Joint moments of characteristic polynomials from the orthogonal and unitary symplectic groups
THEODOROS ASSIOTIS, MUSTAFA ALPER GUNES, JONATHAN P. KEATING and FEI WEI · 2025
Read Joint moments of characteristic polynomials from the orthogonal and unitary symplectic groups by THEODOROS ASSIOTIS, MUSTAFA ALPER GUNES, JONATHAN P. KEATING and FEI WEI (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-2508-09910.txt.
Introduction
Theodoros Assiotis, Mustafa Alper Gunes, Jonathan Keating, and Fei Wei work on random matrix theory, which studies the statistical patterns of large matrices filled by chance. They focus on polynomials tied to each matrix, and on averages of products of them and their rates of change. The matrices come from two symmetry families, one symplectic and one orthogonal, and the question is how the averages behave as matrix size grows.
The authors derive formulas for that large-size limit and tie the leading terms to averages over a third, simpler ensemble. They also uncover an exact link to a differential equation, the Painlevé V equation, that holds at any matrix size. In the large limit a related equation, Painlevé III, governs the leading term instead. These links let them compute some averages exactly and build others step by step. As a payoff, they settle a question raised by earlier researchers.
Catalog record
| CREATOR | THEODOROS ASSIOTIS, MUSTAFA ALPER GUNES, JONATHAN P. KEATING and FEI WEI |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 150 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2508-09910
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Joint moments of characteristic polynomials from the orthogonal and unitary symplectic groups THEODOROS ASSIOTIS, MUSTAFA ALPER GUNES, JONATHAN P. KEATING and FEI WEI Abstract We establish asymptotic formulae for general joint moments of characteristic polynomials and their higher-order derivatives associated with matrices drawn randomly from the groups USp ( 2 N ) \mathrm{USp}(2N) and SO ( 2 N ) \mathrm{SO}(2N) in the limit as N → ∞ N\to\infty . This relates the leading-order asymptotic contribution in each case to averages over the Laguerre ensemble of random matrices. We uncover an exact connection between these joint moments and a solution of the σ \sigma -Painlevé V equation, valid for finite matrix size, as well as a connection between the leading-order asymptotic term and a solution of the σ \sigma -Painlevé III ′ equation in the limit as N → ∞ N\rightarrow\infty . These connections enable us to derive exact formulae for joint moments for finite matrix size and for the joint moments of certain random variables arising from the Bessel point process in a recursive way. As an application, we provide a positive answer to a question proposed by Altuğ et al. [Quarterly Journal of Mathematics, 65 (2014), 1111–1125]. 1 Introduction 1.1 Background and motivation In this paper we address the problem of calculating joint moments of the characteristic polynomials, along with their higher-order derivatives, of random matrices drawn from the unitary symplectic \mathrm{USp}(2N) and even orthogonal \mathrm{SO}(2N) classical compact groups, establishing a connection with Painlevé equations. The general question can be phrased as follows. Problem 1.1 . Given a sequence of N -indexed compact matrix groups (\mathrm{G}(N))_{N=1}^{\infty} of increasing dimension endowed with Haar measure \mu_{\mathrm{Haar}} , denote the characteristic polynomial of \mathbf{A}\in\mathrm{G}(N) by \phi_{\mathbf{A}}(z)=\det(\mathbf{I}-z\mathbf{A}) . Let h_{1},\dots,h_{m}\in\mathbb{R}_{\geq 0} . Then, the task is to establish the asymptotics of the matrix integral \int_{\mathrm{G}(N)}\prod_{k=0}^{m}\left|\phi_{\mathbf{A}}^{(k)}(1)\right|^{h_{k}}\mathrm{d}\mu_{\mathrm{Haar}}(\mathbf{A}), (1) as N\to\infty . We show that the leading order asymptotics of this integral takes the form \mathfrak{c}^{(G)}_{h_{0},\dots,h_{m}}N^{\delta^{(G)}_{h_{0},\dots,h_{m}}} and determine expressions for \mathfrak{c}^{(G)}_{h_{0},\dots,h_{m}} and \delta^{(G)}_{h_{0},\dots,h_{m}} . We also find exact expressions for the integrals in question in terms of solutions of certain Painlevé equations. These considerations, particularly the choice of evaluating \phi_{\mathbf{A}}(z) at z=1 , are motivated by conjectural connections to moments of \mathrm{L} -functions with different symmetry types, as studied in the work of Katz-Sarnak [ 46 ] and Keating-Snaith [ 47 ] . We discuss this further at the end of Section 1.2 . The analogous problem of joint moments over \mathrm{U}(N) , which relates to