arXiv:2611.00931
Binary Quantum Hypothesis Testing with a Rejection Option: The Exact Error-Exponent Trade-off
Yonglong Li
Abstract
We characterize the exact error-exponent trade-off in binary quantum hypothesis testing with a rejection option for faithful states on a finite-dimensional Hilbert space. The probabilities of rejection under the two hypotheses are required to vanish exponentially, with exponents at least $K$ and $L$. We show that a quadruple $(a,b,K,L)$ of positive error and rejection exponents is achievable if and only if it satisfies three single-letter conditions: two conditions involve Petz--Rényi quantities and coincide with the converse bound conjectured to be tight by Ji and Regula; the third involves the $α$--$z$ Rényi quantities on the triangle $\max\{α,1-α\}\le z\le1$. The proof combines a sector decomposition inside Schur--Weyl blocks, an explicit one-shot measurement construction, and an exact minimax argument.
AI Summary
The paper determines the joint trade-off between two error exponents and two rejection exponents in binary i.i.d. quantum hypothesis testing. For distinct faithful states $\rho,\sigma$ in finite dimension, a three-outcome collective measurement may accept either hypothesis or reject both; rejection probabilities must decay with exponents at least $K,L>0$. A positive quadruple $(a,b,K,L)$ is achievable exactly when three single-letter conditions hold. Two are the previously known Petz–Rényi converse: $(\min\{a,K\},b)$ and $(a,\min\{b,L\})$ must lie in the binary quantum Hoeffding region. The additional condition is that, for some $\max\{\alpha,1-\alpha\}\le z\le1$, $(\alpha+z-1)a+(z-\alpha)b+(1-z)(K+L)\le-\log Q_{\alpha,z}(\rho\|\sigma)$. This condition is automatic when $K+L\le-2\log F(\rho,\sigma)$, with $F$ the root fidelity. Above this threshold, the Petz converse can be non-tight. A qubit example exhibits an interior $\alpha$–$z$ optimizer and a discontinuity in the optimal equal-error exponent. The proof combines Schur–Weyl blocks, sector exponents tracking spectral weights and eigenspace overlaps, an exact minimax argument, and an explicit one-shot measurement construction. Achievability bounds all four probabilities by their target exponentials times $800(n+1)^{3(d^2-1)}$. The result assumes faithful, finite-dimensional states and positive exponents. For rejection budgets imposed exactly at every blocklength, the same region is established except on $K+L=-2\log F$, where the general noncommuting case remains unresolved. The bounds are not second-order asymptotics; extensions to multiple, composite, or correlated hypotheses remain open.
The author used ChatGPT 5.6 Sol (OpenAI) to assist with mathematical derivations and proof checking, numerical calculations, language editing, and LaTeX preparation.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.