News & Events

08/2026 Awards: Our CIMS Lab has received a DOE Genesis Mission Award to develop generative AI-enabled digital twins for subsurface fracture systems. In collaboration with PNNL and LBNL, the project will develop AI-enabled digital twins that integrate multiphysics modeling and sensor data to provide rapid, accurate predictions of evolving underground environments. More information is available from the UMN News and CSE News announcements.

Our group has received the prestigious NSF CAREER Award for developing learning-enabled computational multiphysics methods to advance the understanding and prediction of failure processes in coupled civil–geosystems. More information is available in the CEGE News.

Our group contributed 3 poster presentations at AGU Annual Meeting 2025. If you are interested in our work, please check out our posters:

NG31A-VR8968A: Du et al., Differentiable Meshfree Framework for Hybrid ML–Physics Forward and Inverse Modeling of Geomechanics and Geophysical Flow Hazards

H43F-VR8718: Lin et al., Differentiable Physics-constrained subsurface inversion with Latent Diffusion model

H33R-1537A Feng et al., Comparative Study of Physics-Informed and Data-Driven Neural Networks for Compound Flood Simulation at River-Ocean Interfaces: A Case Study of Hurricane Irene

CIMS Lab at USNCCM-18: Showcasing Advances in Differentiable and AI-Augmented Mechanics

We’re excited to share the strong presence of the Computational Intelligence and Mechanics Systems (CIMS) Lab at the 18th U.S. National Congress on Computational Mechanics (USNCCM), held in Chicago, July 20–24:

• Guo, Binyao et al., Attention-Enhanced Fourier Neural Operators for Robust Constitutive Modeling of History-Dependent Nonlinear Materials [Abstract]
• Lin, Zihan et al., A Latent Diffusion Model-Coupled Differentiable Physics Simulator for Inverse Modeling of Heterogeneous Media [Abstract]
• Du, Honghui et al., A Differentiable Meshfree Method for Nonlinear Mechanics and Geomechanics Modeling [Abstract]

Happy to share two recent preprints from my research group:

1. History-Aware Neural Operator (HANO) for robust constitutive modeling of path-dependent materials (https://arxiv.org/abs/2506.10352)

2. JAX-MPM, a learning-augmented differentiable meshfree framework enabling GPU-accelerated Lagrangian simulation and geophysical inverse modeling (https://arxiv.org/abs/2507.04192)

HANO aims to overcome key limitations related to loading resolution dependence and internal variable representations in path-dependent materials modeling. JAX-MPM provides a unified, high-performance framework for forward and inverse simulations in geomechanics and geophysics, in which we particularly focus on complex scenarios involving large deformations, inelastic responses, and fluid-solid interactions.

Both frameworks are fully differentiable and seamlessly integrate with modern deep learning platforms.

Feel free to explore and connect for further discussions!

Dr. He presented recent advancements in a differentiable Eulerian–Lagrangian framework (JAX-MPM) for nonlinear mechanics and landslide hazard modeling at the 2025 ASCE Engineering Mechanics Institute (EMI) Conference in Anaheim, California, from May 27–30, 2025.

As of March 30, 2025, our group is officially named the Computational Intelligence and Multiphysics Simulation Lab (CIMS Lab) — formerly known as the Intelligent Computational Mechanics Group (InCOME).

This name change reflects our broader and evolving vision: to advance the integration of artificial intelligence, physics-informed machine learning, and high-performance simulation science in addressing complex multiphysics challenges across engineering, geoscience, biomechanics, and beyond. Thank you for your continued support — we’re excited for the journey ahead!
 

The project is about leveraging state-of-the-art Generative AI techniques combined with mesoscale modeling to accelerate the discovery and optimization of novel polymer materials. We sincerely thank 3M and our collaborators for their generous support.

Updates in ASME's International Mechanical Engineering Congress & Exposition (IMECE) 2024 in Portland, OR.

Presentations
Our group was involved in two presentations on advancing differentiable mechanics: 
- "A Neural Network-Enhanced Differentiable Meshfree Method for Computational Mechanics" Contributors: Honghui Du, Binyao Guo
- "Neural Topology Optimization Based on Differential Programming with Principled Constrained Optimization" Contributors: Ryan Devera (Computer Science & Engineering) and Binyao Guo

Committees
Dr. He served on the ASME AI/Deep Learning Best Paper Honors Committee, and joined the IMECE Fracture and Fatigue Mechanics Technical Committee (FFMTC)

Awards
Honghui received the ASME Applied Mechanics Division's Robert M. and Mary Haythornthwaite Foundation Student Travel Award for the work on "Neural-Integrated Meshfree Method for computational mechanics". Congratulations!

I was pleased to share our recent research on Differentiable Solid Mechanics and Geophysics with students and colleagues during my visit to Singapore in September. The talk was titled "Neural-Integrated and Data-Driven Approaches for Computational Modeling in Solid Mechanics and Geophysics." I am grateful for the warm invitations from NUS, NTU and A*STAR.

• ME Department Seminar, National University of Singapore (NUS)
• iHPC Seminar, Agency for Science, Technology and Research (A*STAR)
• CEE Department Seminar, Nanyang Technological University (NTU)