Table of Contents
Fetching ...

Numerical spectrums control Cohomological spectrums

Junyi Xie

TL;DR

The work addresses how spectral data of a surjective endomorphism on numerical Chow groups compare to spectral data on $l$-adic cohomology for smooth projective varieties. It develops a unified framework of numerical and cohomological spectra, radii, and polygon encodings via cohomological correspondences, enabling precise comparisons. The main result ties the radii $\alpha_j$ and $\beta_i$ through equalities $\log \alpha^-_{2i}(f)=\log \beta^-_i(f)$ and $\log \beta_i(f)=\log \alpha_{2i}(f)$, and shows the numerical and cohomological convex hulls coincide; this extends Deligne's Weil bound and confirms a Tate conjecture aspect for polarized endomorphisms. Consequences include sharp fixed-point counts for int-amplified and $q$-polarized endomorphisms, as well as moving-target variants, and the results generalize to a broad class of cohomological correspondences, bridging geometric dynamics on numerical groups with arithmetic cohomology.

Abstract

Let $X$ be a smooth irreducible projective variety over a field $\mathbf{k}$ of dimension $d.$ Let $τ: \mathbb{Q}_l\to \mathbb{C}$ be any field embedding. Let $f: X\to X$ be a surjective endomorphism. We show that for every $i=0,\dots,2d$, the spectral radius of $f^*$ on the numerical group $N^i(X)\otimes \mathbb{R}$ and on the $l$-adic cohomology group $H^{2i}(X_{\overline{\mathbf{k}}},\mathbb{Q}_l)\otimes \mathbb{C}$ are the same. As a consequence, if $f$ is $q$-polarized for some $q>1$, we show that the norm of every eigenvalue of $f^*$ on the $j$-th cohomology group is $q^{j/2}$ for all $j=0,\dots, 2d.$ This generalizes Deligne's theorem for Weil's Riemann Hypothesis to arbitary polarized endomorphisms and proves a conjecture of Tate. We also get some applications for the counting of fixed points and its ``moving target" variant. Indeed we studied the more general actions of certain cohomological coorespondences and we get the above results as consequences in the endomorphism setting.

Numerical spectrums control Cohomological spectrums

TL;DR

The work addresses how spectral data of a surjective endomorphism on numerical Chow groups compare to spectral data on -adic cohomology for smooth projective varieties. It develops a unified framework of numerical and cohomological spectra, radii, and polygon encodings via cohomological correspondences, enabling precise comparisons. The main result ties the radii and through equalities and , and shows the numerical and cohomological convex hulls coincide; this extends Deligne's Weil bound and confirms a Tate conjecture aspect for polarized endomorphisms. Consequences include sharp fixed-point counts for int-amplified and -polarized endomorphisms, as well as moving-target variants, and the results generalize to a broad class of cohomological correspondences, bridging geometric dynamics on numerical groups with arithmetic cohomology.

Abstract

Let be a smooth irreducible projective variety over a field of dimension Let be any field embedding. Let be a surjective endomorphism. We show that for every , the spectral radius of on the numerical group and on the -adic cohomology group are the same. As a consequence, if is -polarized for some , we show that the norm of every eigenvalue of on the -th cohomology group is for all This generalizes Deligne's theorem for Weil's Riemann Hypothesis to arbitary polarized endomorphisms and proves a conjecture of Tate. We also get some applications for the counting of fixed points and its ``moving target" variant. Indeed we studied the more general actions of certain cohomological coorespondences and we get the above results as consequences in the endomorphism setting.

Paper Structure

This paper contains 15 sections, 15 theorems, 106 equations.

Key Result

Theorem 1.2

For every $i=0,\dots, d$, we have For every odd $j\in G(X)$, we have

Theorems & Definitions (28)

  • Remark 1.1
  • Theorem 1.2: =Corollary \ref{['corendosum']}
  • Remark 1.3
  • Definition 1.4
  • Corollary 1.5
  • Corollary 1.7
  • Remark 1.8
  • Proposition 1.9: =Proposition \ref{['promovingtaget']}
  • Definition 1.10
  • Definition 1.11
  • ...and 18 more