molsim

▌Theme of the course

Molecular modeling and simulation allow us to calculate macroscopic properties of a system from the knowledge of microscopic details such as forces of interactions. In other words, it is a mathematical way of mimicking the behavior of a collection of molecules. The simulation techniques can be broadly classified into two categories: Molecular Dynamics (MD) and Monte Carlo (MC). MD generates the trajectory (coordinates as a function of time) of a system of molecule as it goes from an initial state to the corresponding final state. From these trajectories, one can measure time-averages of properties of interest. Contrastingly, MC samples the phase space with an underlying distribution dictated by statistical mechanics, and one measures properties averaged over an ensemble of microstates of the system. Moreover, machine learning has remarkable ability to classify, recognize, and characterize complex patterns and trends in large data sets, and help in accelerating molecular simulation and improve the accuracy of molecular simulation-based property calculation. Integration of these two tools - molecular simulation and machine learning - can accelerate materials design and development.

▌Objective and scope

The objective of this course is to disseminate principles of molecular simulation and machine learning techniques and show how these two tools can be used to compute and predict materials properties. The course will highlight best practices, key challenges, and important advances in molecular model development, simulation techniques, advanced sampling methods, use of machine leaning for data analysis. Participants will learn ways to improve and validate the prediction of molecular simulations and machine leaning, usefulness of these tools in materials research and education.