Past Seminars
Past Seminars
Prof. Han Gao
Han Gao is an Assistant Professor of Aerospace Engineering at Iowa State University. His research centers on AI and scientific computing, with particular emphasis on turbulence, fluid dynamics, and multiscale PDE-governed systems. He received his bachelor’s degree from Shanghai University of Electric Power and his master’s degree from Washington University in St. Louis. He earned his Ph.D. in Aerospace and Mechanical Engineering from the University of Notre Dame and completed postdoctoral training at Harvard University.
Generative AI for Turbulent Flow
Turbulent flows are ubiquitous and notoriously difficult to simulate. I will present recent progress on AI-driven generative frameworks that address these challenges by combining probabilistic diffusion models, attention-based dynamics learning, and gradient-guided generation. These approaches capture turbulence stochasticity, forecast effective dynamics, and generate flow sequences across parameters. Applications include synthesizing LES-like sequences from URANS inputs, exploring canonical channel and separation flows, and super-resolving boundary layers. Collectively, the results demonstrate scalable and versatile tools for advancing turbulence research.
Seminar date and time: September 12 (Friday), 11 AM ET.
Zoom recording here.
Dr. Daniel Serino
Dan Serino is a staff scientist in the Applied Math and Plasma Physics group (T-5) at Los Alamos National Laboratory, specializing in scientific and structure-preserving machine learning for dynamical systems and inverse problems. He earned his Ph.D. from Rensselaer Polytechnic Institute in 2019 on stable numerical methods for fluid–structure interaction and free-surface flows. Before joining LANL as a postdoc in 2022, he worked in industry on problems ranging from acoustic scattering to financial optimization. His current research includes dynamic radiography and plasma shape control.
Structure-Preserving Machine Learning for Stiff and Multi-Scale Dynamical Systems
Machine learning for scientific applications faces challenges such as limited and noisy data, multiple scales in time and space, and non-physical predictions from standard deep learning models. Structure-preserving approaches address these issues by embedding physics-based constraints into machine learning models. This talk presents two such methods using neural ODEs: (1) a fast-slow neural network (FSNN) that enforces a trainable slow manifold for multi-scale systems, enhancing integration efficiency and prediction accuracy; and (2) an autoencoder for stiff problems that splits dynamics into linear and nonlinear parts, using a Hurwitz-based parameterization to ensure stability. Stability of the nonlinear dynamics is further guaranteed with a Lipschitz-controlled network. These methods are validated on examples including Kuramoto-Sivashinsky and Abraham-Lorentz systems.
Seminar date and time: August 29 (Friday), 10 AM ET.
Zoom recording here.
Raphael Rossellini
Raphael Rossellini is a rising fourth-year PhD student in the Department of Statistics at the University of Chicago. His research is focused on developing effective statistical inference methods for black-box models. He is advised by Professors Rebecca Willett and Rina Foygel Barber.
Can a calibration metric be both testable and actionable?
Forecast probabilities often serve as critical inputs for binary decision making. In such settings, calibration—ensuring forecasted probabilities match empirical frequencies—is essential. Although the common notion of Expected Calibration Error (ECE) provides actionable insights for decision making, it is not testable: it cannot be empirically estimated in many practical cases. Conversely, the recently proposed Distance from Calibration (dCE) is testable, but it is not actionable since it lacks decision-theoretic guarantees needed for high-stakes applications. To resolve this question, we consider Cutoff Calibration Error, a calibration measure that bridges this gap by assessing calibration over intervals of forecasted probabilities. We show that Cutoff Calibration Error is both testable and actionable, and we examine its implications for popular post-hoc calibration methods, such as isotonic regression and Platt scaling.
Seminar date and time: August 15 (Friday), 10 AM ET.
Zoom recording here.
