Statistics of the Random Matrix Spectral Form Factor
Alex Altland, Francisco Divi, Tobias Micklitz, Silvia Pappalardi, Maedeh Rezaei · 2025
Read Statistics of the Random Matrix Spectral Form Factor by Alex Altland, Francisco Divi, Tobias Micklitz, Silvia Pappalardi, Maedeh Rezaei (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-2503-21386.txt.
Introduction
Alex Altland and four co-authors revisit a key statistic of random matrix theory, a tool for describing disordered and chaotic quantum systems. Their paper came out in 2025. The statistic, the spectral form factor, tracks how energy levels of a system repel and correlate over time.
Its average behavior is roughly bell-shaped, but small corrections matter for modern applications, and earlier studies disagreed about them. The authors attack the problem with two independent methods: one based on kernel functions, one on a supersymmetric field theory. Both give the same next-order corrections, and both match computer simulations. Their kernel analysis also exposes where previous work went wrong. The paper is written as a guide, pointing out new paths and pitfalls for these fine-grained statistical questions.
Catalog record
| CREATOR | Alex Altland, Francisco Divi, Tobias Micklitz, Silvia Pappalardi, Maedeh Rezaei |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 76 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2503-21386
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Statistics of the Random Matrix Spectral Form Factor Alex Altland, Francisco Divi, Tobias Micklitz, Silvia Pappalardi, Maedeh Rezaei Abstract The spectral form factor of random matrix theory plays a key role in the description of disordered and chaotic quantum systems. While its moments are known to be approximately Gaussian, corrections subleading in the matrix dimension, D D , have recently come to attention, with conflicting results in the literature. In this work, we investigate these departures from Gaussianity for both circular and Gaussian ensembles. Using two independent approaches — sine-kernel techniques and supersymmetric field theory — we identify the form factor statistics to next leading order in a D − 1 D^{-1} expansion. Our sine-kernel analysis highlights inconsistencies with previous studies, while the supersymmetric approach backs these findings and suggests an understanding of the statistics from a complementary perspective. Our findings fully agree with numerics. They are presented in a pedagogical way, highlighting pathways (and pitfalls) in the study of statistical signatures at next leading order, which are increasingly becoming important in applications. I Introduction The spectral form factor \displaystyle{\mathrm{SFF}}\equiv|\tr U_{t}|^{2} (1) is one of the most important observables in the characterization of disordered or chaotic quantum systems [ 6 , 42 , 35 , 36 ] . It owes its significance to a combination of factors: The SFF probes the dynamics of the time evolution operator U_{t} , and its Fourier transform energy dependent correlations in spectra, both key signatures of a system. The time dependence of numerous observables, such as correlation functions between operators [ 19 , 45 , 8 ] , out of time order correlations (OTOC) functions [ 17 , 37 ] , or return probabilities [ 43 ] , approximately, or even exactly reduce to the SFF. Finally, the SFF has come to prominence in the context of many-body physics [ 11 , 12 , 30 , 7 , 3 ] and holography [ 26 , 18 , 16 , 41 , 40 , 2 ] , where it features in the effective boundary description of two-dimensional gravitational theories. The spectral form factor is commonly discussed with reference to three temporal regimes: at times t<t_{\textrm{Th}} smaller than an effective Thouless time , it probes physical relaxation processes preceding the entry into an ergodic regime at times t>t_{\textrm{Th}} . (Our focus here will be on systems with late time ergodic behavior.) Once ergodic, it typically increases in a linear or near linear manner — the so called ‘ramp regime’ — up to a time {t_{\textrm{H}}=2\pi/\Delta} , the Heisenberg time defined as the inverse of a system’s characteristic level spacing \Delta . At later times, the form factor approaches constant behavior, the ‘plateau regime’. Our focus in this paper will be on ergodic late time physics, ramp and plateau. Even in the ergodic regime, the spectral form factor remains a statistically distributed quantity, and is subject to three different sources of fluctuations: 1. The SFF of generic chaotic quantum systems typically shows deviations from that of random matrix theory (RMT), even deep in the ergodic regime at times {t\simeq t_{\textrm{H}}} . These deviations are due to large sample-to-sample fluctuations of the spectral density, which in many-body systems generally exceed those of Gaussian matrix ensembles. The deviations from the RMT form factor caused by these fluctuations have become a subject of study in their own right [ 21 , 13 , 44 , 31 , 24 ] . 2. The form factor makes reference to physical time so that, by energy/time uncertainty, it includes an integral over energies (cf. Eqs. ( 2 ) and ( 3 ) below). The non-constancy of the spectral density of a quantum system affects the form factor and its statistical moments via this convolution mechanism [ 10 ] . 3. Even within RMT, the spectral form factor is a famously non self-averaging quantity whose statistical fluctuations are of about the same order as the average, and not diminishing in the large system size limit [ 39 , 9 , 33 ] . In generic interacting quantum systems, these three fluctuation channels act in combination, leading to intricate statistics. Their importance is highlighted by the fact that the quantum systems relevant to present day applications are typically large, D\gg 1 , but not thermodynamically large. Some of the fluctuation channels mentioned above diminish in the large system size limit, D\to\infty , but not all of them do. Finally, statistical data on the form factor is typically collected by sampling over a finite number of runs N_{\textrm{sim}} , either on a computer or in experiments. This opens a fourth fluctuation channel reflecting the statistics of a finite data set. In view of this level of complexity, it is important to understand the statistics of the random matrix SFF as a ‘null-model’ before comparing it to realistic quantum systems. The statistics of the form factor in RMT has been the subject of a number of previous studies [ 27 , 16 , 32 , 23 , 22 , 14 ] , a key finding being that it is approximately Gaussian, which at the second order implies \overline{\mathrm{SFF}^{2}}\simeq 2(\overline{\mathrm{SFF}})^{2} , where \overline{\,\cdot\,} is the ensemble average. Deviations from Gaussianity are subleading in D but otherwise universal signatures of the quantum ergodic phase. They, too, have come under scrutiny, with a plethora of not entirely consistent results: Ref. [ 14 ] rigorously computes \overline{\mathrm{SFF}} and Var (SFF) for Wigner matrices, but only for short times t\ll N^{5/17} , where it is dominated by the edge spectrum. Ref. [ 27 ] agrees with Ref. [ 14 ] when such edge effect is neglected, but it also predicts the emergence of non-Gaussian statistics linearly increasing in time, t , at t=t_{\mathrm{H}}/2 . The sudden (non-analytic) onset of this correction is indicative of a non-perturbative effect, similar to the celebrated ‘ramp-plateau’ transition in the average SFF. Finally, Refs. [ 16 , 32 ] predict a more complicated time dependence containing a linear term throughout the semiclassical regime t<t_{\mathrm{H}}/2 , and contributions of next-leading order in the 1/D -expansion. (We note that for [ 16 ] , this equation is ancillary to the paper’s main results.) In the comparison of these results, it is important to remember that the SFF is subject to energy averaging, the second channel of fluctuations mentioned above: The SFF contains physical time in the dimensionless combination \tau\equiv t\Delta/2\pi , where \Delta\sim D^{-1} is the characteristic level spacing. Being defined in terms of an unconstrained trace {\mathrm{SFF}}=|\textrm{tr}(U_{t})|^{2} the scale \Delta=\Delta(\epsilon) enters averaged over the entire spectrum blurring otherwise sharply defined structures over scales \sim\Delta^{-1} [ 10 ] . This mechanism is absent in circular ensembles , defined in terms of Haar-distributed unitary operators with statistically uniform eigenphase spacing, \Delta=2\pi/D , making them attractive candidates for the study of the more interesting ‘intrinsic’ statistical fluctuations, channel 3. Figure 1: Numerical summary of the paper. The spectral form factor (SFF) of a single realization (red dots) fluctuates over time around its average \overline{\rm SFF} (black line), with a variance {\rm Var(SFF)}=\overline{\rm SFF^{2}}-\overline{\rm SFF}^{2} proportional to the average in the large- D limit (blue line), i.e. Gaussian distribution. The amplitude of these fluctuations is indicated by the shaded red region. The