Computational methods for nonlocal models

NLMech library welcome page.

Develop efficient and parallel computational methods for class of nonlocal models such as Peridynamics and nonlocal diffusion equations


1. Google Summer of Code 2020

Towards the development of a massively parallel computational library for peridynamics and other nonlocal models, together with P. Diehl, we prepared a computational problem where the goal was to develop a massively parallel library for the simple 2D nonlocal heat equation. See the rough description of the problem. This project was sponsored by Google in Google Summer of Code 2020 and was assigned to student P. Gadikar (IIT Madras). Pranav’s work is open source and can be found here. Conference paper on proposed load balancing algorithm is in this link GSoC 2020 conference paper.

2. NLMech library

We have launched peridynamics code NLMech that utilizes HPX library for efficient multi-threading computation. We aim to make the NLMech library more user-friendly and easily extensible in the coming days. Our future goals are to

  • develop a fully parallel library following some of the key ideas and algorithms from our GSoC 2020 work

  • implement Quasistatic discretization

The paper with P. Diehl briefly describing the open-source library is currently under review in JOSS (Journal of Open Source Software). The review of the code and paper is open and can be seen live here.

Prashant K. Jha
Prashant K. Jha
Research Associate

My research focuses on using mechanics, applied mathematics, and computational science to understand and represent the complex behavior of materials, e.g., crack propagation and multiphysics effects in materials and particle breakage and interlocking in granular media. My research expertise and interests include mechanics of solids and granular media, multiphysics and multiscale modeling of engineering and biomedical systems, uncertainty quantification, and applications and advancement of neural networks to engineering problems.