--- title: Scalar Poisson-Summation Certificates url: https://www.emergentmind.com/topics/scalar-poisson-summation-certificates type: topic --- # Scalar Poisson-Summation Certificates Scalar Poisson-summation certificates are scalar Fourier-analytic constructions in which a single function, measure, or scalar identity is used to certify a Poisson-type summation formula or an inequality derived from Poisson summation. In the setting of the Regev–Stephens-Davidowitz Gaussian mass maximality conjecture, the certificate is a single even Schwartz function \(h:\mathbb{R}^n\to\mathbb{R}\) designed to prove \[ \Theta_\Lambda(t)\le \Theta_{\mathbb{Z}^n}(t), \qquad \Theta_\Lambda(t)=\sum_{x\in \Lambda} e^{-t\|x\|^2}, \] simultaneously for all lattices in a prescribed class. The central result is a rigidity phenomenon: in dimensions \(n\ge 4\), any scalar certificate sharp at \(\mathbb{Z}^n\) must saturate every integer shell, and in dimensions \(n\ge 8\) this saturation is incompatible with the strict theta-series gap between \(\mathbb{Z}^8\) and \(E_8\), so no sharp scalar certificate exists [2605.26803]. ## 1. Lattice setting and certificate definition A full-rank lattice \(\Lambda\subset \mathbb{R}^n\) is called integral if \(\langle x,y\rangle\in\mathbb{Z}\) for all \(x,y\in\Lambda\), and unimodular if \(\mathrm{covol}(\Lambda)=1\). Its dual lattice is \[ \Lambda^*=\{\xi\in\mathbb{R}^n:\langle x,\xi\rangle\in\mathbb{Z}\ \text{for all }x\in\Lambda\}. \] If \(\Lambda\) is integral and unimodular, then \(\Lambda=\Lambda^*\) setwise. The conjecture of Regev and Stephens-Davidowitz asserts that for any integral lattice \(\Lambda\subset\mathbb{R}^n\) and all \(t>0\), \[ \Theta_\Lambda(t)\le \Theta_{\mathbb{Z}^n}(t). \] The same source emphasizes that this inequality is false shell-by-shell; in dimension \(8\), for example, \(E_8\) has more short vectors than \(\mathbb{Z}^8\). Any proof must therefore use cancellations across shells rather than termwise comparison [2605.26803]. The scalar Poisson-summation strategy fixes \(t>0\) and chooses a single even Schwartz function \(h:\mathbb{R}^n\to\mathbb{R}\) subject to two inequalities. The first is primal majorization, \[ h(x)\ge e^{-t\|x\|^2} \quad \text{for every nonzero }x\text{ in each unimodular integral }\Lambda\subset\mathbb{R}^n. \] The second is dual nonpositivity, \[ \widehat{h}(\xi)\le 0 \quad \text{for every nonzero }\xi\text{ in each unimodular integral }\Lambda\subset\mathbb{R}^n, \] where the Fourier transform is normalized by \[ \widehat{f}(\xi)=\int_{\mathbb{R}^n} f(x)\,e^{-2\pi i\langle x,\xi\rangle}\,dx. \] Poisson summation, \[ \sum_{x\in\Lambda} f(x)=\frac{1}{\mathrm{covol}(\Lambda)}\sum_{\xi\in \Lambda^*}\widehat{f}(\xi), \] then yields, for unimodular integral \(\Lambda\), \[ \Theta_\Lambda(t)\le 1+\widehat{h}(0)-h(0). \] The certificate is called sharp at \(\mathbb{Z}^n\) if \[ 1+\widehat{h}(0)-h(0)=\Theta_{\mathbb{Z}^n}(t). \] Sharpness is an extremality condition: because \(\Theta_{U\mathbb{Z}^n}(t)=\Theta_{\mathbb{Z}^n}(t)\) for every \(U\in O(n)\), equality must propagate to all rotated copies of \(\mathbb{Z}^n\). ## 2. Integral-shell saturation The rigidity theorem is formulated on the integer squared norm locus \[ \mathcal{I}_n:=\{x\in \mathbb{R}^n\setminus\{0\}:\|x\|^2\in\mathbb{Z}_{>0}\}. \] For \(n\ge 4\), Lagrange’s four-square theorem implies that every positive integer occurs as the squared norm of some \(z\in\mathbb{Z}^n\). Combined with \(O(n)\)-invariance, the union of the nonzero point sets of all rotations \(U\mathbb{Z}^n\) is exactly \(\mathcal{I}_n\) [2605.26803]. The saturation theorem states that if \(n\ge 4\), \(t>0\), and \(h\) is an even Schwartz function satisfying primal majorization and dual nonpositivity on every unimodular integral lattice, then sharpness at \(\mathbb{Z}^n\) forces \[ h(x)=e^{-t\|x\|^2} \quad\text{and}\quad \widehat{h}(x)=0 \qquad\text{for every }x\in\mathcal{I}_n. \] In the terminology of the paper, any sharp scalar certificate exhibits integral-shell saturation. The mechanism is termwise rigidity. One applies the certificate inequalities and Poisson summation on each self-dual lattice \(U\mathbb{Z}^n\). Since sharpness at \(\mathbb{Z}^n\) implies sharpness at \(U\mathbb{Z}^n\), the entire chain of inequalities collapses to equality. The relevant sums are absolutely convergent and consist of terms of fixed sign, so equality of sums forces equality term by term: \[ h(z)-e^{-t\|z\|^2}=0, \qquad \widehat{h}(z)=0 \quad \text{for all }z\in U\mathbb{Z}^n\setminus\{0\}. \] Rotation and four-squares then propagate this to all of \(\mathcal{I}_n\). The significance is that sharpness at one lattice does not remain local: it rigidifies the certificate on every nonzero shell of integer squared radius. ## 3. The \(E_8\) obstruction and the no-go theorem The obstruction in dimensions \(n\ge 8\) comes from the strict theta-series gap between \(\mathbb{Z}^8\) and \(E_8\). Writing \[ \tau=\frac{it}{\pi}, \qquad q=e^{\pi i\tau}=e^{-t}, \] and using the Jacobi theta nullwerte, \[ \theta_2(\tau)=\sum_{m\in\mathbb{Z}} q^{(m+\tfrac12)^2}, \qquad \theta_3(\tau)=\sum_{m\in\mathbb{Z}} q^{m^2}, \qquad \theta_4(\tau)=\sum_{m\in\mathbb{Z}} (-1)^m q^{m^2}, \] one has \[ \Theta_{\mathbb{Z}^8}(t)-\Theta_{E_8}(t)=\theta_2(\tau)^4\,\theta_4(\tau)^4>0 \qquad (t>0). \] The positivity follows from the series expansions for \(\theta_2,\theta_3\) and the product formula \[ \theta_4(\tau)=\prod_{m\ge 1}(1-q^{2m})(1-q^{2m-1})^2 \qquad (00\), every even Schwartz \(h\) satisfying the scalar certificate conditions must obey \[ 1+\widehat{h}(0)-h(0)>\Theta_{\mathbb{Z}^n}(t). \] Thus the sharp \(\mathbb{Z}^n\) Gaussian mass bound is unattainable by scalar Poisson certificates in dimensions \(n\ge 8\). ## 4. Stable lattices, graded families, and near-sharp schemes The obstruction is not confined to unimodular integral lattices in the narrow sense. A full-rank lattice \(\Lambda\subset\mathbb{R}^n\) is called stable if \(\mathrm{covol}(\Lambda)=1\) and every nonzero sublattice \(\Lambda'\subset \Lambda\) has covolume at least \(1\) in its span, equivalently \[ \sqrt{\det(\langle v_i,v_j\rangle)_{i,j=1}^r}\ge 1 \] for any basis \(v_1,\dots,v_r\) of \(\Lambda'\). Unimodular integral lattices are stable. In the stable setting one requires \[ h(x)\ge e^{-t\|x\|^2}\quad (x\in \Lambda\setminus\{0\}), \qquad \widehat{h}(\xi)\le 0\quad (\xi\in \Lambda^*\setminus\{0\}), \] for every stable \(\Lambda\). Because \(U\mathbb{Z}^n\) and \(E_8\oplus \mathbb{Z}^{n-8}\) are stable and self-dual, the saturation and no-go arguments apply verbatim [2605.26803]. The same paper extends the obstruction to orbit-constant graded families. Such a family assigns to each stable lattice \(\Lambda\) an even Schwartz function \(h_\Lambda\) satisfying orbit-constancy, \[ h_{U\Lambda}=h_\Lambda \qquad\text{for all }U\in O(n), \] together with primal majorization and dual nonpositivity on \(\Lambda\) and \(\Lambda^*\). If \[ 1+\widehat{h}_{\mathbb{Z}^n}(0)-h_{\mathbb{Z}^n}(0)=\Theta_{\mathbb{Z}^n}(t), \] then for \(\Lambda_c=E_8\oplus \mathbb{Z}^{n-8}\), \[ 1+\widehat{h}_{\Lambda_c}(0)-h_{\Lambda_c}(0)>\Theta_{\mathbb{Z}^n}(t). \] Hence no orbit-constant graded scalar family can both be sharp at \(\mathbb{Z}^n\) and certify the sharp inequality uniformly over all stable lattices. A further extension excludes uniformly summable near-sharp sequences. If \(\{h_j\}\) are even Schwartz functions satisfying the scalar certificate inequalities for all unimodular integral lattices, with slack \[ \varepsilon_j := 1+\widehat{h}_j(0)-h_j(0)-\Theta_{\mathbb{Z}^n}(t)\to 0^+, \] then on every integral shell one has \[ 0\le h_j(x)-e^{-t\|x\|^2}\le \varepsilon_j, \qquad 0\le -\widehat{h}_j(x)\le \varepsilon_j. \] Thus \(h_j(x)\to e^{-t\|x\|^2}\) and \(\widehat{h}_j(x)\to 0\) for all \(x\neq 0\) with \(\|x\|^2\in\mathbb{Z}\). If, moreover, the restrictions to \(\Lambda_c=E_8\oplus \mathbb{Z}^{n-8}\) are uniformly absolutely summable, dominated convergence plus Poisson summation forces \(\Theta_{\Lambda_c}(t)=\Theta_{\mathbb{Z}^n}(t)\), again contradicting the \(E_8\) gap. | Setting | Result | Range | |---|---|---| | Single ambient \(h\) on unimodular integral lattices | Integral-shell saturation | \(n\ge 4\) | | Single ambient \(h\) on unimodular integral lattices | No sharp certificate for the sharp \(\mathbb{Z}^n\) bound | \(n\ge 8\) | | Stable scalar certificates | Same saturation and no-go | \(n\ge 4\), \(n\ge 8\) | | Orbit-constant graded families | Sharp uniform certification excluded | \(n\ge 8\) | | Uniformly summable near-sharp sequences | Compact near-sharp scheme excluded | \(n\ge 8\) | ## 5. Scope, misconceptions, and routes beyond scalarity Several clarifications are essential. First, the Gaussian itself is not a scalar certificate. Although \[ g_t(x)=e^{-t\|x\|^2} \] satisfies primal majorization with equality, its Fourier transform is \[ \widehat{g}_t(\xi)=\left(\frac{\pi}{t}\right)^{n/2} e^{-\pi^2\|\xi\|^2/t}>0, \] so it violates the required dual nonpositivity. The obstruction therefore concerns the scalar certificate method, not the Gaussian test function itself [2605.26803]. Second, the no-go theorem does not refute the Gaussian mass maximality conjecture. It refutes a particular proof strategy based on a single ambient Schwartz function with pointwise primal and dual sign conditions. The same source explicitly contrasts scalar certificates with higher-order approaches. Scalar certificates observe only single-point data and cannot access pairwise inner products inherent to integrality, \(\langle x,y\rangle\in\mathbb{Z}\). Semidefinite-programming frameworks such as Bachoc–Vallentin retain angular information via harmonic kernels and positive semidefiniteness, and