Hypothesis testing on invariant subspaces of the symmetric group: part I. Quantum Sanov's theorem and arbitrarily varying sources

Abstract

We report a proof of the quantum Sanov theorem by the elementary application of basic facts about representations of the symmetric group, together with a complete characterization of the optimal error exponent in a situation where the null hypothesis is given by an arbitrarily varying quantum source instead. Our approach differs from previous ones in two ways. First, it supports a reasoning inspired by the ‘method of types’. Second, the measurement scheme we propose to distinguish the two alternatives not only does that job asymptotically perfectly, but also yields additional information about the null hypothesis. An example of that is given. The measurement is composed of projections onto permutation-invariant subspaces, thus providing a direct link between one of the most basic tasks in quantum information on the one hand and fundamental objects in representation theory on the other. We additionally connect to the representation theory by proving a relation between Kostka numbers and quantum states, and to state estimation via a generalization of a well-known spectral estimation theorem to non-i.i.d. sequences.