DJ, Haiwen, and Zach present posters at UQ-MLIP with DJ winning the best student poster award! Congratulations to all of them for representing ISCL so brilliantly. Romit also presents a talk on GNNs for SciML with work led by Shivam and Hojin.
Announcing a new preprint for non-Gaussian data assimilation of nonlinear observations led by Zachariah Malik at UC Boulder. A transformation of variables to a latent space enables more accurate application of conditional-Gaussian based Kalman filtering. Read our preprint here.
New preprint related to the use of distributed graph neural networks for spectral super-resolution led by Shivam! Applied for subgrid scale reconstruction of the Taylor-Green Vortex case at high Reynolds numbers. Read more here.
New paper with Professor Valerio Iungo at UT Dallas related to the development and use of data-driven algorithms for the modeling and analysis of wind-turbine data is accepted in Applied Energy!
New preprint on using higher-order quantum reservoir computing for learning dynamical systems is on the Arxiv (link to paper). This work was led by Vinamr Jain, an undergraduate intern in our group from IIT-Delhi.
Our new paper on using ensemble data assimilation to improve the Spalart Allmaras RANS turbulence model is published in Physical Review Fluids. This excellent piece of work was led by Deepinder (as Research Aide at Argonne National Laboratory).
Pleased to announce a new collaborative project with the EVS division of Argonne National Laboratory for Foundation Models in Surface Hyrdology. This will be collaborative with Jeremy Feinstein, Ross Alexander, Hong Zhang, and Rao Kotamarthi.
Zachariah Malik (UC Boulder) and I have been working on data assimilation for systems with non-Gaussian noise and have a new preprint that demonstrates how some classical methods prove to be quite competitive even against state-of-the-art deep learning based techniques. Our new preprint is now available on the Arxiv.
Our paper on learning chaotic dynamical systems using a novel neural ordinary differential equations is now on the Arxiv (link to paper). This work was led by Dibyajyoti Chakraborty and in collaboration with Kevin Chung at Lawrence Livermore National Laboratory.
A new preprint on using the Spherical Fourier Neural Operator to build climate emulators with quantified uncertainty is out (link to paper). This work was led by Haiwen Guan and is in collaboration with Dr. Troy Arcomano (Argonne National Laboratory) and Prof. Ashesh Chattopadhyay (UC Santa Cruz). Update: This work has now been accepted at the ICML workshop on Machine Learning for Earth System Modeling (2024)!
Our paper on multifidelity PINNs is published in CMAME! See a preprint of this paper here. Thank you to Dr. Sunwoong Yang and ISCL member Hojin Kim for their work on this collaboration.
Congratulations to Dibyajyoti Chakraborty and Haiwen Guan for securing visiting researcher positions at Los Alamos National Laboratory and Argonne National Laboratory! They will work on collaborations between ISCL and these institutions for neural differential equations and climate model emulation research respectively.
A collaboration with Prof. Aditya Grover (UCLA) and (Rao Kotamarthi) Argonne National Laboratory leads to a best paper at the ICLR 2024 Climate Change for AI workshop! See a preprint of this paper here. We introduce a novel vision transformer based emulator for the atmosphere.
A new paper published in collaboration with Gianmarco Mengaldo (NUS), Oliver Schmidt (UCSD), Marcin Rogowski, Matteo Parsani, and Lisandro Dalcin (KAUST) is published in Computer Physics Communications. Read our preprint here.
Our paper on the use of the Quantum Approximate Optimization Algorithm to build reduced-order models of fluid flows is published in the Journal of Computational Physics! This work was done in collaboration with Argonne National Laboratory and the University of Glasgow. Read more here.
Shivam Barwey presents work on GNN-based methods for SciML at PASC in Zurich.
Romit presents a guest lecture at Texas A&M's NUEN 689 course ("Deep Learning for Engineering Applications"). Thank you to Professor Yang Liu for the invitation!
Romit presents invited talks related to work on scientific machine learning at the HKUST Mechanical Engineering Seminar Series and the UNRAVEL project in Netherlands (thank you to Prof. Lin Fu and Benjamin Sanderse for the invitation!)
Two papers accepted as posters at the ICLR 2024 AI4DiffEqtnsInSci workshop! The first introduces a scalable vision transformer model for weather forecasting (Stormer) and the second studies the preservation of time-scales in the latent space of autoencoder reduced-order models. Official links to be posted shortly.
A new perspective piece of machine learning for wind energy is published in Theoretical and Applied Mechanics Letters. This work, titled "A Call for Enhanced Data-Driven Insights into Wind Energy Flow Physics", is collaborative with Professor Iungo and Coleman Moss at the University of Texas at Dallas.
New preprint out on a state-of-the-art transformer based model for weather forecasting! This work was collaborative with UCLA (Professor Aditya Grover). Read more here.
