LEAPS-MPS: Memory-Based Simulation Approaches: Developing Computational Methods and Scientists

In this project, funded by the MPS-LEAPS (Launching Early-Career Academic Pathways) Program and the Chemical Structure and Dynamics (CSD) Program in the Division of Chemistry, Professor William C. Pfalzgraff and his students at Chatham University will develop computational methods to accelerate the calculation of molecular properties from computer simulations. Computer simulations of molecules and materials are essential tools in modern chemistry, but calculating properties like viscosity, diffusion rates, and spectra from these simulations can require significant computational time and resources. The challenge associated with obtaining accurate correlation functions that can provide physical insight and allow the calculation of transport coefficients and spectra from simulations often limits researchers' ability to study complex systems of interest. Professor Pfalzgraff and his students will develop formally exact generalized master equation (GME) approaches that can be applied to any system to calculate correlation functions accurately and efficiently from molecular dynamics simulations. Their studies could open the door to more efficient calculations of transport properties and spectra in a wide variety of complex condensed phase systems relevant to drug design, advanced materials, and energy production. This work will also create comprehensive pathways for training undergraduate students in theoretical chemistry research using early research exposure, integrating training in programming and mathematics with research, multi-level mentoring, and open-source code development.

This work seeks to develop methods that extract maximum information from limited simulation data by using memory kernels that encode how the environment influences the dynamics of observable of interest. In many condensed phase systems, these memory kernels decay to zero more rapidly than the correlation functions themselves, allowing researchers to run short simulations to obtain the memory kernel and then use the GME to predict long-time behavior with minimal additional computational cost. The research team will optimize collocation-based numerical methods for extracting memory kernels from simulation data and explore alternative mathematical formulations that accelerate memory kernel decay. The methods will be implemented in open-source software that can be applied to any molecular system.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

Award Number: 2532916
Principal Investigator: William Pfalzgraff
Funds Obligated: $249,950
State: PA