2020/08/24 by Vikram Jadhao, Jadhao, Vikram, JCS Kadupitiya +1
Computer Science · #Online Learning and Analytics
paper · pdf · doi:10.48550/arxiv.2008.13518
We explore the idea of integrating machine learning (ML) with high\nperformance computing (HPC)-driven simulations to address challenges in using\nsimulations to teach computational science and engineering courses. We\ndemonstrate that a ML surrogate, designed using artificial neural networks,\nyields predictions in excellent agreement with explicit simulation, but at far\nless time and computing costs. We develop a web application on nanoHUB that\nsupports both HPC-driven simulation and the ML surrogate methods to produce\nsimulation outputs. This tool is used for both in-classroom instruction and for\nsolving homework problems associated with two courses covering topics in the\nbroad areas of computational materials science, modeling and simulation, and\nengineering applications of HPC-enabled simulations. The evaluation of the tool\nvia in-classroom student feedback and surveys shows that the ML-enhanced tool\nprovides a dynamic and responsive simulation environment that enhances student\nlearning. The improvement in the interactivity with the simulation framework in\nterms of real-time engagement and anytime access enables students to develop\nintuition for the physical system behavior through rapid visualization of\nvariations in output quantities with changes in inputs.\n