New preprint out on how we can construct graph neural networks for surrogate modeling with a-posteriori error tagging - a novel technique that improves the explainability of surrogate models! This work was led by Shivam Barwey. Read more here.
Romit presents a talk on anomaly detection at the Brij Disa Center for Data Science and AI at IIM Ahmedabad. Thank you Anindya Chakrabarti and Debjit Ghatak for the invitation!
Romit has been awarded an "Impact Argonne Award" for "For developing an AI model for climate and creating an innovated data assimilation methodology" by Argonne National Laboratory.
Our paper on interpretable latent subgraph extraction for graph neural networks is published in the Journal of Computational Physics! This work was led by Dr. Shivam Barwey at Argonne National Laboratory. Read more here.
Our paper on using Bayesian online changepoint detection to detect anomalies in complex dynamical systems is published in Chaos! This work was led by Sen Lin (at the University of Houston) in collaboration with Gianmarco Mengaldo at the National University of Singapore. Read more here.
Our paper on using projection pursuit optimal transport for learning stochastic dynamical systems is published in Chaos! This work was led by Jonah Botvinick-Greenhouse in collaboration with Yunan Yang at Cornell. Read more here.
ISCL is funded by the U.S. Department of Energy for developing novel scientific machine learning algorithms for Tokamak Disruption mitigation. More information here here.
Neelappagouda (Neel) and Dibyajyoti (DJ) are the newest graduate students at ISCL-Penn State. They will be working on scalable unstructured scientific machine learning for dynamical systems. Welcome!
Romit visited George Mason University to provide a seminar on the 27th of October (related to Graph Neural Networks and SciML). Thank you Sean Carney and Harbir Antil for the kind invitation!
Romit provides an invited talk at the TU Delft Aerospace Seminar series. Many thanks to Anh Khoa Doan and Richard Dwight for the hospitality!
Romit visis the CWI Autumn School on Scientific Machine Learning in Amsterdam and provides an invited talk on Differentiable turbulence modeling. A big thank you to Benjamin Sanderse for the invitation!
Varun Shankar defends his thesis at Carnegie Mellon! ISCL is grateful to have him and Dr. Vishwanathan as a collaborator.
Romit is appointed Editor of the Elsevier journal, Results in Engineering. We welcome manuscripts from studies related to scientific machine learning for fluid dynamical systems.
New preprint out for differentiable turbulence modeling for complex unstructured geometries with multiscale graph neural networks with Varun Shankar and Venkat Vishwanathan! See here for more details.
Romit and Shivam present work related to PythonFOAM and interpretable unstructured scientific machine learning at USNCCM 2023, in Albuquerque, New Mexico.
Our paper on using neural architecture search ensembles for dynamical systems forecasting and flow reconstruction is published in Physica D. Read more here.
New preprint out for differentiable turbulence modeling with Varun Shankar and Venkat Vishwanathan! See here for more details.
Romit officially starts his faculty position at Penn State University and transitions to a joint appointment at Argonne National Laboratory.
Shivam presented an invited talk on scaling graph neural networks for PDE surrogates at PASC 2023, Zurich.
Matt Poska (co-advised with Dr. Sharon Huang) wins the U.S. DOE SCGSR award for explainable scientific machine learning! Matt will spend the Fall semester working at Argonne National Laboratory to develop a novel multiresolution analysis enhanced dynamical systems forecasting architecture.
New preprint out on the deployment of reduced-order models on quantum computers. We show how a reformulation of DMD can be used to execute simulations of nonlinear dynamical systems on near-term quantum computing devices such as annealers. Read more about this work with Katherine Asztalos (Argonne National Laboratory) and Rene Steijl (University of Glasgow) here.
Romit visited the BIRS SciML workshop from June 18-23, 2023 and presented a cross-section of the group's activities in beautiful Banff Canada.
Romit visited the topical workshop on Mathematical and Scientific Machine Learning at ICERM (Brown University) for an invited talk on neural architecture search for scientific machine learning from June 4-9, 2023. Thank you ICERM for the invitation!
Romit visited the National University of Singapore for the International Workshop on Reduced Order Methods from May 21-25, 2023. Thank you to the organizers (and particularly Gianmarco Mengaldo) for the kind invitation!
Our work on multifidelity reinforcement learning for airfoil shape optimization has been published in the Journal of Computational Physics. Congrats to our summer interns Sahil Bhola and Suraj Pawar! Read more here.
Our workshop paper on the practical implications of invariant-preserving geometric deep learning for computational fluid dynamics is published in the ICLR Physics4ML workshop. Congratulations to Varun Shankar and Shivam Barwey!
Romit visited the Department of Mathematics at University of Wisconsin, Madison on Friday, April 14. Thank you Nan Chen for the invitation!
Romit presented work related to learning stable dynamical systems from data at SIAM-CSE 2023, Amsterdam (Feb 26 - Mar 3).