Prof. Anne Talkington
Anne Talkington is an Assistant Professor at the University at Buffalo Department of Pharmaceutical Sciences. Dr. Talkington works at the intersection of mathematical modeling, immunology, oncology, and pharmacology. She received her B.S. in Mathematics and B.A. in Biology from Duke University, and her M.S. in Applied Mathematics and Ph.D. in Computational Biology from the University of North Carolina at Chapel Hill. She then pursued postdoctoral research at the University of Virginia in Systems Immunology. Dr. Talkington recently completed a National Research Council Fellowship at the National Institute of Standards and Technology, where she developed a multiscale modeling framework for tumor-immune interactions. Since joining UB in early 2025, her emerging research group has focused on understanding spatiotemporal immune activity in the tumor microenvironment and optimization of therapeutic strategies for immune checkpoint inhibition. The Talkington Lab integrates computational models and experimental data to answer questions of optimization in drug activity and delivery strategies.
Modeling tumor-immune interactions and optimal perturbations
Despite recent advances in therapeutic strategies, many cancers are not responsive to treatment. One way in which cancer is consistently able to evade an immune response is through a series of immune checkpoints, or inhibitors. Immune checkpoint blockade seeks to prevent this evasion by competitively binding to the inhibitory receptors on the immune cell surface, thereby reducing the cancer’s opportunities for immune downregulation. We explore the optimization of immune checkpoint blockade from a multiscale perspective. We first consider an ODE system for tumor-immune cellular dynamics and immune checkpoint perturbation at the whole-tumor level. We then consider an agent-based model for tumor-immune interactions at the individual cell level. Finally, we demonstrate the role of immune checkpoint blockade efficiency in our model system. This work holds promise for informing optimal therapeutic design strategies.
Seminar date and time: May 9, 2025. 10 AM ET.
Zoom recording here.
Dr. David Vargas
David Vargas is currently the John von Neumann fellow at Sandia National Labs. He has a PhD in applied mathematics from the University of New Mexico where he studied the application of Parallel-in-Time methods to chaotic systems as part of his dissertation work under Dr Jacob Schroder. David is broadly interested in nonlinear multigrid methods, including space-time parallel multigrid for turbulent systems and multilevel training algorithms for machine learning.
Parallel-in-Time Relaxation for Chaotic Systems with Local Shadowing
Despite the fact that Parallel-in-Time (PinT) methods are predicted to become necessary to fully utilize next-generation exascale machines, there are currently no known practical methods which scale well with the length of the time-domain for chaotic problems, due to exponential dependence of the condition number on the fastest chaotic time-scale. While most prior works applying multigrid-in-time to chaotic systems focus on the coarse grid equation, we investigate the relaxation techniques commonly used and demonstrate that they in fact diverge for chaotic systems. Here the novel Local Shadowing Relaxation (LSR) is presented and proven to be a convergent, PinT relaxation for chaotic PDE systems. Promising preliminary analytical results and numerical experiments with the Lorenz system indicate that LSR may solve the scaling problem for chaotic systems, potentially allowing space-time parallelization of turbulent computational fluid dynamics.
Seminar date and time: May 2. 10 AM ET.
Zoom recording here.
Prof. Shaowu Pan
Shaowu Pan is currently a tenure-track assistant professor in the Department of Mechanical, Aerospace, and Nuclear Engineering at RPI starting from 2022 Fall. He is also affiliated with the Rensselaer-IBM Artificial Intelligence Research Collaboration (AIRC). He received M.S. and Ph.D. in Aerospace Engineering and Scientific Computing from the University of Michigan, Ann Arbor in April 2021. Then he started as a Postdoctoral Scholar in the AI Institute in Dynamic Systems at the University of Washington, Seattle from 2021 to 2022. His research interests are scientific machine learning for large-scale PDE systems and operator-theoretic modeling and control of nonlinear systems.
