Sufficient criteria for absolute separability in arbitrary dimensions via linear map inverses
Jofre Abellanet-Vidal, Guillem Müller-Rigat, Grzegorz Rajchel-Mieldzioć, Anna Sanpera
TL;DR
The paper tackles the problem of characterizing absolute separability (AS) and absolute PPT (AP) across arbitrary dimensions by leveraging inverses of non-CP maps. It develops eigenvalue-based sufficient criteria derived from unitary-covariant, reduction-like maps Λ_α and their inverses, and then unifies these criteria using convex geometry to yield stronger, spectrum-only tests expressible as convex programs. The main contributions include a compact inequality describing the convex hull of AS criteria, an SDP-formulation to combine multiple criteria, and robust extensions to multipartite and symmetric subspaces with explicit eigenvalue bounds for SAS and SAP. These results provide practically verifiable, dimension-agnostic tests that rely only on spectral data, with potential applications in resource theories and invariant-set analyses. The framework also suggests pathways to extend similar convex-geometry methods to other invariant convex sets in quantum information, such as steering and nonlocality criteria.
Abstract
Quantum states that remain separable (i.e., not entangled) under any global unitary transformation are known as absolutely separable and form a convex set. Despite extensive efforts, the complete characterization of this set remains largely unknown. In this work, we employ linear maps and their inverses to derive new sufficient analytical conditions for absolute separability in arbitrary dimensions, providing extremal points of this set and improving its characterization. Additionally, we employ convex geometry optimization to refine the characterization of the set when multiple non-comparable criteria for absolute separability are available. We also address the closely related problem of characterizing the absolute PPT (positive partial transposition) set, which consists of quantum states that remain positive under partial transposition across all unitary transformations. Finally, we extend our results to multipartite states.
