Robustness of quantum spectrum estimation: weak Schur sampling under noisy inputs

2026-10-02

AI Summary

This study addresses the failure of weak Schur sampling in quantum spectrum estimation caused by noise-induced breaking of permutation symmetry. For the first time, we establish the noise robustness of the Keyl–Werner algorithm. By integrating trace distance, quantum Wasserstein distance, and Lipschitz continuity theory, we rigorously prove that the sampling error depends solely on single-copy noise and does not accumulate with increasing sample size. Furthermore, we derive a total variation error upper bound of ε + O(d/√n), formally characterizing the algorithm's resilience to noise. These results validate the practical feasibility of optimal quantum learning under realistic noisy conditions.

Abstract

We initiate the study of weak Schur sampling (WSS) under noisy input. Many optimal quantum learning algorithms rely on this measurement, for instance in such fundamental tasks as quantum spectrum estimation (QSE) and quantum state tomography (QST). The standard analysis of WSS, QSE and QST assumes that the $n$ input copies of the target state $ρ$ are identical. We study what happens when they are not, a more realistic noisy scenario that breaks the very permutation symmetry on which WSS is built. In the simplest noise model, the copies are independent but not necessarily identical, i.e. they form a product state, and each is $ε$-close in trace distance to the $d$-dimensional target state $ρ$. A naive data-processing argument lets the per-copy errors accumulate to $nε$ on measurement output, growing with the input size. We prove that they do not accumulate: the expected output error in total variation is at most $ε+η(n,d)$, where $η(n,d)=O(d/\sqrt n)$ is the noiseless error, and the additive noise term $ε$ is optimal. Beyond noisy product states, for arbitrary inputs $ω$ we show that the outcome observables of WSS are $1/n$-Lipschitz with respect to the quantum Wasserstein distance $W_1$ of De Palma et al., so the error is at most $η(n,d)+\frac{1}{n}\|ω-ρ^{\otimes n}\|_{W_1}$. In particular, our results imply, for the first time, noise-robustness of Keyl and Werner's seminal algorithm for QSE, more than two decades after its introduction. Within our model of robustness, the promise cannot be weakened to closeness of single-copy marginals or to a global trace-distance budget alone, since in both cases some noisy inputs defeat every measurement and estimator. Our results are a necessary step towards optimal quantum learning in practice, and help make progress towards robustness of many other problems in quantum information theory.

Problem

• weak Schur sampling
• quantum spectrum estimation
• noisy inputs
• robustness
• quantum state tomography

Innovation

• Weak Schur sampling
• Quantum spectrum estimation
• Noise robustness
• Quantum Wasserstein distance
• Lipschitz continuity