are presented as a potential route beyond scalar obstructions. Third, the obstruction has definite boundaries. It applies to orbit-constant families \(h_{U\Lambda}=h_\Lambda\), but not to the weaker equivariance condition \[ h_{U\Lambda}(x)=h_\Lambda(U^{-1}x). \] A fully equivariant, lattice-dependent scalar program remains outside the saturation mechanism. Likewise, the present rigidity relies on self-duality, so extending comparable arguments to nonunimodular integral lattices, where \(\Lambda\subsetneq \Lambda^*\) and \(\mathrm{covol}(\Lambda)>1\), is open. These limitations matter because they isolate precisely which scalar architectures are excluded and which remain formally unruled. ## 6. Related scalar certificate paradigms Across adjacent literatures, the phrase “scalar Poisson-summation certificate” denotes several closely related mechanisms rather than a single formal definition. This suggests a common template: scalar data are chosen so that Poisson summation yields either a pointwise identity, a trace identity, or an extremal inequality. | Setting | Scalar datum | Certified statement | |---|---|---| | LCA quotient \(G/\Gamma\) [1602.01252] | Continuous density \(p_t\) | PSF and probabilistic trace formula | | One-dimensional FS-pairs [2312.11185] | Hermite–Biehler data \(E=A-iB\) or holomorphic \(f\) | \(\int \widehat{\phi}\,d\mu=\sum a(x)\phi(x)\) | | \(k\)-spherical eigenmeasures [2405.15620] | Scalar modular-type Fourier series | \(\widehat{\mu}=\lambda\mu\) and Poisson-type summation | | Whittaker/KL index transforms [2407.14174] | Scalar index kernel \(W_{\mu,i\tau}(x)\) or \(K_{i\tau}(x)\) | Explicit PSF in the index variable | | Sum-of-squares formulas [2002.02324] | Odd tempered distribution \(o_k\) on \(\pm\sqrt{n}\) | Exact derivative-corrected Poisson-type identity | In Applebaum’s probabilistic setting on a locally compact abelian group \(G\) with discrete subgroup \(\Gamma\) and compact quotient \(X=G/\Gamma\), the scalar datum is the convolution-semigroup density \(p_t\). Under assumptions (A1) and (A2), one has the pointwise Poisson summation formula \[ \sum_{\gamma\in\Gamma} p_t(x\gamma) = \sum_{\chi\in\Gamma^\perp} e^{-t\psi(\chi)}\chi([x]), \] and at the identity this becomes the probabilistic trace certificate \[ p_t^X([e])=\operatorname{Tr}(T_t) = \sum_{\chi\in\Gamma^\perp} e^{-t\psi(\chi)}. \] The Gaussian on \(\mathbb{R}^d/\mathbb{Z}^d\) satisfies both the PSF and trace certificates; rotationally invariant \(\alpha\)-stable semigroups satisfy the trace formula but may fail the PSF; an adelic semistable construction can make both diverge [1602.01252]. In the one-dimensional classification of Fourier summation formulas, scalar certificates are encoded by almost periodic holomorphic functions and Hermite–Biehler data. A real-antipodal FS-pair \((\mu,a)\) with \(\deg(\mu)\le 2\) is classified through a holomorphic symbol \(f\) on the upper half-plane satisfying bounded-type and almost-periodicity conditions, or, in the nonnegative locally finite case, by a Hermite–Biehler entire function \(E=A-iB\) with \(iA/B\in AP\). The resulting certificate identity is \[ \int_{\mathbb{R}} \widehat{\phi}(t)\,d\mu(t)=\sum_{x\in\mathbb{R}} a(x)\phi(x) \qquad (\phi\in C_c^\infty(\mathbb{R})), \] with \(\mu\) reconstructed from the phase zeros of \(E\) and weights \(2\pi/\phi_E'(x)\) [2312.11185]. A different scalar paradigm appears in modular and metaplectic settings. For \(k\)-spherical measures, \[ \mu=\sum_{m=0}^\infty c_m\,\delta_{S^k_{\lambda_m}}, \qquad \widehat{\mu}=\sum_{m=0}^\infty \tilde c_m\,\delta_{S^k_{\tilde\lambda_m}}, \] the paper on measures and modular forms shows that the theta map \(\mu\mapsto \theta_\mu\) is an \(Mp_2(\mathbb{R})\)-equivariant isomorphism between such measures and modular-type Fourier series. A one-term modular transformation law under \(S\) is exactly a scalar eigenmeasure certificate: \[ \widehat{\mu}=\lambda \mu \quad\Longleftrightarrow\quad \theta_\mu(-1/\tau)=\lambda\,\sqrt{-i\tau}^{\,k}\,\theta_\mu(\tau). \] This recovers Poisson-type formulas associated with holomorphic modular forms, eta-products, and higher-dimensional Hilbert modular forms [2405.15620]. The index-transform literature provides another explicit scalar model. With the normalization \[ \widehat{f}(\xi)=\int_{\mathbb{R}} f(\tau)e^{-i\tau\xi}\,d\tau, \] the Whittaker-index paper proves Poisson summation identities for \(f(\tau)=W_{\mu,i\tau}(x)\) and, in the Kontorovich–Lebedev case, \[ \int_{-\infty}^{\infty} K_{i\tau}(x)e^{-i\tau\xi}\,d\tau=\pi e^{-x\cosh \xi}. \] Applying the scalar PSF gives \[ K_0(x)+2\sum_{n=1}^\infty K_{i\alpha n}(x) = \frac{\pi}{\alpha}e^{-x} + \frac{2\pi}{\alpha}\sum_{n=1}^\infty e^{-x\cosh(2\pi n/\alpha)}, \] together with related Whittaker, Olevskii, and Lommel summation formulas [2407.14174]. Lev–Reti’s odd-dimensional sum-of-squares formulas convert radial Poisson summation in \(\mathbb{R}^k\) into a scalar tempered distribution \[ o_k=-2\delta_0'+\sum_{n=1}^\infty r_k(n)n^{-1/2}(\delta_{\sqrt n}-\delta_{-\sqrt n}), \] whose Fourier transform is again supported on the same nodes \(\pm \sqrt n\) plus a derivative at \(0\). For odd \(k\ge 3\), pairing this identity with an odd Schwartz function yields an exact derivative-corrected Poisson-type summation formula with weights \(r_k(n)\) [2002.02324]. Two further comparison points sharpen the boundary of the scalar paradigm. Nori’s algebraic theory of \(t\)- and \(h\)-summability defines intrinsic scalar lattice sums without analytic continuation and then certifies them by a regularized tempered-distribution Poisson formula; the algebraic \(h\)-sum equals the regularized value \(T_{\mathrm{reg}}u(0,0)\) [1204.3533]. By contrast, the Braverman–Kazhdan and Schubert-variety literatures retain a certificate viewpoint but leave the scalar regime: the certifying objects are Schwartz spaces on nonabelian or quasi-affine spaces, normalized intertwining operators, and Eisenstein-residue boundary terms rather than a single ambient scalar function [1707.06091], [2107.01874]. Taken together, these developments show that scalar Poisson-summation certificates form a broad but sharply stratified category. In some settings they yield exact and highly structured identities; in the Gaussian mass maximality problem they instead produce a rigidity theorem and, beyond dimension \(7\), a definitive no-go result for the sharp scalar strategy. Source: https://www.emergentmind.com/topics/scalar-poisson-summation-certificates