Purdue professor in Indianapolis uses RCAC clusters to study materials, predict failures
Shengfeng Yang, an assistant professor of mechanical engineering in Indianapolis, uses the Rosen Center for Advanced Computing (RCAC)’s Negishi community cluster supercomputer to help with his research simulating complex materials. To see how failure of materials happen at the atomic level, he and his research group simulate systems with more than a million atoms, which requires a great deal of computational power, and wouldn’t be possible without a powerful computer like Negishi.
Currently, Yang and his team focus on semiconductor materials and metals such as copper that are used in semiconductor packaging to study how cracking and deformation happens at the atomic level in critical areas.
Yang and his research group also use the Gilbreth cluster’s GPUs to train machine learning models to predict material behavior and properties. This means in the future they won’t have to run time-consuming simulations because the trained machine learning model will be able to make fast predictions about material behavior and failure.
Yang says there have been no obstacles to using the clusters as a faculty member at the Indianapolis campus, and he’s been able to access the clusters remotely without any difficulties.
He says tapping into RCAC resources has also connected him to faculty in West Lafayette he might not otherwise have met.
“I’ve gotten a lot of connection opportunities, and chances to collaborate with faculty in West Lafayette that are more focused on the experiment side, so we can have that connection between the computational simulation and the experimental science. So that’s been a big benefit as well.”
To learn more about Negishi, Gilbreth and other RCAC resources, contact rcac-help@purdue.edu and visit the RCAC website.