Reduced-Order Learning for Complex Dynamics: Stable Koopman Control and Scalable Surrogates
Nonlinear dynamical systems are ubiquitous in science and engineering. While high-fidelity models based on first principle has been well established, reduced-complexity models are of particular interest in recent years due to its feasibility for many-query tasks, e.g., uncertainty propagation, optimal control and design. In this talk, I will discuss two recent works from my group at RPI in this theme. For the first half, I will talk about learning reduced complexity models in the light of learning Koopman operator for control. We highlight our solutions to address several issues when standard algorithm is applied, e.g., noisy data, instability, efficient curation of training data, choice of observables. We demonstrate the benefits of our proposed framework on model predictive control of a Koopman-based surrogate model for CartPole problem. For the second half, I will talk about learning reduced complexity models in the light of surrogate model of time-dependent partial differential equations from data. In contrast to existing frameworks, our model ensures stability of learned surrogate model through a stable parametrization of Koopman operator and trapping theorem of linear quadratic dynamics. Moreover, our model is agnostic to mesh, scalable to 3D problems and could only require a few sensor measurement during inference stage. We demonstrate the benefits of our model over several state-of-the-art neural operator frameworks on 2D wave propagation, 2D Navier-Stokes in a periodic box, shallow water equations.
Seminar date and time: April 25, 2025. 10 AM ET.
Zoom recording here.
Dr. Christopher Rackauckas
Chris is the VP of Modeling and Simulation at Julia Computing, the Director of Scientific Research at Pumas-AI, Co-PI of the Julia Lab at MIT, and the lead developer of the SciML Open Source Software Organization. He is the lead developer of the Pumas project and has received a top presentation award at every ACoP in the last 3 years for improving methods for uncertainty quantification, automated GPU acceleration of nonlinear mixed effects modeling (NLME), and machine learning assisted construction of NLME models with DeepNLME. For these achievements, Chris received the Emerging Scientist award from ISoP. For his work in mechanistic machine learning, his work is credited for the 15,000x acceleration of NASA Launch Services simulations and recently demonstrated a 60x-570x acceleration over Modelica tools in HVAC simulation, earning Chris the US Air Force Artificial Intelligence Accelerator Scientific Excellence Award.
The Numerical Analysis of Differentiable Simulation: How Automatic Differentiation of Physics Can Give Incorrect Derivatives
Scientific machine learning (SciML) relies heavily on automatic differentiation (AD), the process of constructing gradients which include machine learning integrated into mechanistic models for the purpose of gradient-based optimization. While these differentiable programming approaches pitch an idea of "simply put the simulator into a loss function and use AD", it turns out there are a lot more subtle details to consider in practice. In this talk we will dive into the numerical analysis of differentiable simulation and ask the question: how numerically stable and robust is AD? We will use examples from the Python-based Jax and PyTorch libraries in order to demonstrate how canonical formulations of AD and adjoint methods can give inaccurate gradients in the context of ODEs and PDEs. We demonstrate cases where the methodologies are "mathematically correct", but due to the intricacies of numerical error propagation, their approaches can give 60% and greater error even in simple cases like linear ODEs. We'll then describe some of the non-standard modifications to AD which are done in the Julia SciML libraries to overcome these numerical instabilities, crucially also describing the engineering trade-offs which are required to be made in the process. The audience should leave with a greater appreciation of the greater numerical challenges which still need to be addressed in the field of AD for SciML.
Seminar date and time: April 18, 2025. 10 AM ET.
Zoom recording here.
Prof. Deep Ray
Deep Ray is an Assistant Professor of Mathematics at the University of Maryland, College Park. He obtained his PhD in Mathematics from the Tata Institute of Fundamental Research (Bangalore, India), followed by postdoctoral positions at EPFL (Switzerland), Rice University and University of Southern California. His research interests lie at the interface of conventional numerical analysis and machine learning. He has worked on the judicious integration of deep learning tools to overcome computational bottlenecks in areas such as shock-capturing in fluid flows, Bayesian inference, PDE-constrained optimization, and operator learning.
Can we learn the optimal PDE solution?
