Refuting spectral compatibility of quantum marginals
Felix Huber, Nikolai Wyderka
TL;DR
This work introduces a symmetry-reduced semidefinite programming hierarchy that completely certifies spectral incompatibility among overlapping quantum marginals, with certificates that are valid in all local dimensions when the copy number does not exceed the local dimension ($k\le d$). The approach leverages permutation symmetry, Schur-Weyl duality, and a quantum de Finetti-type argument to guarantee completeness and provide practical scalability (up to modest system sizes). It yields dimension-free incompatibility witnesses, along with concrete numerical results for tripartite and higher-partite systems, and derives entropy- and purity-type inequalities from the dual. Beyond the spectral marginal problem, the framework applies to sums of Hermitian matrices, local unitary invariants, and equivariant state polynomials, suggesting broad applicability to invariant theory and quantum information tasks.
Abstract
The spectral variant of the quantum marginal problem asks: Given prescribed spectra for a set of overlapping quantum marginals, does there exist a compatible joint state? The main idea of this work is a symmetry-reduced semidefinite programming hierarchy that detects when no such joint state exists. The hierarchy is complete, in the sense that it detects every incompatible set of spectra. The refutations it provides are dimension-free, certifying incompatibility in all local dimensions. The hierarchy also applies to the sums of Hermitian matrices problem, the compatibility of local unitary invariants, for certifying vanishing Kronecker coefficients, and to optimize over equivariant state polynomials.
