Dissipative Spectral Form Factor of the Complex Elliptic Ginibre Ensemble across Various Non-Hermiticity Regimes
Gernot Akemann, Sung-Soo Byun, Seungjoon Oh · 2026
Read Dissipative Spectral Form Factor of the Complex Elliptic Ginibre Ensemble across Various Non-Hermiticity Regimes by Gernot Akemann, Sung-Soo Byun, Seungjoon Oh (2026) 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-2605-28319.txt.
Introduction
Gernot Akemann, Sung-Soo Byun, and Seungjoon Oh chart the dissipative spectral form factor of the complex elliptic Ginibre ensemble across non-Hermiticity regimes. They derive precise asymptotics for disconnected and connected parts, characterizing the dip-ramp-plateau shape with dip and Heisenberg times. A mesoscopic regime interpolates between non-Hermitian and Hermitian spectral behavior, with ramp shape varying by scaling.
Catalog record
| CREATOR | Gernot Akemann, Sung-Soo Byun, Seungjoon Oh |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 90952 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.28319
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Dissipative Spectral Form Factor of the Complex Elliptic Ginibre Ensemble across Various Non-Hermiticity Regimes Gernot Akemann, Sung-Soo Byun, Seungjoon Oh Abstract . We study the dissipative spectral form factor (DSFF) at complex time T e i θ Te^{i\theta} for the complex elliptic Ginibre ensemble with non-Hermiticity parameter τ ∈ [ 0 , 1 ) \tau\in[0,1) . As the matrix dimension N → ∞ N\to\infty , we consider the natural scalings in both the time variable and the non-Hermiticity parameter, namely T = O ( N γ ) T=O(N^{\gamma}) and 1 − τ = O ( N − α ) 1-\tau=O(N^{-\alpha}) . For all regimes γ ≥ 0 \gamma\geq 0 and α ≥ 0 \alpha\geq 0 , we derive the precise asymptotic behaviour of both the disconnected and connected components of the DSFF. In particular, we explicitly characterise the dip–ramp–plateau structure, including the dip time and the Heisenberg time. In addition, we identify the mesoscopic regime α ∈ ( 0 , 1 ) \alpha\in(0,1) , which interpolates between the behaviour of the DSFF of non-Hermitian random matrices and the spectral form factor (SFF) of Hermitian ensembles. We further provide an explicit description of the phase diagram, in which the ramp exhibits quadratic, linear, or intermediate behaviour depending on the scaling parameters. 1. Introduction The spectral statistics of quantum systems have long been shaped by two fundamental conjectures. The Berry–Tabor conjecture asserts that generic quantum integrable systems exhibit locally Poissonian level statistics, reflecting the absence of significant correlations between neighbouring energy levels. By contrast, the Bohigas–Giannoni–Schmit conjecture predicts that quantum systems whose classical counterparts are chaotic display local level statistics governed by Hermitian random matrix theory (RMT), in accordance with the relevant symmetry class. This paradigm of quantum chaos has been extended to open and dissipative quantum systems, where the spectrum becomes complex. In this setting, generic integrable systems are expected to exhibit two-dimensional (2D) Poisson statistics, whereas open, dissipative quantum chaotic systems display correlated spectra described by non-Hermitian RMT [ 28 ] . In this work, we focus on the latter regime. There are different tools available from RMT to quantify such statistics, on different scales. On a local or microscopic scale, the nearest-neighbour spacing distribution (in radial distance in 2D) is amongst the most popular, due to the Wigner surmise for RMT in 1D. Much more recently, the consecutive spacing ratio has been introduced [ 37 , 4 ] in 1D and in [ 40 ] in 2D, which is particularly convenient in comparison to data. Here, the dependence on the local scale, given by the inverse mean spectral density, drops out without so-called unfolding [ 29 ] . Nevertheless, little rigorous knowledge is available for this quantity so far, see however [ 36 , 7 ] in 1D, even if a surmise is also effective in the Hermitian case in 1D [ 4 ] . This is despite the fact that, in principle, all local quantities should be accessible from knowledge of the local k -point functions. At least in the classical Hermitian ensembles of RMT the latter follow from standard asymptotics of classical orthogonal polynomials and their corresponding kernel, being given by determinantal or Pfaffian point processes [ 15 ] . An entirely different tool that probes spectral correlations on all scales—microscopic, mesoscopic, and macroscopic—is the spectral form factor (SFF) [ 30 ] , defined as the Fourier transform of the two-point function. While spectra are typically not universal on macroscopic scales, quantum chaotic systems nonetheless exhibit the characteristic dip–ramp–plateau profile predicted by Hermitian RMT; see [ 29 ] and references therein. The literature on the SFF is extensive, and we refrain from attempting a comprehensive survey. Perhaps surprisingly, the mathematically rigorous literature on the SFF remains limited; see, for instance, [ 16 , 17 , 18 , 10 ] and references therein, with early contributions going back to [ 6 ] . This is largely due to the difficulty of obtaining rigorous asymptotics, as multiple scales arise and must be matched. To complete the picture, one may also consider other tools probing multiple scales, such as the number variance and related quantities [ 35 , Chapter 16] , as well as linear statistics more generally. In this work, we focus on the recently introduced dissipative spectral form factor (DSFF) [ 31 ] , which has been found to be applicable to a large variety of open, dissipative quantum chaotic systems [ 32 ] . The DSFF is defined as a two-variable Fourier transform of the two-point function of complex eigenvalues in the appropriate non-Hermitian symmetry class. It also displays a dip–ramp–plateau behaviour; however, the ramp is quadratic instead. The universality of the results in [ 31 ] has been established for real and complex non-Hermitian random matrices with i.i.d. entries [ 11 ] . Note that different proposals have also been made, e.g. the dissipative form factor [ 9 ] , a one-parameter Fourier transform of the two-point function of the Liouvillian (rather than the Hamiltonian), and a fidelity-type deformation [ 34 ] . Given that the study of open dissipative systems in comparison with non-Hermitian RMT has become very popular, see [ 8 , Section 6.2] for a recent review of quantum chaos from that perspective, our goal is to derive mathematically rigorous asymptotics of the DSFF in 2D, and its interpolation to the SFF in 1D proposed in [ 41 ] . For that purpose, we consider the simplest ensemble without further symmetries, introduced by Ginibre [ 26 ] , with complex Gaussian matrix entries (GinUE). Its one-parameter-dependent extension by Girko [ 27 ] and Sommers et al. [ 42 ] , namely the elliptic (or Ginibre–Girko) ensemble, denoted by eGinUE, furthermore allows one to interpolate between the strongly non-Hermitian Ginibre case and the Hermitian limit given by the Gaussian Unitary Ensemble (GUE), thereby recovering the SFF in 1D. Of particular importance here is the weak non-Hermiticity regime introduced by Fyodorov, Khoruzhenko, and Sommers [ 22 , 20 ] . Previous heuristic results on this interpolation between DSFF and SFF exist in [ 41 ] . Apart from making these results mathematically rigorous, we also uncover several different scaling regimes for the dip, ramp, and plateau regions, in analogy with recently discovered weak non-Hermiticity regimes for the number variance in the eGinUE [ 2 ] . 1.1. Elliptic Ginibre ensemble and DSFF Let us now describe more precisely the model under consideration. We consider the eGinUE (see [ 8 , Section 2.3] and references therein) defined by (1.1) X:=\frac{\sqrt{1+\tau}}{2}(G+G^{*})+\frac{\sqrt{1-\tau}}{2}(G-G^{*}),\qquad\tau\in[0,1), where G is a complex Ginibre matrix, i.e. an N\times N matrix with i.i.d. complex Gaussian entries of mean zero and variance 1/N . Let \{z_{j}\}_{j=1}^{N} denote the (complex) eigenvalues of X . It is well known that, as N\to\infty , the empirical spectral measure \frac{1}{N}\sum_{j=1}^{N}\delta_{z_{j}} converges in probability to the elliptic law (1.2) \frac{1}{1-\tau^{2}}\mathbbm{1}_{S}(z)\,dA(z),\qquad S:=\Big\{(x,y)\in\mathbb{R}^{2}:\Big(\frac{x}{1+\tau}\Big)^{2}+\Big(\frac{y}{1-\tau}\Big)^{2}\leq 1\Big\}. Here, dA(z)=d^{2}z/\pi denotes the normalised area measure. By definition, the k -point correlation function R_{N,k} is characterised as follows: for a suitable test function f:\mathbb{C}^{k}\to\mathbb{C} , (1.3) \Big\langle\sum_{\begin{subarray}{c}1\leq i_{1},\dots,i_{k}\leq N\\ \text{distinct}\end{subarray}}f(z_{i_{1}},\dots,z_{i_{k}})\Big\rangle=\int_{\mathbb{C}^{k}}f(z_{1},\dots,z_{k})\,R_{N,k}(z_{1},\dots,z_{k})\,dA(z_{1})\cdots dA(z_{k}), where \langle\cdot\rangle denotes the ensemble average. For the eGinUE, the eigenvalues form a determinantal point process; see again [ 8 , Section 2.3]