The optimal Petrov-Galerkin formulation to solve PDEs was first introduced in the 1980’s. Its goal was, given a trial basis, to approximate the finite-dimensional solution that is optimal with respect to a suitable norm. The theory is elegant and allows for the recovery of optimal convergence rates, even for problems where standard Galerkin methods fail. However, recovery of the optimal solution is contingent on being able to construct the optimal weighting functions associated with the trial basis. While explicit constructions are available for simple 1D and 2D problems, such constructions for a general multidimensional problem remain elusive. As a result, interest in this approach waned with the rise of alternative finite element strategies.
In this talk, we revisit optimal weighting functions through the lens of deep learning. We propose an operator network framework called PG-VarMiON, that emulates the optimal Petrov-Galerkin weak form of the underlying PDE. Given a suitable trial basis and a norm, the PG-VarMiON approximates the optimal finite-dimensional while implicitly learning the optimal weighting functions. We derive an approximation error estimate for PG-VarMiON, highlighting the contributions of various error sources. Several numerical results for the advection-diffusion equation are presented to demonstrate the efficacy of the proposed method. By embedding the Petrov-Galerkin structure into the network architecture, PG- VarMiON exhibits greater robustness and improved generalization compared to other deep operator frameworks, particularly when training data is limited. The proposed approach harnesses the knowledge of traditional numerical methods to solve PDEs, paving the way for constructing mathematical-sound and efficient deep surrogates.
Seminar date and time: April 11, 2025. 11 AM ET (note delayed start!)
Zoom recording here (starts a bit late because Romit forgot to hit record -sorry)
Prof. Paris Perdikaris
Paris Perdikaris is an Associate Professor in the Department of Mechanical Engineering and Applied Mechanics at the University of Pennsylvania. He received his PhD in Applied Mathematics at Brown University in 2015, and, prior to joining Penn in 2018, he was a postdoctoral researcher at the department of Mechanical Engineering at the Massachusetts Institute of Technology working on physics-informed machine learning and design optimization under uncertainty. His work spans a wide range of areas in computational science and engineering, with a particular focus on the analysis and design of complex physical and biological systems using machine learning, stochastic modeling, computational mechanics, and high-performance computing. Current research thrusts include physics-informed machine learning, uncertainty quantification in deep learning, engineering design optimization, and data-driven non-invasive medical diagnostics. His work and service has received several distinctions including the DOE Early Career Award (2018), the AFOSR Young Investigator Award (2019), the Ford Motor Company Award for Faculty Advising (2020), and the SIAG/CSE Early Career Prize (2021).
Aurora: A Foundation Model for the Earth System
Reliable forecasts of the Earth system are crucial for human progress and safety from natural disasters. Artificial intelligence offers substantial potential to improve prediction accuracy and computational efficiency in this field, however this remains underexplored in many domains. Here we introduce Aurora, a large-scale foundation model for the Earth system trained on over a million hours of diverse data. Aurora outperforms operational forecasts for air quality, ocean waves, tropical cyclone tracks, and high-resolution weather forecasting at orders of magnitude smaller computational expense than dedicated existing systems. With the ability to fine-tune Aurora to diverse application domains at only modest computational cost, Aurora represents significant progress in making actionable Earth system predictions accessible to anyone.
Seminar date and time: April 7, 2025. 11 AM ET.
Zoom recording here.
Dr. Patrick Blonigan
Patrick Blonigan is a principal member of the technical staff at Sandia National Laboratories. He currently leads or contributes to several research projects on model order reduction and sensitivity analysis with application to computational models of high-speed aerodynamics and thermal protection systems. Prior to joining Sandia in 2018, Patrick was a postdoctoral fellow in the advanced supercomputing division at NASA Ames Research Center. Patrick holds a B.S. in mechanical engineering from Cornell University, a M.S. in aeronautics and astronautics from MIT, and a PhD in aerospace computational engineering from MIT.
