Molecular dynamics (MD) simulations accelerated by high-performance computing (HPC) methods are powerful tools for investigating and extracting the microscopic mechanisms characterizing the properties of soft materials such as self-assembled nanoparticles, virus capsids, confined electrolytes, and polymeric fluids. However, despite the employment of optimal parallelization, scientific simulations can often take hours or days to furnish accurate information, and deep learning (DL) has the potential to address this critical need. In this talk, I will discuss ideas on integrating DL methods with HPC-accelerated MD simulations of soft materials in order to enhance their predictive power and advance their applications for research and educational activities. Parallelized MD simulations of self-assembling ions in nanoconfinement are employed to demonstrate our approach. Through this demonstration, I will introduce ``machine learning surrogates'' for MD simulations of soft-matter systems. I will also demonstrate a deployed web application on nanoHUB to realize the advantages associated with the Machine Learning surrogates.

Google CoLab notebook and dataset

Engineered nanoBIO Node at Indiana University

Researchers should cite this work as follows:

• JCS Kadupitiya (2021), "Designing Machine Learning Surrogates for Molecular Dynamics Simulations," https://nanohub.org/resources/35650.

  BibTex | EndNote

1. high performance computing (HPC)
2. machine learning
3. material properties
4. molecular dynamics (MD)
5. nano/bio
6. NanoBio Node at Indiana
7. nanoparticles
8. self-assembly
9. Simulation
10. soft materials