The notebook can simulate Ising model on a chain/square/cube/N-d hypercubic lattice. The Ising model consists of +/-1 at each lattice site. A clean model at a finite temperature can be simulated using Metropolis-Hashtings algorithm. The update can be parallelized using a checkerboard lattice, since it has nearest-neighbor interactions only. As a result, it can be simulated on a GPU very efficiently. Tensorflow library is used to demonstrate this on a square lattice Ising model with uniform Ferromagnetic interaction and a uniform field at a non-zero temperature.
- tensorflow 2.x
- numpy
- matplotlib
- ipywidgets