A new method for the partition function of discrete systems with application to the 3D Ising model

Abstract We describe a new method to find numerically the density of states of discrete statistical systems. We apply the method to the zero-field, three-dimensional Ising model on a 5 3 lattice. Our method yields an excellent approximation to the partition function and can accurately predict its zeros near the critical points. We argue that our method can be used very effectively even on much larger systems.