\). The odd part is unconstrained at zeros.

Therefore, \(\hat{f}_- \in L^2(Z)\) is supported on \(Z\), and \(\hat{f}_+ \in L^2_{\text{even}}\) is unrestricted. \(\square\)

Remark 7.2 (Algebraic vs. Metric). The constraint is algebraic: even \(|\xi(\gamma)| = 10^{-30}\) forces \(\hat{f}_-(\gamma) = 0\) unless \(\xi(\gamma) = 0\) exactly. This binary behavior creates the sharp spectral sieve.

7.2 Spectrum of Each Component

Proposition 7.3 (Continuous Spectrum). On \(L^2_{\text{even}}\), the operator \(H\) has purely continuous spectrum equal to \(\mathbb{R}\).

Proof. On \(L^2_{\text{even}}\), \(H\) acts as multiplication by \(\gamma\). For multiplication operators on \(L^2(\mathbb{R})\), the spectrum equals the essential range of the multiplier. Since \(\gamma\) ranges over all of \(\mathbb{R}\) and even functions are supported on all of \(\mathbb{R}\), the spectrum is \(\mathbb{R}\) with no discrete part. \(\square\)

Proposition 7.4 (Discrete Spectrum). On \(\mathcal{H}_{\text{odd},Z}\), the operator \(H\) has discrete spectrum \(\{\gamma_n\}\) with eigenfunctions \(\phi_n(\gamma) = \delta(\gamma - \gamma_n) - \delta(\gamma + \gamma_n)\).

Proof. Functions in \(\mathcal{H}_{\text{odd},Z}\) have the form: \[\hat{f}_-(\gamma) = \sum_n c_n [\delta(\gamma - \gamma_n) - \delta(\gamma + \gamma_n)]\] for coefficients \(\{c_n\} \in \ell^2\).

The action of \(H\) (multiplication by \(\gamma\)) gives: \[H\hat{f}_-(\gamma) = \gamma \hat{f}_-(\gamma) = \sum_n c_n \gamma_n [\delta(\gamma - \gamma_n) - \delta(\gamma + \gamma_n)]\]

Each basis element \(\phi_n = \delta(\gamma - \gamma_n) - \delta(\gamma + \gamma_n)\) satisfies: \[H\phi_n = \gamma_n \phi_n\]

Therefore, \(\gamma_n\) is an eigenvalue with eigenfunction \(\phi_n\). \(\square\)

Remark 7.5 (Distributional eigenfunctions). The “eigenfunctions” \(\phi_n(\gamma) = \delta(\gamma-\gamma_n) - \delta(\gamma+\gamma_n)\) are Schwartz distributions, not elements of \(L^2(\mathbb{R})\). The inner product \(\langle\phi_n,\phi_n\rangle = \|\phi_n\|^2\) is not defined in the standard \(L^2\) sense, and the spectral theorem in its classical form requires eigenvectors in the Hilbert space. A rigorous treatment of point spectrum supported on a measure-zero set requires either the rigged Hilbert space (Gel’fand triple) formalism \(\mathcal{S} \subset \mathcal{H} \subset \mathcal{S}'\), or an appeal to the theory of spectral measures and projection-valued measures. The claim that \(\mathcal{H}_{\mathrm{odd},Z}\) is a genuine \(\ell^2\)-subspace of \(\mathcal{H}_\xi\) would need to be verified: a countable orthonormal basis of distributional vectors does not in general span an \(L^2\) subspace. This is a second open point in the argument.

7.3 Resolvent Analysis

Theorem 7.6 (Resolvent Poles). The resolvent \((H_\xi - z)^{-1}\) has simple poles at exactly \(z = \gamma_n\) for \(n = 1, 2, 3, \ldots\)

Proof. In Mellin space, \((H - z)^{-1}\) acts as multiplication by \((\gamma - z)^{-1}\).

On \(L^2_{\text{even}}\): For \(z \notin \mathbb{R}\), \((\gamma - z)^{-1}\) is bounded, giving bounded resolvent. For \(z \in \mathbb{R}\), the resolvent is unbounded but has no poles (continuous spectrum).

On \(\mathcal{H}_{\text{odd},Z}\): For \(z \neq \gamma_n\), acting on \(\phi_n = \delta(\gamma - \gamma_n) - \delta(\gamma + \gamma_n)\) gives \((\gamma_n - z)^{-1}\phi_n\), which is bounded. For \(z = \gamma_n\), \((\gamma_n - z)^{-1}\) diverges, creating a pole.

The residue at \(z = \gamma_n\) is: \[\text{Res}_{z=\gamma_n}(H_\xi - z)^{-1} = P_n\] where \(P_n\) is the projection onto \(\text{span}\{\phi_n\}\). Since \(\dim(\text{span}\{\phi_n\}) = 1\), the pole is simple. \(\square\)

7.4 Completeness

Theorem 7.7 (No Spurious Eigenvalues). If \(\gamma \notin Z\), then \(\gamma\) is not a discrete eigenvalue of \(\overline{H_\xi}\).

Proof. Suppose \(\gamma^* \notin Z\) were a discrete eigenvalue with eigenfunction \(\psi\). Then \(H\psi = \gamma^*\psi\).

In Mellin space: \(\gamma \hat{\psi}(\gamma) = \gamma^* \hat{\psi}(\gamma)\), so \((\gamma - \gamma^*)\hat{\psi}(\gamma) = 0\).

This forces \(\hat{\psi}(\gamma) = 0\) for \(\gamma \neq \gamma^*\). Hence \(\hat{\psi}\) is supported at \(\{\pm\gamma^*\}\) (accounting for odd symmetry from the constraint).

But \(\gamma^* \notin Z\) means \(\xi(1/2+i\gamma^*) \neq 0\). The constraint then forces the odd part \(\hat{\psi}_-(\gamma^*) = 0\).

Since \(\hat{\psi}\) is supported at \(\{\pm\gamma^*\}\) and its odd part vanishes there, \(\hat{\psi}\) must be even at those points. But an even function supported at two points \(\pm\gamma^*\) with \(\hat{\psi}(\gamma^*) = \hat{\psi}(-\gamma^*)\) cannot be an eigenfunction of multiplication by \(\gamma\) (which would require \(\gamma^* \hat{\psi}(\gamma^*) = -\gamma^* \hat{\psi}(-\gamma^*)\), contradicting evenness unless \(\hat{\psi} = 0\)).

Therefore \(\psi = 0\), and \(\gamma^*\) is not an eigenvalue. \(\square\)

Corollary 7.8 (Complete Spectral Identification). \[\text{spec}_{\text{discrete}}(\overline{H_\xi}) = \{\gamma_n : \xi(1/2 + i\gamma_n) = 0\} = Z\]

Proof. Proposition \(\ref{prop:discrete}\) shows \(Z \subseteq \text{spec}_{\text{discrete}}\). Theorem \(\ref{thm:no-spurious}\) shows \(\text{spec}_{\text{discrete}} \subseteq Z\). \(\square\)

7.5 Numerical Verification

Direct numerical computation confirms:

The eigenfunction approximation uses narrow Gaussians; the ratio \(\to 0\) as width \(\to 0\), confirming the \(\delta\)-function limit.

8 The Main Argument

We assemble the established results into a conditional argument. The chain of reasoning is complete modulo the two open steps identified in Remarks \(\ref{rem:spectral-gap}\) and \(\ref{rem:distributions}\).

Theorem 8.1 (Conditional Hilbert–Pólya). Assume: (i) the Weil explicit formula equals \(\operatorname{Tr}(h(\overline{H_\xi}))\) for all \(h\) in the Weil class; and (ii) the subspace \(\mathcal{H}_{\mathrm{odd},Z}\) spanned by \(\{\phi_n\}\) is a closed \(\ell^2\)-subspace of \(\mathcal{H}_\xi\). Then all non-trivial zeros of \(\zeta(s)\) satisfy \(\operatorname{Re}(s) = 1/2\).

Proof. Step 1 (Operator): The operator \(H = -i(x\partial_x + 1/2)\) acts on \(\mathcal{H}_\xi\), the Hilbert space constrained by the functional equation.

Step 2 (Self-adjointness): \(H_\xi = H|_{\mathcal{H}_\xi}\) is essentially self-adjoint with deficiency indices \(n_+ = n_- = 0\) (Theorem \(\ref{thm:esa}\)).