An Overview of Shadowing-Based Sensitivity Analysis
Adjoint methods are powerful engineering design tools for computational fluid dynamics. These methods efficiently compute the sensitivities of an objective function with respect to many input parameters. Adjoint methods have been successfully used for design optimization, error-estimation, flow control, and uncertainty quantification of steady flow simulations such as Reynolds-averaged Navier-Stokes (RANS) solvers. Recently, engineers have started to use scale-resolving simulations including Detached-eddy simulations (DES) and Large-eddy simulations (LES) for flows with unsteady separation and jets. DES and LES model these flows more accurately than RANS and capture the chaotic dynamics inherent in turbulence, making the application of adjoint methods challenging. One of the main challenges is the exponential growth in magnitude of adjoint variables with time in a chaotic dynamical system [1]. This exponential growth causes the conventional adjoint method to compute large, unusable sensitivities for scale-resolving simulations.
This talk discusses the application of adjoint methods to chaotic systems, especially scale-resolved turbulent flow simulations, using several shadowing methods including non-intrusive least-squares shadowing (NILSS) [2]. NILSS is formulated so that the exponential growth observed in conventional adjoint approaches is controlled and a bounded, useful adjoint is computed. Adjoint computations of time-averaged objective sensitivities, and some preliminary work on dual-weighted residual error estimates will be presented for the minimal flow unit for near-wall turbulence, a channel flow through a truncated domain. This talk concludes with some thoughts on the current state of research on chaotic sensitivity analysis and the future of shadowing-based sensitivity analysis.
[1] Q. Wang, R. Hui, and P. Blonigan. Least Squares Shadowing sensitivity analysis of chaotic limit cycle oscillations. Journal of Computational Physics, 267:210–224, June 2014
[2] P. Blonigan. Adjoint sensitivity analysis of chaotic dynamical systems with non-intrusive least squares shadowing. Journal of Computational Physics, 348:803–826, Nov. 2017
Seminar date and time: March 28, 2025. 10 AM ET.
Zoom recording here.
Prof. Joongoo Jeon
Joongoo Jeon is an Assistant Department of Quantum System Engineering and the Graduate School of Integrated Energy-AI at Jeonbuk National University, Korea. He received his B.S. and Ph.D. from Hanyang University (Korea). His academic journey also includes postdoctoral positions at Seoul National University (Korea) and KTH Royal Institute of Technology (Sweden). His research focuses on developing nuclear reactor digital twins by leveraging coupling of scientific machine learning and traditional numerical analysis. He also has a strong interest in fundamentals of fluid dynamics including hydrogen combustion. He is a Director of the Korean Society of Mechanical Engineers-Artificial Intelligence Division (2024 -)
Three level SciML application framework for nuclear reactor digital twin
This talk focuses on how scientific machine learning (SciML) can be effectively applied to develop digital twin technology for nuclear reactors. A key component of digital twin implementation is achieving accurate and efficient computational fluid dynamics (CFD). Recent research highlights the need for SciML techniques that can simultaneously satisfy the following key objectives: accurate prediction of unseen (future) time series in long-term simulations, accelerated computation, and a manageable amount of training data and time. To address these challenges, we propose a three-level SciML application framework: (1) real-time PDE solvers for low-fidelity CFD, (2) hybrid CFD solvers for high-fidelity simulations, and (3) effective flow control for optimal-fidelity CFD. Advanced physics-informed neural networks (PINNs) and deep operator networks (DeepONets) are highlighted as strong candidates for real-time PDE solvers. The residual-based physics-informed transfer learning (RePIT) strategy will be introduced as a hybrid solver. Lastly, this talk will cover inductive-biased deep reinforcement learning to exploit the full potential of the constructed digital twin.
Seminar date and time: March 21, 2025. 10 AM ET.
Zoom recording here.
