Research Experiences for Undergraduates in Computational Methods with Applications in Materials Science
MSU Affiliation
James Worth Bagley College of Engineering; Department of Mechanical Engineering; Center for Computational Sciences
Research Mentor
Doyl E. Dickel
Creation Date
7-25-2025
Abstract
Machine-learned interatomic potentials (ML IAP) for alloy systems have been sought after due to their ability to reduce experimentation costs and time, accelerating alloy development and discovery. However, the explicit inclusion of magnetism in these potentials has been both a difficult and important problem to solve, due to the complexity of spin-lattice dynamics and its significance in the properties of magnetic alloys. We present here the development of an explicitly magnetic Fe-Mn ML IAP using a physics informed neural network (PINN) extension of the rapid artificial neural network (RANN) formalism. It is shown that the potential is capable of reproducing a number of energetic, mechanical, and magnetic properties of the Fe-Mn system, including phase stability and magnetic ordering, as well as thermal and elastic properties. The presented formalism and potential should provide a useful platform for the exploration of magnetic alloy systems and their properties at the atomistic scale.
Presentation Date
Summer 7-31-2025
Keywords
magnetic alloys, magnetism, machine learning, interatomic potential
Recommended Citation
Race, Robert D. H.; Dickel, Doyl; and Messaoud, Hala Ben, "Developing a Neural Network for Prediction of Interatomic Energies in Iron-Manganese Alloys" (2025). Research Experiences for Undergraduates in Computational Methods with Applications in Materials Science. 12.
https://scholarsjunction.msstate.edu/ccs-reu/12