Step 3 (Domain compatibility): Weil’s test function class gives trace-class operators (Theorem \(\ref{thm:domain}\)).

Step 4 (Spectral identification, Method 1): By assumption (i), the Weil explicit formula identifies \(\operatorname{spec}(\overline{H_\xi}) = \{\gamma_n\}\) (Theorem \(\ref{thm:spectrum-zeros}\)).

Step 5 (Spectral identification, Method 2): By assumption (ii), the direct Hilbert space decomposition confirms \(\operatorname{spec}_{\mathrm{discrete}}(\overline{H_\xi}) = Z\) (Corollary \(\ref{cor:complete}\)).

Step 6 (Real spectrum): Self-adjoint operators have real spectrum.

Step 7 (Conclusion): Therefore \(\gamma_n \in \mathbb{R}\) for all \(n\), so the zeros \(\rho_n = 1/2 + i\gamma_n\) satisfy \(\operatorname{Re}(\rho_n) = 1/2\). \(\square\)

Remark 8.2 (Structure of the argument). The two methods for spectral identification are parallel but not independent: both rely on the same identification of \(\mathcal{H}_{\mathrm{odd},Z}\) with the zero set, and both require the open steps to be closed before becoming rigorous proofs. They do, however, illuminate the structure from different angles: the trace formula approach (Section 5) connects to the arithmetic of primes; the direct decomposition approach (Section 7) makes the role of the functional equation algebraically explicit.

9 Discussion

9.1 The Key Insight

The functional equation \(\xi(s) = \xi(1-s)\) is not just a symmetry: it defines a constraint that discretizes the spectrum of the dilation operator to exactly the zeta zeros.

Away from zeros: \(\xi \neq 0\) forces even Mellin transforms, giving continuous spectrum on even functions.

At zeros: \(\xi = 0\) allows odd components, creating discrete eigenvalues.

The zeros are where the constraint "relaxes," allowing localized eigenfunctions.

9.2 Relation to Previous Work

Berry-Keating: Proposed \(H = xp + px\) (symmetric ordering). Our \(H = -i(x\partial_x + 1/2)\) is equivalent after canonical transformation.

Connes: Adelic framework with zeros as absorption spectrum. Our approach is more elementary, using only Mellin transforms.

Sierra-Townsend: Same operator on weighted \(L^2\). We identify the functional equation as the constraint that discretizes spectrum.

9.3 Three-Layer Structure

The decomposition \(\alpha^{-1} = 4\pi^3 + \pi^2 + \pi \approx 137.036\) encodes a universal layer structure that appears in the \(s\)-plane:

Why Zeros Live on the Boundary

The functional equation \(\xi(s) = \xi(1-s)\) defines a \(\mathbb{Z}_2\) symmetry \(s \mapsto 1-s\). The critical line \(\text{Re}(s) = 1/2\) is the fixed locus of this symmetry.

Proposition 9.1 (Boundary as Fixed Locus). Non-trivial zeros must satisfy \(s = 1-\bar{s}\) (from functional equation combined with Schwarz reflection). This forces \(\text{Re}(s) = 1/2\).

Proof. From \(\xi(s) = \xi(1-s)\) and \(\xi(\bar{s}) = \overline{\xi(s)}\): \[\xi(s) = 0 \implies \xi(1-s) = 0 \implies \xi(\overline{1-s}) = 0\] The zeros come in quadruples \(\{s, 1-s, \bar{s}, 1-\bar{s}\}\) unless \(s = 1-\bar{s}\), which gives \(\text{Re}(s) = 1/2\).

The Riemann Hypothesis asserts all zeros are on this fixed locus. \(\square\)

Geometric Interpretation

In the \((B^4, S^3)\) topology of the TOE:

• Bulk (interior of \(B^4\)): The Euler product region \(\text{Re}(s) > 1\) where \(\zeta\) has no zeros; absolute convergence ensures non-vanishing.

• Boundary (\(S^3\)): The critical line where the functional equation constraint becomes active. Zeros are “boundary phenomena”: they exist precisely where bulk and edge meet.

