A Szegő Limit Theorem for Radially-Compressed Toeplitz Operators

Trevor Camper · 2024

Read A Szegő Limit Theorem for Radially-Compressed Toeplitz Operators by Trevor Camper (2024) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2412-00612.txt.

Introduction

Trevor Camper proves limit theorems for compressed multiplication operators in a 2024 math study. Toeplitz operators encode multiplication followed by projection onto analytic functions, and compressing them in a radial way shrinks their domain. A classic theorem of Szego describes how eigenvalues spread for large truncations of such operators, a cornerstone of analysis.

Camper obtains the analogue inside reproducing kernel Hilbert spaces on discs. The eigenvalue density for these operators follows from that limit formula. Worked cases cover the Bergman space of the disc and the Segal-Bargmann-Fock space of the plane, two settings used across complex analysis and physics.

Catalog record

CREATORTrevor Camper
YEAR2024
COLLECTIONPAPERS
FORMATTXT
FILE SIZE22 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2412-00612

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

A Szegő Limit Theorem for Radially-Compressed Toeplitz Operators Trevor Camper Abstract . We obtain Szegő-type Limit Theorems in the setting of Reproducing Kernel Hilbert Spaces on discs in ℂ \mathbb{C} . From this, we derive a formula for the density of the eigenvalues of compressions of Toeplitz operators. Examples for the Bergman and Segal-Bargmann-Fock space are also presented. 1. Introduction Let c:=\{c_{n}\}_{n\geq 0} be a strictly positive sequence, i.e. c_{n}>0 for each n\in\mathbb{N}\cup\{0\} , such that the sequence \{c_{n+1}/c_{n}\} is monotonic. Let \displaystyle R=\lim_{n\to\infty}\frac{c_{n+1}}{c_{n}}, where R=\infty if the corresponding limit diverges. We assume that c:=\{c_{n}\}_{n\geq 0} is the moment sequence of an absolutely continuous, finite measure d\mu(r)=\mu(r)dr with \text{supp}(\mu)=[0,R] , i.e. \displaystyle c_{n}=\int_{0}^{R}r^{n}\mu(r)dr. Denote by \mathcal{H}_{c} the space of analytic functions with expansion \displaystyle f(z)=\sum_{n=0}^{\infty}a_{n}z^{n} for |z|<R such that \displaystyle\left\lVert f\right\rVert^{2}_{c}:=\sum_{n=0}^{\infty}|a_{n}|^{2}c_{2n+1}<\infty. \mathcal{H}_{c} is a Reproducing Kernel Hilbert space (RKHS) with inner product inherited from \left\lVert\cdot\right\rVert_{c} . Note that for the measure d\nu(z)=\mu(|z|)\frac{dz}{2\pi} we see \displaystyle\left\langle{\frac{z^{n}}{\sqrt{c_{2n+1}}}},{\frac{z^{\ell}}{\sqrt{c_{2\ell+1}}}}\right\rangle_{L^{2}(\nu)}=\frac{1}{\sqrt{c_{2k+1}c_{2\ell+1}}}\int_{0}^{R}r^{k+\ell+1}d\mu(r)\int_{0}^{2\pi}e^{i(k-\ell)\theta}\frac{d\theta}{2\pi}=\delta_{\ell,k}. Thus for f\in\mathcal{H}_{c} we see \displaystyle\left\langle{f},{\frac{z^{k}}{\sqrt{c_{2k+1}}}}\right\rangle_{L^{2}(\nu)}=a_{k}\frac{1}{\sqrt{c_{2k+1}}}\int_{0}^{R}r^{2k+1}d\mu(r)=a_{k}\sqrt{c_{2k+1}}. Hence, \displaystyle\left\lVert f\right\rVert^{2}_{c}=\sum_{n=0}^{\infty}|a_{k}|^{2}c_{2k+1}=\sum_{n=0}^{\infty}\left|\left\langle{f},{\frac{z^{n}}{\sqrt{c_{2n+1}}}}\right\rangle_{L^{2}(\nu)}\right|^{2}=\left\lVert f\right\rVert^{2}_{L^{2}(\mathbb{D}(0,R),\nu)}. From this, we can see that \mathcal{H}_{c} can be viewed as a closed subspace of L^{2}(\mathbb{D}(0,R),\nu) . The most interesting such spaces are when \mu(r)=2 (i.e., c_{2n+1}=(n+1)^{-1} and \mathcal{H}_{c} is the Bergman space A^{2}(\mathbb{D}) ), and when \mu(r)=2e^{-r^{2}} (i.e., c_{2n+1}=n! and \mathcal{H}_{c} is the Segal-Bargmann-Fock Space \mathcal{F}^{2}(\mathbb{C}) ). Our main object of study will be Toeplitz operators T_{\sigma} with \sigma\in L^{\infty}(\mathbb{D}(0,R)) acting on \mathcal{H}_{c} . Definition 1.1 . Let \sigma\in L^{\infty}(\mathbb{D}(0,R),\nu) , and let P:L^{2}(\mathbb{D}(0,R),\nu)\to\mathcal{H}_{c} be the projection onto \mathcal{H}_{c} . A Toeplitz operator T_{\sigma}:\mathcal{H}_{c}\to\mathcal{H}_{c} is the operator PM_{\sigma} , where M_{\sigma} is the multiplication operator on L^{2}(\mathbb{D}(0,R),\nu) associated to \sigma . The spectral asymptotics of operators is a well studied field, see [ 5 ] for a history of the subject and [ 2 , 4 ] for the case of Toeplitz operators acting on spaces similar to \mathcal{H}_{c} . In the case of the Hardy space \mathcal{H}^{2}(\mathbb{D}) , one of the most classical result concerning the spectral asymptotics of Toeplitz operators is the Szegő Limit Theorem [ 7 ] : if \psi:\mathbb{R}\to\mathbb{R} is continuous, and \sigma:\mathbb{T}\to\mathbb{R} is continuous, then (1) \displaystyle\lim_{N\to\infty}\frac{1}{N}\text{tr}\left(\psi\left(P_{N}T_{\sigma}P_{N}\right)\right)=\frac{1}{2\pi}\int_{0}^{2\pi}\psi\left(\sigma(x)\right)dx, where P_{N} denotes the projection onto \text{span}\{e^{ik\theta}:k=0,\pm 1,\cdots,\pm N\} . In this note, we will be interested in limits of a similar form. In the context of the Bergman and Segal-Bargmann-Fock spaces, the radial Toeplitz operators are those Toeplitz operators which are diagonalized by the monomials z^{n} . In keeping with the classical case, we will be concerned with compressions onto these eigenfunctions which motivates the following definition. Definition 1.2 . Let \mathcal{K}_{N}=\text{span}\{\frac{z^{n}}{\sqrt{c_{2n+1}}}:n=0,1,\cdots,N\} , let P_{N} denote the orthogonal projection onto \mathcal{K}_{N} , and let T_{\sigma} denote the Toeplitz operator with symbol \sigma\in L^{\infty}(\mathbb{D}) acting on \mathcal{H}_{c} . We call the operator P_{N}T_{\sigma}P_{N} a radially-compressed Toeplitz operator. As compared to the classical Szegő Limit Theorem, we will obtain results of the form: for \psi:\mathbb{R}\to\mathbb{C} and \sigma:\mathbb{D}(0,R)\to\mathbb{R} continuous, \displaystyle\lim_{N\to\infty}\frac{1}{N}\text{tr}\left(\psi\left(P_{N}T_{\sigma}P_{N}\right)\right)=\lim_{r\to R}\frac{1}{2\pi}\int_{0}^{2\pi}\psi\left(\sigma(re^{i\theta})\right)d\theta. Essentially, this statement says that the asymptotic spectral averages of radially-compressed Toeplitz operators is determined by the boundary behavior of the symbol. From this result, as is standard in the literature, we obtain the following Weyl law: for \lambda>0 , \displaystyle\lim_{N\to\infty}\frac{\#\{0\leq j\leq N:\lambda_{j}^{N}(\sigma)>\lambda\}}{N+1}=\frac{|\{\tilde{\sigma}>\lambda\}|}{2\pi}, where \{\lambda_{j}^{N}(\sigma)\}_{j} are the eigenvalues of the compression P_{N}T_{\sigma}P_{N} . The structure of this note is as follows. In Section 2, we will prove some preliminary facts that will be used throughout. In Section 3, we will present the proof strategy for our Szegő limit and obtain the associated spectral density for \mathcal{H}_{c} . In Section 4, we present the corresponding results for the Bergman and Segal-Bargmann-Fock spaces. 