You're invited to attend

"Interpretable Data-Driven Modeling in Solid Mechanics with Limited Data"

by

Behador Bahmani

Columbia University

Monday, February 26
2 - 3 p.m.
Weber Lecture Hall 2
 

About the Seminar: 
The accuracy of physics-based simulations in solid mechanics is critically dependent on the fidelity of constitutive laws, which vary significantly across different materials. Traditional models often struggle to capture the complexity of material data, a challenge exacerbated by the emergence of unconventional materials such as metamaterials. This necessitates a paradigm shift towards more adaptive, data-driven modeling techniques. My research goal is to make physics simulations autonomous given diverse material data while ensuring robustness, accuracy, interpretability, and scalability. In this talk, I will begin by introducing a manifold embedding approach that enhances the robustness of model-free methods against scarce or noisy data. This strategy offers a seamless transition between fully model-based and entirely model-free frameworks based on the data’s availability, providing a flexible continuum of modeling approaches. Further, I will present an interpretable data-driven technique for discovering constitutive laws. By combining the strengths of neural networks and symbolic regression, this method efficiently uncovers symbolic, physics-constrained equations without compromising scalability. I will discuss how the proposed algorithms perform in terms of robustness and extrapolation, using both experimental data and high-fidelity sub-scale simulations.

About the Speaker: 
Bahador Bahmani has recently obtained his Ph.D. in Engineering Mechanics from Columbia University. Bahador’s research lies at the intersection of computational mechanics and scientific machine learning, focusing on developing scalable and robust algorithms for forward and inverse problems in data-driven solid mechanics. He also has industrial research experience in computational geometry and machine learning within additive manufacturing and data security sectors.