WorksPhysics and Astronomy
Efficient numerical measurement design and certified risk bounds for multiparameter quantum estimation
Abstract
Multiparameter quantum estimation requires efficient measurement design and reliable bounds on the attainable precision. We present Q-StARC, a numerical framework that combines measurement optimization with certified lower bounds on the weighted variance of locally unbiased estimators. The measurement search parameterizes rank-one measurements on a complex Stiefel manifold and combines projected ambient Adam updates, QR retraction, and Riemannian conjugate-gradient refinement. Independent risk bounds are obtained from a denominator-based matrix sum-of-squares hierarchy. When its support coefficients are optimized jointly, the hierarchy starts at the weighted Nagaoka-Hayashi (NH) bound and converges to the optimal locally unbiased risk, including for singular states. Numerical applications certify that the NH bound is unattainable for several finite-copy qubit models and two full eight-parameter qutrit models. For the qutrit examples, the certified relative gaps above the NH bound exceed and , while the level- lower bounds agree with the best numerical measurement costs to relative differences of and , respectively.