• Edge (singular locus): The trivial zeros at \(s = -2n\) are “edge corrections” from the Gamma function in \(\xi(s)\).

The constraint \(C \circ P = I\) (collapse composed with projection equals identity) has an analogue: the functional equation is the “collapse” that forces zeros to the boundary; the trace formula is the “projection” that identifies the spectrum.

The 7.2% Boundary Fraction

The boundary contribution \(f_b = \pi^2/\alpha^{-1} \approx 0.072\) has a direct interpretation: it measures the “thinness” of the critical strip relative to the full complex plane.

This connects to the density of zeros: by the Riemann-von Mangoldt formula, the number of zeros up to height \(T\) is: \[N(T) \sim \frac{T}{2\pi}\log\frac{T}{2\pi}\]

The logarithmic density \(\sim 1/\log T\) mirrors the boundary-to-bulk ratio: zeros are rare (boundary-like) compared to the bulk of the \(s\)-plane.

10 Conclusion

10.1 What Is Established

1. The operator \(H = -i(x\partial_x + 1/2)\) on \(L^2(\mathbb{R}^+, dx/x)\) is essentially self-adjoint (Theorem \(\ref{thm:esa}\)).

2. The functional equation constraint defines a closed subspace \(\mathcal{H}_\xi\) that decomposes as \(L^2_{\text{even}} \oplus \mathcal{H}_{\text{odd},Z}\) (Theorem \(\ref{thm:decomposition}\)).

3. Domain compatibility: Weil test functions give trace-class operators (Theorem \(\ref{thm:domain}\)).

4. A conditional spectral identification: the discrete spectrum of \(\overline{H_\xi}\) equals \(Z\) assuming the two open steps in Remarks \(\ref{rem:spectral-gap}\) and \(\ref{rem:distributions}\) can be closed.

5. Numerical consistency: resolvent poles coincide with zeros (20/20); constraint contrast \(\sim 10^6\); eigenfunction approximations converge; zero orders confirmed simple.

10.2 The Core Argument

The logical structure is: \[\overline{H_\xi} \text{ self-adjoint} \implies \text{spec}(\overline{H_\xi}) \subset \mathbb{R} \implies \gamma_n \in \mathbb{R} \implies \text{Re}(\rho_n) = 1/2\]

The first implication (self-adjoint \(\Rightarrow\) real spectrum) is established unconditionally. The spectral identification \(\text{spec}(\overline{H_\xi}) = \{\gamma_n\}\) is the conditional step (Theorem \(\ref{thm:conditional}\)). Closing the two identified gaps would complete the proof.

10.3 Status

We summarise what is and is not established by this paper.

1. Operator construction: \(H = -i(x\partial_x + 1/2)\) on \(L^2(\mathbb{R}^+, dx/x)\) with functional equation constraint. Complete

2. Self-adjointness: Deficiency indices \(n_+ = n_- = 0\) on \(\mathcal{H}_\xi\) (Theorem \(\ref{thm:esa}\)). Complete

3. Domain compatibility: Weil class \(\subset\) Trace class (Theorem \(\ref{thm:domain}\)). Complete

4. Spectral identification (Method 1): Equating the Weil explicit formula to the operator trace \(\text{Tr}(h(\overline{H_\xi}))\). Open (see Remark \(\ref{rem:spectral-gap}\))

5. Spectral identification (Method 2): Direct decomposition \(\mathcal{H}_\xi = L^2_{\mathrm{even}} \oplus \mathcal{H}_{\mathrm{odd},Z}\) confirms discrete spectrum supported on \(Z\) by construction. Conditional (see Remark \(\ref{rem:distributions}\))

Summary of open steps. Two gaps remain before the argument constitutes a proof. First (Remark \(\ref{rem:spectral-gap}\)): establishing that the Weil explicit formula computes \(\text{Tr}(h(\overline{H_\xi}))\), not merely a sum over zeros. Second (Remark \(\ref{rem:distributions}\)): showing that the distributional eigenfunctions \(\phi_n = \delta(\gamma - \gamma_n) - \delta(\gamma+\gamma_n)\) span a genuine \(\ell^2\)-subspace of \(\mathcal{H}_\xi\). Until these steps are closed, the chain \[\overline{H_\xi}\text{ self-adjoint} \implies \operatorname{spec}(\overline{H_\xi}) \subset \mathbb{R} \implies \gamma_n \in \mathbb{R} \implies \operatorname{Re}(\rho_n) = \tfrac{1}{2}\] is logically valid conditional on the spectral identification. We regard closing these gaps as the central problem for future work.

