We are located in the Department of Civil and Systems Engineering at the Latrobe Hall. Dr. Somdatta Goswami leads the group. Somdatta’s research interests lie at the intersection of Scientific Machine Learning and Computational Mechanics, focusing on developing algorithms and architectures to accelerate traditional numerical solvers by applying deep learning methodologies. 

Latest News!

• Somdatta presented in 2025 ICTP Advanced School on Foundation Models for Scientific Discovery, hosted by the International Centre for Theoretical Physics. The presentation is titled: Slides are available on GitHub.

• Somdatta presented on IVADO and CRM workshop on Accuracy and Efficiency in Scientific Machine Learning. Slides are available on GitHub.

• Dibakar won the joint 3rd position in The NASA and DNV Challenge on Optimization under Uncertainty 2025. Details are available on GitHub.

• Dibakar and Somdatta presented at EMI 2025 on various topics in neural operators and hybrid solvers . The slides are available on GitHub.

• Sharmila, Maryam and Dibakar presented at MACH 2025 on various topics in Sceintific Machine Learning. The slides are available on GitHub.

• Somdatta presented at SIAM CSE 2025 on "Pushing the Boundaries of Surrogate Modeling: Neural Operators Integrated Numerical Simulators." The slides are available on GitHub.

• Somdatta presented at BIRS workshop Uncertainty Quantification in Neural Network Models on "Scalable Surrogate Models for High-Dimensional Physics-based Systems." The slides are available on GitHub.

• Somdatta presented at Joint Mathematics Meeting 2025 on "Learning Physics-Informed Operators on Latent Spaces." The slides are available on GitHub.

• Somdatta presented at Case Western Reserve University on "Learning for Real-Time Physical System Inference: Bridging Physics and Observation." The slides are available on GitHub.

• Somdatta hosted a webinar at the Training series of NHERI DesignSafe on "Neural Operators: Advancing Real-Time Structural Response." The slides are available on GitHub. 

• Somdatta gave a talk on "Physics Informed Operator Learning on Latent Spaces" at the Interdisciplinary Scientific Computing Laboratory of Prof. Romit Maulik at Pennsylvania State University. Slides are available on GitHub. The recording of the presentation is available here.

• Somdatta gave a plenary talk on "Employing Machine Learning Approaches to Solve PDEs in Mechanics" at the 5th International Conference on Computational Engineering (ICCE 2024) in TU Darmstadt. Slides are available on GitHub.

• Dibakar Roy Sarkar is awarded the Creel Family Engineering Fellowship.

• We are awarded a grant from NSF to research "Cardiac Digital Twins; but better!", in collaboration with Yannis Kevrekidis, Natalia Trayanova, Mauro Maggioni, and Dimitris Giovanis.

• We are awarded a grant from NSF to research "Enabling parametric sweeps on exascale AI-integrated simulations through federated learning", in collaboration with Krishna Kumar, University of Texas at Austin.

• Somdatta Goswami talked on “Advancing Real-Time Predictions in Science and Engineering: Training Neural Operators in Latent Spaces for Complex Dynamics” at the Workshop on Hybrid Machine Learning Methods for Cavitation Erosion Measurements & Predictions at Johns Hopkins University Bloomberg Center. (August 15, 2024)

• We have been awarded the Johns Hopkins Discovery Award for our research on "Data-Driven, Deep Learning Architectures for Multiscale Modeling of Biological Systems" in collaboration with Michael Lapera, Melissa Yates, and Vicky Nguyen.

• We were awarded a DOE grant to research "Physics and Uncertainty Informed Latent Operator Learning" in collaboration with Michael Shields, Lori Graham-Brady, Yannis Kevrekidis, Tamer Zaki, and Dimitris Giovanis.

• Nat Trask, Somdatta Goswami, and Michael Shields organized a mini tutorial session on "Multi-modal data-driven and physics-informed machine learning with uncertainty for materials applications" at the SIAM Conference on Mathematical Aspects of Materials Science (MS24) on May 19 – 23, 2024. The details are available on GitHub.

• Somdatta Goswami talked on “Deep Learning for Simulating Fracture in Materials” at the Princeton Center for Theoretical Science for a workshop for Fracture across Fields: Insights from Materials Science, Biology. (May 8, 2024)

• Somdatta Goswami gave a talk on “Phase Field Modeling with Neural Operators” at the SIAM Conference on Mathematical Aspects of Materials Science. (May 22, 2024)

• Somdatta Goswami gave a talk on “Employing Deep Learning to Solve Engineering Problems” at the University of Texas at Austin at the Chishiki AI Summit. (Apr. 5, 2024)