Orbital maximin equals size of set for abelian groups
Suppose is a natural number greater than . Then, for any abelian group acting on a set of size , the maximum possible value of the minimum of the sizes of the orbitals is . Further, there exists an abelian group (namely, the cyclic group of order ) for which this maximum is achieved.
We assume that is at least . The result holds trivially for .
Reducing to the case of a transitive action
Suppose the action of on a set of size has more than one orbit. Let be the size of the smallest orbit, which we call . restricts to a permutation action on . In particular, the orbit of any orbital with both points in lies completely inside the orbital set for .
Thus, if we prove the result for all for a transitive action, then using would complete the proof. We can thus assume that the action is transitive.
Every element must act freely (semiregularly), i.e., without fixed points
We assume acts transitively on a set of size .
Suppose we have an element in and elements such that and . Since acts transitively, there exists such that . Then, whereas . Thus, , contradicting abelianness.
Note that we can in fact deduce that the action must be a regular group action since it is both transitive and semiregular. This is part of a more general idea: any action that centralizes a transitive group action must be semiregular.
The size of an orbital is exactly for a transitive action
As before, acts transitively on a set of size .
Consider any pair . We know that for any two elements , if , then . Since the action is free, this forces , also forcing . Thus, no two distinct elements of the orbital of can have the same first coordinate. Since the number of first coordinates is , we obtain that the size of the orbital is exactly .