The central heat trace on large compact classical groups

Thibaut Lemoine, Mylène Maïda · 2025

Read The central heat trace on large compact classical groups by Thibaut Lemoine, Mylène Maïda (2025) 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-2511-08288.txt.

Introduction

Thibaut Lemoine and Mylene Maida study the large size asymptotics of the central heat trace on compact classic groups, objects central to random matrix theory and mathematical physics. For each classic family inside the general linear group, they prove a full asymptotic expansion as the rank grows. The proof adapts a correspondence between highest weights and partitions, under which Laplace Beltrami eigenvalues stabilize as observables in an algebra of shifted symmetric functions.

A random surface representation then expresses the trace through ramified coverings of the torus. Applications include an explicit counting law for the Casimir spectrum. That law grows exponentially, unlike the polynomial fixed rank law. A rigorous Yang Mills Hurwitz duality on a torus follows, extended further toward Gromov Witten invariants.

Catalog record

CREATORThibaut Lemoine, Mylène Maïda
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE106 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2511-08288

Browse the full library · About the Tower of Babel

Full text

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

The central heat trace on large compact classical groups Thibaut Lemoine, Mylène Maïda Abstract We establish the large- N N asymptotic expansion of the (central) trace of the heat kernel on any compact classical group G N ⊂ GL N ​ ( ℂ ) G_{N}\subset\mathrm{GL}_{N}(\mathbb{C}) , which extends a previous result known only for U ⁡ ( N ) \mathrm{U}(N) [ 28 ] . It admits two new interpretations of the trace: in terms of ramified coverings of the torus, and Gromov–Witten invariants on elliptic curves. These connections allow us to explore several aspects of the gauge/string duality in two dimensions: a Yang–Mills/Hurwitz duality, and Yang–Mills/Gromov–Witten duality. 1 Introduction Let G_{N}\subset{\mathrm{GL}}_{N}(\mathbb{C}) be a compact matrix Lie group, such that its Lie algebra \mathfrak{g}_{N} is endowed with an invariant inner product (the explicit choice will be given in ( 5 )). It yields a Riemannian structure on G_{N} , so that the heat kernel p_{t}=e^{\frac{t}{2}\Delta_{G_{N}}} is expressed in terms of the corresponding Laplace–Beltrami operator. By standard results of representation theory, the characters \{\chi_{\lambda},\lambda\in\widehat{G}_{N}\} of the irreducible representations of G_{N} form an orthonormal family of eigenfunctions of \Delta_{G_{N}} , with associated eigenvalues \{-c_{2}(\lambda),\lambda\in\widehat{G}_{N}\} : \Delta\chi_{\lambda}(g)=-c_{2}(\lambda)\chi_{\lambda}(g),\quad\forall\lambda\in\widehat{G}_{N}\text{ and }\forall g\in G_{N}. The number c_{2}(\lambda)\geqslant 0 is called Casimir number 1 1 1 because it is the scalar representative of the (quadratic) Casimir element in the enveloping algebra \mathcal{U}(\mathfrak{g}_{N}) which is directly related to the Laplace–Beltrami operator. . For any fixed time t>0 , p_{t} is a convolution operator, acting in particular on two Hilbert spaces: L^{2}(G_{N}) on the one hand, and \mathcal{H}_{G_{N}}=\{f\in L^{2}(G_{N}):f(hgh^{-1})=f(g),\ \forall g,h\in G_{N}\} on the other hand. As an operator on L^{2}(G_{N}) , its trace is \mathrm{Tr}_{L^{2}(G_{N})}(p_{t})=\sum_{\lambda\in\widehat{G}_{N}}d_{\lambda}^{2}e^{-\frac{t}{2}c_{2}(\lambda)}, whereas as an operator on \mathcal{H}_{G_{N}} it is (1) \mathrm{Tr}_{\mathcal{H}_{G_{N}}}(p_{t})=\sum_{\lambda\in\widehat{G}_{N}}e^{-\frac{t}{2}c_{2}(\lambda)},\quad\forall t>0. In this paper, we will denote by \mathrm{Tr}(p_{t})=\mathrm{Tr}(e^{\frac{t}{2}\Delta}) the trace over \mathcal{H}_{G_{N}} , which will be the main protagonist of our results (Theorem 1.1 , Theorem 1.2 and Theorem 1.3 ). To make a distinction with the usual heat trace \mathrm{Tr}_{L^{2}(G_{N})}(p_{t}) , we shall call it the central heat trace on G_{N} , because we restrict to square integrable central functions. 