Generalized Gleason theorem and finite amount of information for the context

• Published in: arXiv.org
• Main Authors: Montina, A, Wolf, S
• Format: Paper Journal Article
• Language: English
• Published: Ithaca Cornell University Library, arXiv.org 23.09.2022
• Subjects: Context; Evolution; Linear functions; Physics - Quantum Physics; Projectors; Quantum theory; Theorems
• ISSN: 2331-8422
• DOI: 10.48550/arxiv.2206.11830

Quantum processes cannot be reduced, in a nontrivial way, to classical processes without specifying the context in the description of a measurement procedure. This requirement is implied by the Kochen-Specker theorem in the outcome-deterministic case and, more generally, by the Gleason theorem. The latter establishes that there is only one non-contextual classical model compatible with quantum theory, the one that trivially identifies the quantum state with the classical state. However, this model requires a breaking of the unitary evolution to account for macroscopic realism. Thus, a causal classical model compatible with the unitary evolution of the quantum state is necessarily contextual at some extent. Inspired by well-known results in quantum communication complexity, we consider a particular class of hidden variable theories by assuming that the amount of information about the measurement context is finite. Aiming at establishing some general features of these theories, we first present a generalized version of the Gleason theorem and provide a simple proof of it. Assuming that Gleason's hypotheses hold only locally for `small' changes of the measurement procedure, we obtain almost the same conclusion of the original theorem about the functional form of the probability measure. An additional constant and a relaxed property of the `density operator' are the only two differences from the original result. By this generalization of the Gleason theorem and the assumption of finite information for the context, we prove that the probabilities over three or more outcomes of a projective measurement must be linear functions of the projectors associated with the outcomes, given the information on the context.