Prof. Alexander Heinlein
Alexander Heinlein is assistant professor in the Numerical Analysis group of the Delft Institute of Applied Mathematics (DIAM), Faculty of Electrical Engineering, Mathematics & Computer Science (EEMCS), at the Delft University of Technology (TU Delft). He did his PhD at the University of Cologne, where he subsequently also spent several years as a postdoc. In his final years at the University of Cologne, he was the managing coordinator of the Center for Data and Simulation Science (CDS). After being acting full professor for Numerical Mathematics for High Performance Computing at the University of Stuttgart, he started his current position at TU Delft.
His main research areas are numerical methods for partial differential equations and scientific computing, in particular, solvers and discretizations based on domain decomposition and multiscale approaches. He is interested in high-performance computing (HPC) and solving challenging problems involving, e.g., complex geometries, highly heterogeneous coefficient functions, or the coupling of multiple physics. More recently, Alexander also started focusing on the combination of scientific computing and machine learning, a new research area also known as scientific machine learning (SciML). Generally, his work includes the development of new methods and their theoretical foundation as well as their implementation on current computer architectures (CPUs, GPUs) and application to real world problems.
Domain decomposition for physics-informed neural networks: linear and nonlinear function approximation and operator learning
Physics-Informed Neural Networks (PINNs) provide a flexible, mesh-free approach to solving differential equations. While their implementation is straightforward and they hold great potential for high-dimensional, inverse, and nonlinear problems, training remains challenging due to their sensitivity to weight initialization, hyperparameter selection, scalability issues, spectral bias, and ill-conditioning. This talk explores how overlapping domain decomposition (DD) techniques can improve the convergence and efficiency of PINNs.
Different strategies for integrating DD with PINNs are investigated. First, classical PINNs are enhanced using multilevel DD-based architectures to improve performance. Additionally, this strategy is combined with multifidelity stacking PINNs for time-dependent problems, demonstrating clear improvements over reference results without DD. Second, randomized neural networks are considered, where hidden layer weights are randomly initialized and fixed, reducing the problem to a linear least-squares formulation for linear differential operators. In this setting, DD-based architectures, combined with overlapping Schwarz preconditioning, accelerate convergence. Finally, improvements in physics-informed neural operators through DD-based architectures are explored, enabling the efficient approximation of solution operators for parameterized problems.
Numerical experiments on multiscale and wave phenomena demonstrate the effectiveness of DD techniques in PINNs. These results highlight the potential of DD methods to significantly enhance computational efficiency and accuracy in physics-informed machine learning.
Seminar date and time: March 14, 2025. 10 AM ET.
Zoom recording here.
Prof. Souvik Chakraborty
Neural Operators and Beyond: The Changing Landscape of Scientific Computing
This is a special hybrid seminar with Professor Chakraborty presenting in-person at E339 Westgate Building, Pennsylvania State University, University Park.
Seminar date and time: March 10, 2025. 10.30 AM ET.
Zoom recording here.
Dr. Kamyar Azizzadenesheli
Kamyar Azizzadenesheli has been a Research Staff at Nvidia since the Summer of 2022. Prior to his role at Nvidia, he was an assistant professor at Purdue University, Department of Computer Science, from Fall 2020 to Fall 2022. Prior to his faculty position, he was at the California Institute of Technology (Caltech) as a Postdoctoral Scholar, Special Student Researcher, and Visiting Student Researcher. Azizzadenesheli is a former Visiting Student Researcher at Stanford University and a researcher at Simons Institute, UC Berkeley. In addition, he is a former guest researcher at INRIA France (SequeL team), as well as a visitor at Microsoft Research Lab, New England, and New York. He received his Ph.D. at the University of California, Irvine.