1.1 Large- N expansion of the trace The trace of the heat kernel acting on L^{2}(G_{N}) is known to admit a short-time asymptotic expansion [ 16 ] , and large deviations when G_{N}={\mathrm{U}}(N) and N\to\infty [ 32 ] , but there are fewer results about the one acting on \mathcal{H}_{G_{N}} : for instance, as far as we are aware, its short-time asymptotics are unknown, and its large- N asymptotics have been mainly studied by theoretical physicists at the level of formal power series (see Section 4 ). The purpose of this paper is to investigate the large- N asymptotics of the central heat trace at a quantitative level. In the sequel, we shall use the following modified version of Landau’s asymptotic notations: if f,g:(0,\infty)\times{\mathbb{N}}\to\mathbb{R} are functions that depend on a real number t>0 and an integer N , we write f(t,N)=O_{t}(g(t,N)) as N\to\infty if for any t , there exists a constant C(t) that may depend on t and an integer N_{0} , such that |f(t,N)|\leqslant C(t)|g(t,N)|,\quad\forall t>0,\ \forall N\geqslant N_{0}. The fact that \mathrm{Tr}(e^{\frac{t}{2}\Delta_{G_{N}}}) converges to a finite value as N\to\infty has been first conjectured in physics [ 19 , 14 ] , then proved in [ 29 , 12 ] . Our first result is a far-reaching extension of the results of [ 12 ] , namely a full asymptotic expansion of the trace of the heat kernel for compact classical groups (including unitary, special unitary, special orthogonal and symplectic groups). Theorem 1.1 (See Theorem 2.9 ) . Let (G_{N})_{N\geqslant 1} be a sequence of compact classical groups with G_{N}\subset{\mathrm{GL}}_{N}(\mathbb{C}) . For any t>0 , there is an explicit family of coefficients (a_{k}(t))_{k\geqslant 0} (depending on the group type) such that for any p\geqslant 1 , the following asymptotic expansion holds as N tends to infinity: (2) \mathrm{Tr}(e^{\frac{t}{2}\Delta_{G_{N}}})=a_{0}(t)+\frac{a_{1}(t)}{N}+\ldots+\frac{a_{p}(t)}{N^{p}}+O_{t}\left(N^{-p-1}\right). The case of {\mathrm{U}}(N) was recently proved by the same authors [ 28 , Theorem 1.2] , and the proof of Theorem 1.1 relies on the same tools: a description of \mathrm{Tr}(e^{\frac{t}{2}\Delta}) in terms of random partitions, and a precise control of the remainder in terms of tail probabilities with respect to the corresponding measure. Although the expansion involves only even powers of \frac{1}{N} in the (special) unitary case, all powers of \frac{1}{N} appear for the special orthogonal and symplectic cases. By taking p=0 and letting N\to\infty , we also obtain that \mathrm{Tr}(e^{\frac{t}{2}\Delta_{G_{N}}})\to a_{0}(t) , and the limits agree with the results of [ 29 , 12 ] , see Corollary 2.12 . We must insist on the fact that it is only an asymptotic expansion , in the sense that there is no hope to let p tend to infinity with fixed N ; indeed, our estimates only work for N large enough, which is a condition that actually depends on p . However, the coefficients of the expansion do not depend on N or p . Although Theorem 1.1 is stated and proved for the heat kernel, it can be readily extended to more general measures, such as the central symmetric probability measures induced by infinitely divisible probability measures on \mathbb{R} introduced by Applebaum [ 6 ] — see Remark 2.11 . In the development of preliminary tools for the proof, we shall also encounter another result of asymptotic spectral theory of compact groups, namely an asymptotic spectral gap for the negative Laplacian -\Delta_{G_{N}} , given by Theorem 2.5 . 1.2 Random surface interpretation The second main result of this paper is a relation between the previous asymptotic expansion and a model of random surfaces, more specifically random ramified coverings. Let n\geqslant 1 and k\geqslant 0 be two fixed integers, and \Sigma_{g} be a compact connected orientable surface of genus g\geqslant 1 . The Hurwitz space \mathcal{H}_{g}(n,k) is the set of equivalence classes of ramified coverings X\to\Sigma_{g} of degree n with k ramification points, all assumed to be generic. It is finite (its size is the Hurwitz number H_{g}(n,k) ), and can be endowed with the uniform probability measure. Associated random models have already been studied, in particular in the large- n limit: for g=1 [ 15 , 1 ] , or g\geqslant 2 and k=0 [ 34 , 36 ] . In the present paper, we shall fix g=1 , but the novelty compared to the previously mentioned references is that we assume that n and k are random (following respectively a geometric distribution of parameter 1-e^{-\frac{t}{2}} and an even Poisson distribution of parameter t ). Conditionally to n and k , and assuming only generic ramifications, the random (equivalence classes of) ramified coverings are taken uniformly in \mathcal{H}_{1}(n,k) . The corresponding measure \rho_{t} (for a fixed real parameter t>0 ) will be defined and studied in Section 3.1 . Theorem 1.2 (See Theorem 3.3 ) . Let (G_{N})_{N\geqslant 1} be a sequence of compact classical groups with G_{N}\subset{\mathrm{GL}}_{N}(\mathbb{C}) . For any p\geqslant 1 and any t>0 : – If G_{N}={\mathrm{U}}(N) or {\mathrm{SU}}(N) , there is an explicit function \Phi_{t,N}^{p} (that depends on the group type) such that, as N tends to infinity, (3) \mathrm{Tr}(e^{\frac{t}{2}\Delta_{G_{N}}})=\int\Phi_{t,N}^{p}(X_{1},X_{2})N^{\chi(X_{1})+\chi(X_{2})}d\rho_{t}^{\otimes 2}(X_{1},X_{2})+O_{t}(N^{-2p-2}). – Otherwise, there is an explicit function \Phi_{t,N} (that depends on the group

Read the complete text