arXiv:2610.01209
Fermionic Gaussian Scrooge Ensembles in Deep Thermalization
Ning Sun, Pengfei Zhang
Abstract
Measuring part of a many-body wave function generates a projected ensemble of pure quantum states on the unmeasured subsystem. Recent advances in deep thermalization have shown that, in chaotic systems, this ensemble universally converges to the maximally random ensemble compatible with its average density matrix, known as the Scrooge ensemble. By contrast, free-fermion systems are nonchaotic, and their quantum states are constrained to remain Gaussian. Motivated by this distinction, we introduce the fermionic Gaussian Scrooge ensemble to describe deep thermalization in generic free-fermion systems. This ensemble is defined as a distortion of the Gaussian Haar ensemble by the average density matrix, and its moments reveal an enlarged symmetry of Gaussian states in the replicated Hilbert space. We demonstrate the emergence of the fermionic Gaussian Scrooge ensemble in two complementary settings: (1) we analytically prove that the projected ensemble of the SYK$_2$ model follows the fermionic Gaussian Scrooge ensemble at arbitrary evolution times; (2) we provide numerical evidence that it emerges at sufficiently long times in random Gaussian circuits with charge conservation. Our results establish the fermionic Gaussian Scrooge ensemble as a universal description of deep thermalization in free-fermion systems.
AI Summary
The paper asks what universal ensemble describes states obtained by measuring most of a free-fermion system, where Gaussianity prevents convergence to the conventional Scrooge ensemble. It defines a fermionic Gaussian Scrooge (fGS) ensemble by applying $\sqrt{\rho}$ to Gaussian Haar-random states and reweighting their probabilities, separately in each fermion-parity sector. Its average state is $\rho$. Higher moments involve continuous $SO(n)$ replica rotations rather than the replica permutations appearing for conventional Scrooge ensembles. Two settings support this description. For the quadratic SYK model initialized in a thermofield-double state, a disorder-averaged, large-$N$ replica saddle-point calculation identifies the projected-ensemble moments with fGS moments at arbitrary evolution times, with subsystem size $N_A=O(1)$. The identification follows from a continuous family of replica-rotated saddle points. For one-dimensional, charge-conserving random Gaussian circuits, correlation-matrix simulations show convergence toward fGS moments at long times. These use $N_A=4$, $N_B=300$, and $10^5$ sampled measurement outcomes, testing four initial-state parameter combinations in a single initial-state and circuit realization. The analytic result concerns the specified thermofield-double preparation and measurement protocol, rather than arbitrary free-fermion dynamics. The circuit evidence uses initial states with extensive charge fluctuations; these prevent measurements from sharply constraining subsystem charge, so the result does not establish the same target for fixed-charge preparations. Extensions to nonunitary dynamics remain open. The paper also argues that weak interactions throughout the system gap the replica-rotation modes, leaving the conventional Scrooge structure.
Central claim may not hold: The SYK₂ argument establishes, at most, an asymptotic disorder-averaged result for the specified thermofield-double setting, rather than finite-system or individual-realization equality. The supplied SYK₂ derivation does not establish that its symmetry-generated saddle family exhausts the dominant contributions, leaving the central moment-matching argument conditional. The circuit simulations support the proposed ensemble in a restricted regime with extensive initial charge fluctuations, rather than establish universality across free-fermion systems and initial states. The displayed thermofield-double state has an incorrect normalization prefactor: it should contain 2^(N/4), not 2^(N/2), when |EPR⟩ is normalized.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 10 pages, 3 figures + supplementary materials