Local Friendliness Polytopes In Multipartite Scenarios

Recently the Local Friendliness (LF) no-go theorem has gained a lot of attention, owing to its deep foundational implications. This no-go theorem applies to scenarios which combine Bell experiments with Wigner's friend-type set ups, containing space-like separated superobservers who are assumed to be capable of performing quantum operations on a local observer, also known as their "friend". Analogously to the hypothesis of local hidden variables in Bell scenarios, a set of assumptions termed "Local Friendliness" constrains the space of probabilistic behaviours accessible to the superobservers to be a particular subset of the no-signalling polytope in such scenarios. It has additionally been shown, that there are scenarios where the set of behaviours compatible with Local Friendliness is strictly larger than the Bell-local polytope, while in some scenarios those sets are equal. In this work, we complete the picture by identifying all the canonical Local Friendliness scenarios, with arbitrary but finite numbers of superobservers, friends, measurements and outcomes, where the set of LF correlations admits a local hidden variable model, and where they do not. Our proof is constructive in the sense that we also demonstrate how a local hidden variable model can be constructed, given a behaviour compatible with LF in the appropriate scenarios. While our principal motivation is the foundational question of better understanding the constraints from Local Friendliness, the same inequalities constraining LF polytopes have been shown to arise in a priori unrelated contexts of device-independent information processing. Our results may thus find use in those research areas as well.