Coarse-graining using machine learning based methods

My Ph.D. research focuses on coarse-graining using machine learning-based methods and analytical theory for building physical models for multi-scale modeling. There are many reasons for performing simulations on simpler systems. In complex systems with large scales, we are concerned with the effective interactions and driving forces that emerge from the detailed descriptions of the systems. Coarse-grained models enable us to simulate complex systems by focusing on the most essential features and averaging over less important details. The recent rise of machine learning-based computational discovery provides a new philosophy for chemical, physical, and materials science. We recently proposed the use of deep learning methods and Gaussian process to build a bridge between the microscopic and macroscopic worlds for the identification of unavailable coarse-scale equations. We use data from molecular dynamics simulations (lattice-Boltzmann simulations) to approximate an unavailable macroscopic physical equation. My collaboration with Professor George Em Karniadakis, who is an expert in machine learning methods, would provide me with an excellent opportunity to learn the details of the critical techniques of machine learning methods which will be of critical importance for me and will accelerate my Ph.D. research.

Faculty Supervisor:

Mikko Karttunen

Student:

Partner:

Brown University

Discipline:

Mathematics

Sector:

Artificial Intelligence; Energy and Utilities; Health and Related Sciences & Technology

University:

The University of Western Ontario

Program: