Register

Abstract: Python is the dominant language in scientific computing thanks to its simplicity and rich ecosystem, but it often lags behind compiled languages in performance. This course explores practical techniques for accelerating Python workflows, including NumPy vectorization and just-in-time compilation with Numba. We then cover parallel computing, from multithreading and multiprocessing to distributed computing with Ray for scaling CPU- and I/O-bound workloads. Finally, we introduce GPU acceleration with Numba CUDA, including array operations, custom kernels, memory management, and performance tuning, culminating in a GPU-based prime factorization project.