An NP-hard generalization of Nim
Chunlei Liu
TL;DR
The paper introduces occupation games $(X,S,O)$ as a generalization of Nim and Subtraction and defines the Nash-equilibrium indicator ${\rm Truth}(A)\in\{0,1\}$. It analyzes the complexity of computing ${\rm Truth}(A)$, establishing NP-hardness via a subset-sum based reduction. The core contribution is a constructive reduction from Subset Sum that encodes the existence of a subset summing to $t$ into the truth value ${\rm Truth}(X)$. This work delineates the computational limits of equilibrium computation in broad classes of impartial games and underscores the intractability of generalizations of Nim.
Abstract
A new combinatorial game is given. It generalizes both Substraction and Nim. It is proved the computation of Nash equilibrium points in this new game is NP-hard.
