On estimating operator norm distance, with optimal trace distance estimation when one state is pure
Abstract
We investigate the computational complexity of estimating the operator norm distance , defined via the operator norm , given -size state-preparation circuits of -qubit quantum states and . We provide efficient quantum estimators for the operator norm distance whose complexity is independent of the rank (and thus the dimension) of the states: 1. When one state is pure, we establish an optimal quantum estimator using queries to the state-preparation circuits. Consequently, for constant additive error, say , our estimator runs in time. Since the operator norm distance is exactly half of the trace distance , our result also gives rank-independent query complexity for estimating both quantities, whereas the approaches due to van Apeldoorn, Cornelissen, Gily{\'{e}}n, and Nannicini (SODA 2023) and Wang and Zhang (TIT 2024) have query complexity scaling at least linearly with , which can be in general. 2. For general quantum states, we also provide a quantum estimator using queries to the state-preparation circuits, which shows that the corresponding promise problem is -complete and improves the upper bound sketched by Liu and Wang (ESA 2025). Together with an quantum query complexity lower bound, this leaves only square-root room for improvement. The key intuition behind our estimators is that, when one state is pure, the pure state has overlap at least with the top unit eigenvector of , reflecting a structural feature specific to the operator norm distance.
Cite
@article{arxiv.2607.03905,
title = {On estimating operator norm distance, with optimal trace distance estimation when one state is pure},
author = {Yupan Liu and Qisheng Wang and Zhan Yu},
journal= {arXiv preprint arXiv:2607.03905},
year = {2026}
}
Comments
28 pages, 2 algorithms. To appear in ESA 2026