Graphics Processing Unit Cluster for Molecular Design of High-Temperature Polymers- Building, Searching, and Modeling of Chemical Space

Abstract

The molecular design of thermosetting polymers such as polyimides traditionally has been an experimentally driven, trial-and-error process, guided by experience, intuition, and conceptual insights. Notably, the chemical space of all possible organic molecules is estimated to contain 10180 compounds, and only with computational tools, the exploration of the chemical space is possible to some extent. The vastness of the chemical space and the inefficient computing using central processing unit (CPU) clusters bring us into a dilemma, as we aim to explore unlimited space in a limited time. Leveraging the computing power of the Graphics Processing Unit (GPU) in accelerating machine learning (ML) modeling and molecular dynamics (MD) simulations, this project will tackle the significant challenges associated with the chemical spaces of polymeric materials. The efficient processing of the practically infinite chemical space requires the advanced computing power of GPU-accelerated clusters, especially their data-parallel computing to accelerate ML and MD simulations. This proposal is to acquire and build a GPU-accelerated research cluster for provide the computing power needed for AI-guided search of chemical space.

Document Details

Document Type
DoD Grant Award
Publication Date
Feb 29, 2024
Source ID
FA95502310069

Entities

People

  • Hongyi Xu

Organizations

  • Air Force Office of Scientific Research
  • United States Air Force
  • University of Connecticut

Tags

Readers

  • Parallel and Distributed Computing.
  • Quantum Chemistry
  • Systems Analysis and Design

Technology Areas

  • AI & ML
  • AI & ML - Machine Learning Algorithms
  • Space