Poster Presentation on Computational Physics
Conference proceedings talk, Proceedings of the Samahang Pisika ng Pilipinas 41st Samahang Pisika ng Pilipinas Physics Conference, Del Carmen, Siargao Island, Philippines
In my poster presentation, I discuss a method in simulating the 2D Ising model using both CPUs and mobile GPUs. By implementing the checkerboard Metropolis algorithm on an Intel i5-11400H CPU and an Nvidia RTX 3050 Laptop GPU, and coding in Python with Jupyter notebooks, we achieved up to 390 times faster performance compared to standard Python and 2.61 times faster than using the Python compiler with Numba. For smaller lattice sizes, the parallel-CPU approach was more efficient than the mobile GPU, but for larger lattice sizes, the mobile GPU offered significantly better speed. We also address the memory challenges for very large lattice sizes and validate my results by comparing them to Onsager’s analytical solution.