The purpose of this course is to introduce the recent progresses in stochastic computational method for physics and machine learning. It is designed for the postgraduates with background in either classical (PDE, optimization) numerical method or with statistical learning for data analysis. The focus is the efficient and fast numerical methods in solving the stochastic model (as a forward model) and the statistical model (as an inverse problem). The contents include: advanced Monte Carlo method, stochastic simulation of jump and diffusion process, Bayesian computations, stochastic optimization for large scale problem and big data, etc. This course emphasizes the algorithms and the practical applications to standard models. The basic knowledge of stochastic process and numerical analysis is prerequisite. The prior knowledge of statistical physics or machine learning is a plus. The experience of computer programming related to numerical methods or scientific computing is required.

Continuous Assessment: 70%

Examination: 30%

Examination Duration: 2 hours