Welcome to the group page of the Interdisciplinary Scientific Computing Laboratory (ISCL)!. ISCL performs research at the intersection of data science, applied mathematics, and high-performance computing to enhance the understanding of complex multifidelity and multiphysics phenomena in various applications. In other words - we create science-based AI algorithms for applications such as weather and climate modeling, disruption mitigation in nuclear fusion, data and model fusion for complex fluid flows, surrogate models for chaotic dynamical systems, and more. ISCL also runs a scientific machine learning seminar series with leading researchers in our area of study - check it out here!
ISCL is housed in the College of Information Sciences and Technology at Pennsylvania State University, as well as the Mathematics and Computer Science Division at Argonne National Laboratory. A high level overview of our research may be found in the following talks: [1], [2], [3], [4], [5], [6]. Further information about publications can be found on Google Scholar and our software contributions are available on Github. ISCL has access to multiple HPC resources such as Bebop/Swing/Polaris (at Argonne), Roar (Penn State), and Perlmutter (NERSC). ISCL eagerly welcomes possibilities for education, collaboration, and consulting! Feel free to reach out to us for any questions.
News
- Pleased to announce a new preprint from our group, led by Dibyajyoti Chakraborty, on the use of generative modeling for zero-shot multimodal data assimilation for recovering the state of the state of the atmosphere from sparse data. By doing score-matching during inference, rapid state reconstructions are possible in seconds given observations from varying modalities. We are able to recover ERA5 reanalysis using partial observations of the same as well as sparse and unstructured observations from the IGRA radiosonde dataset. Read more on the Arxiv.
- Announcing a new preprint titled "FIGNN: Feature-Specific Interpretability for Graph Neural Network Surrogate Models" where we extend previous work on interpretable subgraph extraction to feature-specific adaptivity. Our graph neural network models not only make predictions but also identify feature-specific spatially coherent subgraphs where a-posteriori errors are high. A preprint is available on the Arxiv.
- We are pleased to announce a new award providing 32000 A100 GPU hours and 20000 CPU node hours on Perlmutter through the NERSC AI4Science program!
- Congratulations to Dr. Xuyang Li for getting accepted to the DL4Science Workshop at Lawrence Berkeley National Laboratory. Congratulations also to Hojin Kim for getting accepted to ATPESC 2025 for scaling our differentiable physics codes!
- Romit presents an invited talk at the IUTAM Workshop on Machine Learning for Diverse Fluid Mechanics at Okinawa. Thank you Prof. Koji Fukagata for the invitation! He also presents an invited talk at Seoul National University related to AI for Weather and Climate modeling (thank you Prof. Youngjoon Hong!) and a seminar related to differentiable physics at Inha University (thank you Prof. Sangseung Lee!).
- Congratulations to Dr. Xuyang Li for securing a Tenure-Track faculty position at UNC Charlotte! Also, congratulations to Dibyajyoti for an internship at Nvidia!
- Romit presents a Keynote talk at NCAR ISS 2025 related to DeepHyper! Learn more here.
- See other archived news here.