Romit presented at the Rutgers Efficient AI Seminar Series on the 23rd of February on his work at the intersection of high-performance computing and scientific machine learning.
Dr. Faez Ahmed from MIT visited us at Argonne on the 14th of February for discussing research in generative machine learning for mechanical design.
Shivam presented a talk to the ALCF Datascience team on a novel interpretable and multiscale graph neural network architecture on Feb 9, 2023.
Our minisymposium proposal titled "MS 423 - Recent Advances in Data-Intensive Physics-Informed Machine Learning for Accelerating Computational Science", jointly chaired by Qi Tang, Joshua Burby, and Romit Maulik, was accepted by USNCCM 2023 at Albuquerque New Mexico. Please consider submitting an abstract here. The deadline for submission is January 30.
Dr. Jianxun Wang from Notre Dame visited us at Argonne on the 26th of January for discussing research in Physics-informed machine learning.
Our paper on using differentiable physics closure models for Burgers turbulence is accepted in Machine Learning Science and Technology. This work was led by Varun Shankar, a PhD candidate at Carnegie Mellon University. Read more here .
Alec's paper on building stable neural ordinary differential equations for chaotic dynamical systems is accepted in the Journal of Computational Physics! Congratulations to Alec and our co-authors. An arxiv preprint is available here.
Our work on deep-learning based surrogate modeling for internal combustion engine simulations is awarded the best paper in ASME Internal Combustion Engine Fall Conference, 2021. This was work in collaboration with Sudeepta Mondal, Gina Magnotti, Bethany Lusch, and Roberto Torelli.
Our large multi-institute collaborative project for Community Research on Climate and Urban Science (CROCUS) has been accepted! We will be working on developing novel parameterizations and surrogates for assessing the impact of tree canopies on street level simulations in collaboration with Drs. Kotamarthi, Fytanidis, and Fernando.
Romit visited the SIAM Convening on Climate Science, Sustainability, and Clean Energy (DMS 2227218), Tysons Corner, Virginia, October 10-12, 2022. More details here.
Romit provided an invited talk at SIAM Mathematics of Data Science in San Diego on the 28th of September titled: "Stabilized Neural Ordinary Differential Equations for Learning Chaotic Dynamical Systems". See details here.
Romit visited the Department of Mathematics at the University of Pittsburgh on the 20th of September for a Computational Mathematics Seminar.
Dr. Shivam Barwey started as a postdoc in our group. Read more about his research here. Shivam will be working on graph-based scientific machine learning methods for surrogate modeling.
Romit provided an invited talk on ensemble-based uncertainty quantification for machine learning and was an early-career panelist at USACM UQ-MLIP. See slides here.
Romit provided a lecture on machine learning for dynamical systems at ATPESC 2022. See his talk slides here.
Romit provided an invited talk for surrogate modeling with neural ODEs at Virginia Tech for the ARIA conference.
Our joint work with the University of Texas at Dallas and the University of Utah for data analysis of wind-turbine wakes is 'Scilighted' and featured on the cover of the Journal of Sustainable and Renewable Energy! Read more here.
Romit (in collaboration with Bethany Lusch, Saumil Patel, Bulut Tekgul, and Dimitrios Fytanidis) organized a 4 hour workshop on coupling a Python-based data science ecosystem with OpenFOAM in the first ever PythonFOAM workshop! Get course materials including lecture videos and code here.
Romit has been awarded an "Impact Argonne Award" for "tackling several climate model challenges and advancing the field of downscaled climate modeling and impact analysis" by Argonne National Laboratory.
Our novel deep learning architecture for time-varying unstructured data is highlighted by TechXplore.
Sahil Bhola is in the news! His work on exploring multifidelity deep reinforcement learning for aerodynamic optimization is highlighted by Argonne.
Our collaborative proposal to study the mathematics of surrogate modeling for nonlinear dynamical systems is accepted for funding by DOE.
Our collaborative proposal to study mathematical, computational, and hardware considerations for surrogate modeling of PDEs is accepted for funding by NSF.
Janah Richardson, an intern from Walter Payton College Prep High School, wins a gold medal for her ACT-SO project! She investigated structural inequalities using statistical modeling in Chicago. (see news story here).
Our research into using neural ordinary differential equations for reduced-order modeling is a Physica D. highly cited article! (scroll down the page to find our entry).
Our Physics of Fluids paper for surrogate modeling of advection-dominated flows using deep convolutional autoencoders is chosen as Editors' pick!
Our studies of simulation and data science interoperability have been selected for an Exascale Computing Project Proxy application.
NSF MSGI intern Dominic Skinner's excellent work is featured by DOE!
Our research on geophysical surrogate modeling using recurrent neural architecture search is covered by HPCWire, Newswise, and Insidehpc.
Our Physics of Fluids paper for stable non-intrusive reduced-order model using bidirectional LSTMs is chosen as Editors' pick!
Check out a feature story on Romit's Margaret Butler Fellowship project.