joint moments of the Riemann zeta and Hardy’s function [ 21 , 24 , 44 ] , has a long history starting from the work and conjectures of Hughes [ 43 ] from 2001, see [ 13 , 50 , 10 , 8 , 12 , 14 , 25 , 30 , 31 , 38 , 35 , 56 , 59 ] for part of the growing literature. The question for \mathrm{USp}(2N) and \mathrm{SO}(2N) has also received attention, see [ 1 , 4 , 40 ] and also [ 3 , 2 ] for the related problem of moments of the logarithmic derivative, but it has been overall much less well studied. In this paper we first settle the convergence aspect of the general question, see Theorem 1.3 , and give an expression for the limit in terms of concrete finite-dimensional averages. Secondly, we establish an exact connection between these joint moments and a solution of the \sigma -Painlevé V equation for finite matrix size, as well as a connection between the leading-order coefficients and a solution of the \sigma -Painlevé III ′ equation in the limit as N\rightarrow\infty . Finally, as an application, we give a positive answer to a question posed by Altuğ et al. in [ 1 ] . We remark that for the odd orthogonal group O^{-}(2N) , analogous results to those above can similarly be obtained by almost identical methods. To avoid excessive length, we omit a detailed discussion of this case. Our starting point is that the matrix integral ( 1 ) can, in the cases when \mathrm{G}(N)=\mathrm{USp}(2N) or \mathrm{G}(N)=\mathrm{SO}(2N) , be written in terms of averages involving singular negative power sum statistics of the Jacobi unitary ensemble of random matrices with certain parameters. Various such singular statistics of classical random matrix ensembles have been studied in the literature. For previous work and connections to integrable systems, orthogonal polynomials and combinatorics, see [ 53 , 52 , 17 , 26 , 27 , 35 , 20 , 18 ] . For physical motivations to study such quantities in relation to quantum transport in chaotic cavities, see [ 28 , 29 ] . The papers closest to our setup, which study the special case of the first negative power sum of the Jacobi ensemble, are the work of Forrester [ 35 ] , whose impetus is connections to multivariate hypergeometric functions, and [ 20 , 18 ] , which prove connections to Painlevé equations. Our approach to proving convergence of moments in [ 8 ] was based on ideas from the theory of exchangeable arrays. In some sense the role of the Jacobi ensemble in [ 8 ] is played by the Cauchy ensemble [ 60 , 34 ] , also known as the Hua-Pickrell ensemble [ 16 ] . However, these exchangeability techniques do not seem to apply to the Jacobi ensemble. Fortunately, mainly due to the positivity of eigenvalues in this setting (this is not present in the Cauchy case for \mathrm{U}(N) ), which we exploit at several places, we can take a more direct approach which uses as input symmetric function theory computations of Forrester [ 35 ] . Nevertheless, after we get a probabilistic expression for the limit it is possible to obtain some explicit evaluations by appealing again to exchangeability in a non-obvious way, see Proposition 1.4 and Corollary 1.8 . Regarding establishing the connection to Painlevé equations, we adapt our previous approach from [ 8 ] to the present setting. This approach is based on certain “partition-shifted” Hankel determinants, with the main difference being that the Hankel entries considered herein are truncated variants of the multivariate generating functions appearing in [ 8 ] . A complication arising from this truncation is that the “partition-shifted” Hankel determinants can no longer be expressed entirely in terms of the partial derivatives of the original Hankel determinants, as was the case in [ 8 ] . Instead, additional terms emerge, which are related to the truncation length. To ensure that this issue does not obstruct the inductive procedure central to our analysis, we proceed in two steps. First, we derive explicit expressions for the additional terms, expressed in terms of the truncation lengths. Second, since our inductive process proceeds with the evaluations of the partial derivatives of the “shifted Hankel determinants” at the origin, we choose truncation lengths that depend appropriately on the orders of the partial derivatives. This ensures that, after evaluation, the contribution from the aforementioned extra terms is zero. A more precise sketch of these strategies is provided in Section 4.1 . Finally, in a different direction, the limits of rescaled negative power sums and corresponding elementary symmetric polynomials are closely related to the Taylor coefficients