Infinite Hat Problems and Large Cardinals

TL;DR

This paper investigates infinite hat guessing problems involving large cardinals, characterizing when logicians can guarantee a correct guess based on the size of the set of hat colors and their observational constraints.

Contribution

It generalizes classical hat problems to infinite settings, linking the existence of winning strategies to large cardinal properties and set-theoretic principles.

Findings

Winning strategies depend on the cardinality of colors and observational restrictions.

Connections established between hat problems and large cardinal axioms.

Some cases are undecidable within ZFC, requiring additional set-theoretic assumptions.

Abstract

Picture countably many logicians all wearing a hat in one of $\kappa$-many colours. They each get to look at finitely many other hats and afterwards make finitely many guesses for their own hat's colour. For which $\kappa$ can the logicians guarantee that at least one of them guesses correctly? This will be the archetypical hat problem we analyse and solve here. We generalise this by varying the amount of logicians as well as the number of allowed guesses and describe exactly for which combinations the logicians have a winning strategy. We also solve these hat problems under the additional restriction that their vision is restrained in terms of a partial order. Picture e.g.~countably many logicians standing on the real number line and each logician is only allowed to look at finitely many others in front of them. In many cases, the least $\kappa$ for which the logicians start losing can be described by an instance of the free subset property which in turn is connected to large cardinals. In particular, $\mathrm{ZFC}$ can sometimes not decide whether or not the logicians can win for every possible set of colours.