A positional game is a sort of tic-tac-toe played on a hypergraph
— a generalized graph where each “edge” may connect any number of vertices
in ,
not just two. The Erdős–Selfridge theorem gives a simple criterion for the
existence of a winning strategy for the first player on a finite hypergraph.
The bound in the Erdős–Selfridge theorem is tight, and numerous
extremalhypergraphs exist that realize the bound. While characterizing all extremal
hypergraphs for the Erdős–Selfridge theorem is still an open problem, we
make progress in the following way. Many extremal hypergraphs can be formed
by labeling the nodes of a rooted binary tree and defining each edge to be
the set of labels on a root-to-leaf path. For a restricted set of labelings, we
characterize when a given labeling will realize an extremal hypergraph for the
Erdős–Selfridge theorem; therefore, we determine when one side or the other
will have a winning strategy for such labelings.
Keywords
positional games, generalized tic-tac-toe, maker-breaker
games