arXiv:2610.18954
Optimal Disorder Strength for Approximate Unitary $k$-Designs in Spin Chains
Tian-Gang Zhou, Yi-Neng Zhou, Thierry Giamarchi, Julian Sonner
Abstract
Quantum simulators with disordered spin Hamiltonians offer a natural route to Haar-like unitary ensembles, but face a competition. Disorder enhances randomness, yet strong disorder suppresses dynamics and localizes the system in the computational basis. This suggests an optimal intermediate disorder strength for approximating a unitary $k$-design, which has so far been identified mainly empirically. Here, we develop a perturbative theory for the moment gap $Δ^{(k)}$, the deviation of the $k$-th moment operator from its Haar value, for the full spectrum of the interacting XXZ chain in a fixed-magnetization sector. We find the non-monotonic asymptotic scalings $\ln Δ^{(k)}\sim -σ^2$ at weak disorder and $\ln Δ^{(k)}\sim -1/σ$ at strong disorder, whose crossover sets an optimal disorder strength $σ_c\sim J_{xy}$, where $J_{xy}$ is the energy scale of the non-random interaction. We further show that a multi-quench protocol further suppresses the gap exponentially in the number of quenches. Numerical simulations confirm both scalings and this exponential suppression, and we demonstrate the usefulness of the protocol for classical shadow tomography. This work suggests that strong disorder may block both thermalization and the formation of unitary designs through a common localization mechanism.
AI Summary
The paper asks how disorder strength affects the generation of approximate unitary $k$-designs by analog spin-chain dynamics, and whether repeated disorder quenches improve convergence. It studies an XXZ chain with independent Gaussian longitudinal fields of standard deviation $\sigma$, restricted to the zero-magnetization sector. Randomness is quantified by $\Delta^{(k)}=\|M_\nu^{(k)}-M_{\mathrm{Haar}}^{(k)}\|_\infty$. Replica-space perturbation theory gives leading logarithmic corrections $\ln\Delta^{(k)}\sim-\sigma^2$ at weak disorder and $\ln\Delta^{(k)}\sim-1/\sigma$ at strong disorder. These tails indicate an intermediate optimum of order $J_{xy}$; simulations for $L=6$, $J_{xy}=1$, and $J_z=0.6$ locate it near $\sigma_c=1.76$. Independent disorder realizations across $q+1$ evolution segments satisfy the exact bound $\Delta_q^{(k)}\leq\prod_a\Delta^{(k)}(\Delta t_a)$, giving exponential suppression for equal-duration segments with gap below one. A numerical shadow-tomography demonstration on ten physical spins reduces biased fidelity estimates for encoded cluster and GHZ states toward unity after about four quenches. The perturbative expansions do not control the intermediate-disorder optimum. The strong-disorder expression describes initial decay but not the nonzero late-time plateau of single-quench evolution; its disorder scaling at the plateau is supported numerically. Most gap calculations use the first moment on six spins, while direct second-moment comparisons use four spins. At those sizes, interacting first- and second-moment gaps coincide after a transient, whereas the free-fermion case has $\Delta^{(k)}=1$ for $k\geq2$. Joint optimization of switching times, higher-moment frame potentials, and shadow sampling costs remain open.
Central claim may not hold: The reported optimum σ_c≈1.76 is numerical for a particular finite-chain protocol, while the perturbative tails provide only a qualitative crossover argument where neither expansion is controlled. For k≥2, the strong-disorder derivation omits nonzero Ising evolution inside the slow subspace and incorrectly applies the Gaussian hopping kernel to zero-imbalance intermediate states, leaving its advertised arbitrary-k gap formula unsupported. The perturbative disorder scalings are established only for small corrections, with their extension to late-time plateaus supported empirically and explicit second-moment verification restricted to L=4. The supplementary exact multi-quench equality rests on an invalid Hermiticity/positivity inference and neglects intervening clean evolution, although the main-text norm inequality is valid.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- main text: 5 pages, 3 figures. Supplementary material 9 pages, 3 figures