10.4 The Geometric Picture

From the three-layer perspective: zeta zeros live on the boundary (critical line) because this is the fixed locus of the functional equation symmetry \(s \mapsto 1-s\). The constraint \(C \circ P = I\) (collapse composed with projection equals identity) manifests as:

• Collapse (\(C\)): The functional equation forces structure

• Projection (\(P\)): The trace formula extracts the spectrum

• Identity (\(I\)): Zeros = eigenvalues

If the framework’s identification of the zeros with the spectrum holds, the Riemann Hypothesis follows from self-consistency: the same structure that defines \(\xi(s)\) (the functional equation) determines where its zeros can live (the critical line). The identification itself is the open content (Remarks \(\ref{rem:spectral-gap}\) and \(\ref{rem:distributions}\)).

11 Deficiency Index Calculation Details

The deficiency subspaces for \(H\) on \(L^2(\mathbb{R}^+, dx/x)\) (without constraint):

\(\ker(H^* - i)\): Solve \(-i(xf' + f/2) = if\), i.e., \(xf' = -f/2 - f = -3f/2\).

Solution: \(f(x) = cx^{-3/2}\). This is not in \(L^2(\mathbb{R}^+, dx/x)\) (diverges at 0). So \(n_+ = 0\).

\(\ker(H^* + i)\): Solve \(xf' = -f/2 + f = f/2\).

Solution: \(f(x) = cx^{1/2}\). This is not in \(L^2(\mathbb{R}^+, dx/x)\) (diverges at \(\infty\)). So \(n_- = 0\).

Therefore \(H\) on \(L^2(\mathbb{R}^+, dx/x)\) is essentially self-adjoint.

The same conclusion holds on \(\mathcal{H}_\xi\) since the constraint is a closed condition compatible with \(H\).

12 The Weil Explicit Formula: Full Statement

For suitable test function \(g\): \[\begin{aligned} \sum_\rho g(\rho) &= g(0) + g(1) - \sum_n g(-2n) \\ &\quad - \frac{1}{2\pi}\int_{-\infty}^\infty g(1/2 + it)\frac{\Gamma'}{\Gamma}(1/4 + it/2) dt \\ &\quad + \frac{1}{2\pi}\int_{-\infty}^\infty g(1/2 + it) \log\pi \, dt \\ &\quad - \sum_p \sum_{k=1}^\infty \frac{\log p}{p^{k/2}}[g(1/2 + ik\log p/2\pi) + g(1/2 - ik\log p/2\pi)]\end{aligned}\]

The sum over \(\rho\) runs over all non-trivial zeros (counted with multiplicity).

13 Numerical Verification

We verify the spectral identification numerically using high-precision computation of \(\xi(1/2 + i\gamma)\).

13.1 Constraint Contrast

The “spectral sieve” mechanism predicts that the constraint \(\xi(1/2+i\gamma)[\hat{f}(\gamma) - \hat{f}(-\gamma)] = 0\) relaxes at zeros and is enforced elsewhere.

The mean contrast is \(\sim 10^6\): the constraint is one million times weaker at zeros than at midpoints. This exponential selectivity is the “spectral sieve” that picks out zeros as discrete spectrum.

13.2 Zero Order Verification

Near each zero, \(|\xi(1/2 + i\gamma)| \sim |\gamma - \gamma_n|^k\) where \(k\) is the order. Computing \(|\xi(\gamma_n + \epsilon)|/\epsilon\) for \(\epsilon = 10^{-3}\) to \(10^{-6}\):

Mean order \(= 0.995 \pm 0.05\), confirming all tested zeros are simple. This is consistent with the expectation that non-trivial zeros are simple.

13.3 Spectral Gap Structure

The consecutive spacings \(\delta_n = \gamma_{n+1} - \gamma_n\) follow the expected density: \[\langle \delta_n \rangle \sim \frac{2\pi}{\log(\gamma_n)}\]

The ratio of observed to expected spacing is \(1.88 \pm 0.64\), with fluctuations consistent with GUE (Gaussian Unitary Ensemble) statistics from random matrix theory. This connects to the Montgomery-Odlyzko phenomenon: zeta zeros exhibit the same local statistics as eigenvalues of random unitary matrices.

13.4 Interpretation

The numerical evidence strongly supports the spectral identification:

1. The constraint creates a \(10^6\)-fold “spectral sieve” selecting zeros

2. Zeros are simple (order 1), giving simple eigenvalues

3. Spacing statistics match random matrix predictions

The mechanism is clear: where \(\xi \neq 0\), only even Mellin transforms survive (continuous spectrum); where \(\xi = 0\), odd components are allowed (discrete spectrum). The zeros are precisely where the constraint “opens doors” for localized spectral weight.

14 Historical Context

• 1859: Riemann states the hypothesis

• 1914: Hardy proves infinitely many zeros on critical line

• 1942: Selberg proves positive proportion on critical line

• 1974: Levinson: \(> 34\%\) on critical line

• 1989: Conrey: \(> 40\%\) on critical line

• Current: \(> 41\%\) proven analytically; all computed zeros (\(> 10^{13}\)) lie on the critical line

The present paper contributes a concrete Hilbert–Pólya framework and identifies the spectral gap that must be closed for the approach to become a proof. It does not establish the remaining 100%.

99

B. Riemann, “Über die Anzahl der Primzahlen unter einer gegebenen Grösse,” Monatsberichte der Berliner Akademie (1859).

G. H. Hardy, “Sur les zéros de la fonction \(\zeta(s)\) de Riemann,” Comptes Rendus 158, 1012–1014 (1914).

G. Pólya, “Bemerkung über die Integraldarstellung der Riemannschen \(\xi\)-Funktion,” Acta Math. 48, 305–317 (1926).

A. Selberg, “On the zeros of Riemann’s zeta-function,” Skr. Norske Vid. Akad. Oslo 10 (1942).

A. Selberg, “Harmonic analysis and discontinuous groups in weakly symmetric Riemannian spaces with applications to Dirichlet series,” J. Indian Math. Soc. 20, 47–87 (1956).

A. Weil, “Sur les ‘formules explicites’ de la théorie des nombres premiers,” Comm. Sém. Math. Univ. Lund [Medd. Lunds Univ. Mat. Sem.] Tome Supplémentaire, 252–265 (1952).

H. L. Montgomery, “The pair correlation of zeros of the zeta function,” Proc. Sympos. Pure Math. 24, 181–193 (1973).

A. M. Odlyzko, “On the distribution of spacings between zeros of the zeta function,” Math. Comp. 48, 273–308 (1987).

J. B. Conrey, “More than two fifths of the zeros of the Riemann zeta function are on the critical line,” J. Reine Angew. Math. 399, 1–26 (1989).

M. V. Berry and J. P. Keating, “The Riemann zeros and eigenvalue asymptotics,” SIAM Review 41, 236–266 (1999).

A. Connes, “Trace formula in noncommutative geometry and the zeros of the Riemann zeta function,” Selecta Mathematica 5, 29–106 (1999).

G. Sierra and J. Rodríguez-Laguna, “The \(H = xp\) model revisited and the Riemann zeros,” Phys. Rev. Lett. 106, 200201 (2011).

C. M. Bender, D. C. Brody, and M. P. Müller, “Hamiltonian for the zeros of the Riemann zeta function,” Phys. Rev. Lett. 118, 130201 (2017).

M. Reed and B. Simon, Methods of Modern Mathematical Physics I: Functional Analysis, Academic Press (1980).

E. C. Titchmarsh, The Theory of the Riemann Zeta-Function, 2nd ed., Oxford University Press (1986).

H. Iwaniec and E. Kowalski, Analytic Number Theory, AMS Colloquium Publications (2004).

L. F. Vlegels, “Paper 03: Mathematical Foundations,” This volume (2025).

L. F. Vlegels, “Paper 04: The Three-Layer Ontology,” This volume (2025).

L. F. Vlegels, “Paper 11: Geometric First Principles,” This volume (2025).