Flexible Polarizable Water Model Parameterized via Gaussian Process Regression

Flexible Polarizable Water Model Parameterized via Gaussian Process Regression
Topic
Flexible Polarizable Water Model Parameterized via Gaussian Process Regression
Date & Time
Friday, May 09, 2025 - 09:00 - 10:00
Speaker
Ying-Lung Steve Tse, The Chinese University of Hong Kong
Location
Room W934, NYU Shanghai New Bund Campus & Hosted via Zoom (Meeting ID: 989 8502 9456; Passcode: 025699)

- RSVP if attending onsite -

- Join via Zoom -

Abstract:

Water is often a crucial component of many molecular simulations. To be able to accurately model chemical reactions near/at water interfaces using classical force fields, it is essential to employ a water model that is both polarizable and flexible.

SWM4/Fw, our in-house flexible polarizable water model [1], will be presented. It was optimized using Gaussian process regression, a machine-learning method that can predict a model’s properties without requiring further simulation whenever the force field parameters are changed. SWM4/Fw reproduces many reference water properties and adds flexibility important for modeling chemical reactions involving bond stretching or breaking and for calculating vibrational spectra. The model is also computationally efficient thanks to the use of extended Lagrangian with Drude oscillators to represent explicit electronic polarization. The approach to parameterize the water model based on Gaussian process regression should also be useful for developing other force fields.

Reference:

[1] Wang, X.; Tse, Y.-L. S.*, Flexible Polarizable Water Model Parameterized via Gaussian Process Regression, J. Chem. Theory Comput., 2022, 18 (12), 7155–7165.

Seminar Series by the NYU-ECNU Center for Computational Chemistry at NYU Shanghai

This event is open to the NYU Shanghai, NYU, ECNU community and the computational chemistry community.