BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Deep Generative Modeling for Computational Mechanics: Inverse Desi
 gn and PDE-Based Simulations
DTSTART:20250605T161500
DTSTAMP:20260416T182026Z
UID:b0692707964179486799130f74d9866165438976456bb95419bd7813
CATEGORIES:Conferences - Seminars
DESCRIPTION:Dr. Yaohua Zang\, Technical University of Munich\nAbstract: 
  \n \n\nInverse materials design and simulation are at the heart of acce
 lerating innovation in computational materials science. In this talk\, we 
 introduce a novel two-part research framework that leverages deep generati
 ve modeling to tackle both material discovery and the simulation of underl
 ying physical systems.\n\n \n\nWe first introduce PSP-GEN\, a probabilist
 ic generative model for inverse materials design that operates across the 
 full Processing–Structure–Property (PSP) chain. By learning low-dimens
 ional\, microstructure- and property-aware latent representations\, PSP-GE
 N enables the generation of process-parameterized microstructures that are
  both high-performing and physically realizable. This approach is particul
 arly effective under small training data\, high-dimensional design spaces\
 , and out-of-distribution target properties\, making it a powerful tool fo
 r practical material discovery.\n\n \n\nNext\, we present DGenNO (Deep Ge
 nerative Neural Operator)\, a novel physics-informed framework for solving
  parametric PDEs and their associated inverse problems. Unlike existing De
 ep  Neural Operator (DNO) solvers\, DGenNO combines probabilistic generat
 ive modeling with neural operator learning to produce fast\, uncertainty-a
 ware\, and physics-consistent solutions. It employs a new NO architecture\
 , MultiONet\, and a physics-driven training strategy that embeds weak-form
  residuals as virtual observations. This allows DGenNO to learn from both 
 labeled and unlabeled data\, significantly improving performance in noisy\
 , discontinuous\, and data-limited scenarios.\n\n \n\nWhile developed ind
 ependently\, PSP-GEN and DGenNO are fundamentally synergistic. PSP-GEN pro
 vides a high-level inverse design framework\, while DGenNO serves as a rob
 ust\, data-efficient solver for the SP (Structure–Property) link by solv
 ing PDEs governing material behavior. Their shared foundation in generativ
 e\, probabilistic modeling paves the way for a future unified framework th
 at combines design\, simulation\, and uncertainty quantification in a phys
 ically consistent loop.\n\n \n\nThis research was conducted in collaborat
 ion with Prof. Faidon-Stelios Koutsourelakis at the Technical University o
 f Munich (TUM).
LOCATION:MA A1 12 https://plan.epfl.ch/?room==MA%20A1%2012
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR