Regensburg 2025 – scientific programme
Parts | Days | Selection | Search | Updates | Downloads | Help
MM: Fachverband Metall- und Materialphysik
MM 9: Poster
MM 9.64: Poster
Monday, March 17, 2025, 18:30–20:30, P1
Investigating the Impact of Optimization Algorithms on Element-Substitution Based Materials Discovery — •David Greten1, Konstantin Jakob1, Karsten Reuter1, and Johannes T. Margraf2 — 1Fritz-Haber-Institut der MPG, Berlin — 2University of Bayreuth
In this study, we investigate how different optimization algorithms affect the relaxation of inorganic structures using general-purpose machine-learned interatomic potentials (MLIPs) like MACE-MP-0. Assessing computational efficiency and relaxation quality via structural similarity and kernel distance metrics, we find that optimizer choice significantly influences performance and can lead to different equilibrium structures. Analyzing both DFT-relaxed structures from the Materials Project and element-substitution based trial structures, we highlight the optimizer's impact in different scenarios. Our findings emphasize the critical role of optimizer selection in large-scale computational materials science workflows, particularly in the context of element-substitution based materials discovery. This can hopefully guide the community towards choosing appropriate algorithms for efficient and reliable structure relaxations.
Keywords: MACE; Relaxation; Machine Learning Interatomic Potentials; Optimization Algorithms