Optimal estimation of pure states with displaced-null measurements

TL;DR

This paper introduces a displaced-null measurement strategy for estimating pure quantum states, overcoming non-identifiability issues, and achieves asymptotic optimality in multi-parameter quantum estimation.

Contribution

It proposes a novel displaced-null measurement method that ensures parameter identifiability and attains the quantum Cramér-Rao and Holevo bounds asymptotically.

Findings

Displaced-null measurements achieve asymptotic optimality.

Naive null-measurements fail to attain standard estimation scaling.

The method extends to multi-parameter models with asymptotic optimality.

Abstract

We revisit the problem of estimating an unknown parameter of a pure quantum state, and investigate `null-measurement' strategies in which the experimenter aims to measure in a basis that contains a vector close to the true system state. Such strategies are known to approach the quantum Fisher information for models where the quantum Cram\'{e}r-Rao bound is achievable but a detailed adaptive strategy for achieving the bound in the multi-copy setting has been lacking. We first show that the following naive null-measurement implementation fails to attain even the standard estimation scaling: estimate the parameter on a small sub-sample, and apply the null-measurement corresponding to the estimated value on the rest of the systems. This is due to non-identifiability issues specific to null-measurements, which arise when the true and reference parameters are close to each other. To avoid this, we propose the alternative displaced-null measurement strategy in which the reference parameter is altered by a small amount which is sufficient to ensure parameter identifiability. We use this strategy to devise asymptotically optimal measurements for models where the quantum Cram\'{e}r-Rao bound is achievable. More generally, we extend the method to arbitrary multi-parameter models and prove the asymptotic achievability of the the Holevo bound. An important tool in our analysis is the theory of quantum local asymptotic normality which provides a clear intuition about the design of the proposed estimators, and shows that they have asymptotically normal distributions.