bibkey: connesconsani2020quasiinner authors: Alain Connes and Caterina Consani year: 2020 title: Quasi-inner functions and local factors doi: null url: https://arxiv.org/abs/2008.10974v1 claim: Products of local-factor ratios containing the archimedean place have compact off-diagonal Hardy blocks, and their Sonin kernels form an injective inductive system. This supplies neither a signed trace comparison nor an unsmoothed trace-class estimate. strata_touched: [] license: citation-only triage: anchor

Quasi-inner local products and Sonin transport

The inspected primary manuscript is arXiv:2008.10974v1, 25 August 2020, 25 pages, SHA-256 c2f116d90e93600909449cb89983e02feea2ed950bf4ad6d91cb6c7cac7e1141. All locators below refer to this version; no publisher-edition identity is asserted. These are reused source results, without a new proof, numerical reproduction or Lean declaration. Selected statement and argument inspection is not a complete proof audit.

Quantitative compactness of the actual local product

For a half-plane or disk $\Omega$, the source calls a unimodular boundary function $u$ quasi-inner when $(I-P)uP$ is compact, with $P$ the orthogonal projection onto $H^2(\Omega )$ and $u$ acting by multiplication. This $P$ is a Hardy projection, distinct from the orthogonal projection onto physical Sonin vectors.

Write ${\rho }_v(z)={\gamma }_v(z)/{\gamma }_v(1-z)$ for the local-factor ratio. Theorem 4.8, manuscript p.22, treats $m\ge 1$ finite primes together with the archimedean place. It states that

$$u_F(z)={\rho }_{\infty }(z)\prod _{p\in F\setminus \{\infty \}}{\rho }_p(z)$$

is quasi-inner relative to $\mathrm{Re}z<1/2$, and that its off-diagonal block $(I-P)u_{F}P$ is an infinitesimal of order $1/(2m)$. Fact 3.6 records that an individual finite-place ratio is not quasi-inner. The full product and the individual factors therefore have different operator contracts.

The stated singular-value order is a compactness estimate. It does not by itself give a trace-class bound for that unsmoothed block, nor the sign of a dilation-smoothed trace correction. No positive ordering is part of Theorem 4.8.

Sonin kernels and the actual transport maps

Definition 5.2, manuscript p.23, defines the abstract Sonin space

$$\mathcal{S}(u)=\ker ((I-P)u(I-P))$$

on the complementary Hardy space. Theorem 5.3 on the same page proves that these kernels are infinite-dimensional for finite place sets containing $\infty$. For $F⊊F^{\prime }$, it gives an injective map

$$\mathcal{S}(u_F)⟶\mathcal{S}(u_{F^{\prime }}),\xi ⟼\left(\prod _{p\in F^{\prime }\setminus F}(1-p^{-z})\right)\xi .$$

This is an injective inductive system; the source does not assert that these multiplication maps are unitary or preserve the needed orthogonal trace.

Proposition 5.5, manuscript p.24, identifies the physical archimedean Sonin space $\mathcal{S}(1,1)$ with $\mathcal{S}({\rho }_{\infty })$ under the source unitary half-density and Mellin maps, equations (24)–(25). This supplies a physical-to-Hardy interface at the archimedean place. The later semilocal source supplies the physical ${\theta }_S$ isomorphism and explicitly different entire-function inner products.

What the compactness result does not settle

The physical projection supplier gives an explicit unweighted Sonin resolvent. These Hardy-block and transport results add genuine operator structure, but neither source estimates the orthogonal projection correction in the Euler-factor metric on a fixed common integrating test.

The archimedean signed comparison remains support-restricted; local time delay retains the arithmetic sign and correction. Combining separate compactness, positivity and transport statements does not supply the missing full same-test signed comparison. No such comparison or RH proof is claimed here.