arXiv:2609.30268
Quantum Channel Stein Theorem beyond Definite Causal Order
Chengkai Zhu, Xin Wang
Abstract
Distinguishability is a central concept in information theory and extends naturally to quantum systems. This work shows that, for any two finite-dimensional memoryless quantum channels, parallel, adaptive and general testing strategies achieve the same Stein exponent at every fixed type-I error tolerance $ε\in(0,1)$, namely the regularized channel relative entropy. When this rate is finite, any larger type-II exponent forces the probability of correctly accepting the null hypothesis to decay exponentially. Thus, adaptivity provides no asymptotic advantage. The key step is proving the continuity of the regularized sandwiched Rényi channel divergence at order one. We also derive the exact strong-converse exponent for general testers.
AI Summary
The paper asks whether adaptivity or indefinite causal order improves asymmetric discrimination of finite-dimensional memoryless quantum channels. It considers parallel probes, adaptive strategies with quantum memory, and positive process testers normalized on arbitrary products of local channels. For every channel pair and fixed type-I error tolerance $\varepsilon\in(0,1)$, all three classes have Stein exponent $D^\infty(\mathcal N\Vert\mathcal M)$, the regularized reference-assisted channel relative entropy, including when it is infinite. At finite rate, requiring a type-II exponent $r>D^\infty$ makes correct null acceptance decay exponentially. Under Choi support inclusion, the exact strong-converse exponent is $\sup_{p>1}\frac{p-1}{p}[r-\widetilde D_p^\infty(\mathcal N\Vert\mathcal M)]$ for all three classes; parallel achievability comes from an existing result. The converse combines two ingredients. An SDP expresses the optimal parallel weighted testing score as a normalized positive Choi slack. A fixed-marginal de Finetti reduction transfers this bound to general testers with polynomial loss $g_n=\binom{n+d_A^2d_B^2-1}{n}$. Separately, a common amplitude approximation, tensor moment bound valid for entangled probes, and three-amplitude decomposition establish $\lim_{p\downarrow1}\widetilde D_p^\infty=D^\infty$. Additional results give exact testing–smoothing dualities and asymptotic equipartition identities for four modified smoothing sets. The conclusions concern asymptotic exponents, not equality of finite-use performance; the explicit parallel conversion divides both acceptance probabilities by $g_n$. Zero type-I error is excluded. The endpoint proof gives no quantitative convergence rate or second-order expansion. General testers form a mathematical class without an asserted circuit realization for every feasible element.
GPT-5.6 sol then generated the initial version of~\Cref{thm:endpoint}.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 20 pages