2. Preliminaries The strategy for proving our results relies on the following proposition proved below. Essentially, the vectors z^{n}/\sqrt{c_{2n+1}} give rise to probability measures \mu_{n} which experience mass escaping at the boundary of \mathbb{D}(0,R) and hence sample \sigma along this boundary. This motivates the following definition. Definition 2.1 . Let \mu_{n} be the family of probability measures on [0,R) satisfying d\mu_{n}(r)=\frac{1}{c_{2n+1}}r^{2n+1}d\mu(r) . Following this, we prove an approximation result which allows us to compare the trace of our radially-compressed Toeplitz operator with the trace of a compressed Toeplitz operator whose symbol is independent of the magnitude of the argument. Finally, identifying this Toeplitz operator with a matrix and invoking a general Szegő limit result of [ 1 ] gives the result. First, we make rigorous this notion of mass escaping at the boundary with the following lemma. Proposition 2.2 . Let \mu_{n} be defined as above. Then, 1. For any \tilde{R}<R , we have \mu_{n}\left([0,\tilde{R})\right)\to 0 as n\to\infty . 2. If R<\infty , then for any m\in\mathbb{N} \displaystyle\frac{c_{2l+m+1}}{\sqrt{c_{2l+2m+1}}\sqrt{c_{2l+1}}}\to 1, as l\to\infty . Proof. For the first point, choose \tilde{R}<R_{1}<R . Then, \displaystyle 0\leq\mu_{n}\left([0,\tilde{R})\right)=\frac{\int_{0}^{\tilde{R}}r^{2n+1}d\mu(r)}{\int_{0}^{R}r^{2n+1}d\mu(r)}\leq\frac{\mu([0,\tilde{R}))\left(\tilde{R}\right)^{2n+1}}{\int_{R_{1}}^{R}r^{2n+1}d\mu(r)}\lesssim\left(\frac{\tilde{R}}{R_{1}}\right)^{2n+1}\to 0 as n\to\infty . For the second point, note that \displaystyle\frac{c_{2l+m+1}}{\sqrt{c_{2l+2m+1}}\sqrt{c_{2l+1}}}=\frac{\frac{c_{2l+m+1}}{c_{2l+1}R^{m}}}{\sqrt{\frac{c_{2l+2m+1}}{c_{2l+1}R^{2m}}}}=\frac{\frac{c_{2l+2}}{c_{2l+1}R}\cdots\frac{c_{2l+m+1}}{c_{2l+m}R}}{\sqrt{\frac{c_{2l+2}}{c_{2l+1}R}\cdots\frac{c_{2l+2m+1}}{c_{2l+2m}R}}}\to 1, by the radius of convergence assumption. ∎ When R=\infty , we make the additional assumption that Part 2 of Proposition 2.2 is true. It is not obvious to the author whether or not this is true in general. However, the result is true in the case of the Segal-Bargmann-Fock space. Continuing on, the following lemma captures the mass escaping at the boundary property. Lemma 2.3 . The family of probability measures d\mu_{n} on [0,R) has a weak limit \delta_{R} in the following sense: if \sigma\in L^{\infty}([0,R)) such that \sigma(r)\to c<\infty as r\to R^{-} , then \displaystyle\int_{0}^{R}\sigma(r)d\mu_{n}(r)\to c, as n\to\infty . Proof. Let \sigma\in L^{\infty}([0,R)) such that \sigma(r)\to c<\infty as r\to R^{-} . Let \varepsilon>0 . Then there is an \tilde{R}<R such that |\sigma(r)-c|<\varepsilon/2 when r\in[\tilde{R},R) . Furthermore, by Proposition 2.2, there is an N such that for n\geq N \displaystyle\mu_{n}\left([0,\tilde{R})\right)<\frac{\varepsilon}{2\max\{c,\left\lVert\sigma\right\rVert_{\infty}\}}. Then, \displaystyle\left|\int_{0}^{R}\sigma(r)d\mu_{n}(r)-c\right|\leq \displaystyle\left|\int_{0}^{\tilde{R}}\left(\sigma(r)-c\right)d\mu_{n}(r)\right|+\left|\int_{\tilde{R}}^{R}\left(\sigma(r)-c\right)d\mu_{n}(r)\right| \displaystyle\leq \displaystyle 2\text{max}\{c,\left\lVert\sigma\right\rVert_{\infty}\}\mu_{n}\left([0,\tilde{R})\right)+\varepsilon/2\int_{\tilde{R}}^{R}d\mu_{n}(r) \displaystyle< \displaystyle\varepsilon, and the result is established. ∎ The Szegő limits presented in the introduction can be interpreted as spectral averages in some sense, which we will take advantage of going forward. In particular, we will need a weak convergence for an average of measures which our next lemma provides. Lemma 2.4 . Suppose \sigma\in L^{\infty}([0,R)) is such that \sigma(r)\to c as r\to R^{-} . Let \mu_{n} be as above. Then, \displaystyle\frac{1}{N+1}\sum_{n=0}^{N}\int_{0}^{R}\sigma(r)d\mu_{n}(r)\to c, as N\to\infty . Proof. This is a simple consequence of Lemma 3.1 and Cesàro convergence of a sequence. ∎ We now make the following simple observation. Note that \displaystyle\left\langle{T_{\sigma}\frac{z^{n}}{\sqrt{c_{2n+1}}}},{\frac{z^{n}}{\sqrt{c_{2n+1}}}}\right\rangle=\int_{\mathbb{D}(0,R)}\sigma(z)\frac{1}{c_{2n+1}}|z|^{2n}d\nu(z)=\int_{0}^{R}\hat{\sigma}(r)d\mu_{n}(r), where \displaystyle\hat{\sigma}(r)=\int_{0}^{2\pi}\sigma(re^{i\theta})\frac{d\theta}{2\pi}. Hence, if \lim_{r\to R^{-}}\hat{\sigma}(r) exists, we may apply the above results to obtain the required Szegő limit. This motivates the following definition. Definition 2.5 . Let \sigma\in L^{\infty}(\mathbb{D}(0,R)) be real-valued. We say \sigma has a radial limit if \lim_{r\to R^{-}}\sigma(re^{i\theta}) is well defined and we write \tilde{\sigma}(\theta):=\lim_{r\to R^{-}}\sigma(re^{i\theta}) . It is well known that such limits exist pointwise a.e. in the holomorphic and continuous cases. We now begin to construct our Szegő Limit Theorem. 3. A Szegő-type Limit Theorem Proposition 3.1 (Averaging Theorem) . Suppose \sigma\in L^{\infty}(\mathbb{D}(0,R)) has a radial limit \tilde{\sigma} . Then, \displaystyle\lim_{N\to\infty}\frac{1}{N+1}\text{tr}\left(P_{N}T_{\sigma}P_{N}\right)=\int_{0}^{2\pi}\tilde{\sigma}(\theta)\frac{d\theta}{2\pi}. Proof. First, note that \displaystyle\frac{1}{N+1}\text{tr}\left(P_{N}T_{\sigma}P_{N}\right)= \displaystyle\frac{1}{N+1}\sum_{n=0}^{N}\left\langle{T_{\sigma}\frac{z^{n}}{\sqrt{c_{2n+1}}}},{\frac{z^{n}}{\sqrt{c_{2n+1}}}}\right\rangle \displaystyle= \displaystyle\frac{1}{N+1}\sum_{n=0}^{N}\int_{\mathbb{D}(0,R)}\sigma(z)\frac{1}{c_{2n+1}}|z|^{2n}\mu(|z|)\frac{dz}{\pi} \displaystyle= \displaystyle\frac{1}{N+1}\sum_{n=0}^{N}\int_{0}^{R}\hat{\sigma}(r)d\mu_{n}(r), where \displaystyle\hat{\sigma}(r)=\int_{0}^{2\pi}\sigma(re^{i\theta})\frac{d\theta}{2\pi}. Applying Lemmas 2.3 and 2.4 give the result. ∎ Before proceeding to our Szegő limit result and the spectral density, we require two approximation results. The first, which is essentially a result of Janssen and Zelditch [ 6 ] , gives an approximation of the difference of the traces of monomials of the radially-compressed Toeplitz operators. Lemma 3.2 . Let

Read the complete text