ML on Functions: Neural Operators for Science and Engineering
The fabric of our daily lives, from weather forecasts to stock market predictions, from the aerodynamics of vehicles to the development of innovative materials, and even in the realms of medicine and space exploration, relies heavily on scientific and engineering computing. While super-intelligence in AI has made significant strides in language processing, visual recognition, and audio analysis, its potential in the vast domain of natural sciences and engineering remains largely untapped. In this talk, we delve into the evolution of AI from neural networks to neural operators, unlocking new frontiers in advanced scientific computing. Join this talk as we explore how these cutting-edge technologies are revolutionizing our approach to understanding and modeling the complexities of the natural world, paving the way for groundbreaking discoveries and innovations.
Seminar date and time: March 7, 2025. 10 AM ET.
Zoom recording here (part 1) and here (part 2).
Dr. Youngsoo Choi
Youngsoo is a staff scientist at LLNL’s CASC group, where he develops efficient foundation models for computational science. His research focuses on creating surrogates and reduced-order models to accelerate time-critical simulations in areas such as inverse problems, design optimization, and uncertainty quantification. He has pioneered advanced ROM techniques—including machine learning-based nonlinear manifolds, space-time ROMs, component-wise ROM optimization, and latent space dynamics identification—and currently leads the libROM team in data-driven surrogate modeling. His contributions extend to open source projects such as libROM, pylibROM, LaghosROM, ScaleupROM, LaSDI, NM-ROM, DD-NM-ROM, and gappyAE. Youngsoo earned his BS from Cornell and his PhD from Stanford, and he was a postdoc at Sandia and Stanford before joining LLNL in 2017.
Foundation models in computational science
Computational science is at the forefront of modern technology, enabling groundbreaking simulations—from quantum molecular dynamics and magnetic fusion to complex wave phenomena. These advances have ushered in the era of digital twins, accelerating the design, build, test, and learn (DBTL) process across diverse applications. Yet even the most advanced simulations remain computationally expensive, often pushing high-performance computing systems to their limits. Fortunately, machine learning (ML) and artificial intelligence (AI) offer promising strategies to enhance simulation speed without sacrificing accuracy. In this talk, I will introduce several foundation models designed to accelerate computational simulations. I will critically evaluate these models, distinguishing between those that offer nice, yet incomplete gains and those that deliver genuine, comprehensive performance improvements. A highlight of the discussion will be the data-driven finite element method (DD-FEM), a robust foundation model whose effectiveness I will demonstrate across three applications: lattice-type structure design, steady Navier–Stokes porous media flow, and time-dependent 2D Burgers advective flow with multiple disturbances, achieving roughly 1000x speed-up and 100x scale-up with a relative error of O(1%).
Seminar date and time: February 28, 2025. 10 AM ET.
Zoom recording here.
Dr. Katherine Asztalos
Katherine Asztalos is a research scientist at Argonne National Laboratory. Her work focuses on computational fluid dynamics of multiphase and aerodynamic flowfields, optimization of dynamical systems through reduced-dimensionality models, and high-performance computing to solve complex engineering challenges in transportation. She received her Ph.D. in Mechanical and Aerospace Engineering from the Illinois Institute of Technology, where her thesis explored the aerodynamic response to impulsive active flow control. Asztalos is passionate about using advanced simulation techniques to enhance the design and efficiency of transportation systems, particularly in the aviation sector.
Interpretable Reduced-order Models for Engineering Applications
The development of reduced-order models has gained significant traction in the field of fluid mechanics. These models offer a means to represent complex systems within an optimally reduced-dimensional space, enabling the creation of predictive models with lower computational costs and facilitating the discovery of underlying dynamical insights. Reduced-order models can be constructed using either a physics-based approach, which leverages governing equations, or data-driven methods, which utilize available training data for the system. These models are particularly well-suited for control applications, where their reduced complexity allows for real-time implementation and efficient control strategy development. This presentation explores the application of both data-driven and physics-based modeling techniques to effectively capture and predict complex fluid dynamical behaviors, with particular emphasis on challenges pertinent to engineering applications, especially in the realm of unsteady aerodynamics.
Seminar date and time: February 21, 2025. 10 AM ET.
Zoom recording here.
Prof. