English

Non-iid hypothesis testing: from classical to quantum

Quantum Physics 2025-10-08 v1 Data Structures and Algorithms Machine Learning

Abstract

We study hypothesis testing (aka state certification) in the non-identically distributed setting. A recent work (Garg et al. 2023) considered the classical case, in which one is given (independent) samples from TT unknown probability distributions p1,,pTp_1, \dots, p_T on [d]={1,2,,d}[d] = \{1, 2, \dots, d\}, and one wishes to accept/reject the hypothesis that their average pavgp_{\mathrm{avg}} equals a known hypothesis distribution qq. Garg et al. showed that if one has just c=2c = 2 samples from each pip_i, and provided Tdϵ2+1ϵ4T \gg \frac{\sqrt{d}}{\epsilon^2} + \frac{1}{\epsilon^4}, one can (whp) distinguish pavg=qp_{\mathrm{avg}} = q from dTV(pavg,q)>ϵd_{\mathrm{TV}}(p_{\mathrm{avg}},q) > \epsilon. This nearly matches the optimal result for the classical iid setting (namely, Tdϵ2T \gg \frac{\sqrt{d}}{\epsilon^2}). Besides optimally improving this result (and generalizing to tolerant testing with more stringent distance measures), we study the analogous problem of hypothesis testing for non-identical quantum states. Here we uncover an unexpected phenomenon: for any dd-dimensional hypothesis state σ\sigma, and given just a single copy (c=1c = 1) of each state ρ1,,ρT\rho_1, \dots, \rho_T, one can distinguish ρavg=σ\rho_{\mathrm{avg}} = \sigma from Dtr(ρavg,σ)>ϵD_{\mathrm{tr}}(\rho_{\mathrm{avg}},\sigma) > \epsilon provided Td/ϵ2T \gg d/\epsilon^2. (Again, we generalize to tolerant testing with more stringent distance measures.) This matches the optimal result for the iid case, which is surprising because doing this with c=1c = 1 is provably impossible in the classical case. We also show that the analogous phenomenon happens for the non-iid extension of identity testing between unknown states. A technical tool we introduce may be of independent interest: an Efron-Stein inequality, and more generally an Efron-Stein decomposition, in the quantum setting.

Cite

@article{arxiv.2510.06147,
  title  = {Non-iid hypothesis testing: from classical to quantum},
  author = {Giacomo De Palma and Marco Fanizza and Connor Mowry and Ryan O'Donnell},
  journal= {arXiv preprint arXiv:2510.06147},
  year   = {2025}
}

Comments

33 pages, 2 figures

R2 v1 2026-07-01T06:21:57.625Z