Relatedly, does anyone believe that MIP* could be undecidable?

]]>If the commuting operator value of a game is equal to the (“standard”) tensor-product value of a game, then combining the NPA algorithm to give an upper bound on the quantum value, plus the exhaustive search algorithm to give a lower bound, will converge to the true value of the game.

However we don’t know if the commuting operator value of games are always equal to the tensor product value, so it is not clear that this NPA + exhaustive search combo algorithm always works to approximate the quantum value of a game.

]]>https://en.wikipedia.org/wiki/Kelmans-Seymour_conjecture

The wiki page points to a series of legit-looking Arxiv preprints claiming to have a proof, but they are from 2016 and don’t mention having been submitted anywhere or that refereeing is in progress or anything like that. So of course, this comment section is the logical place to ask ;-).

]]>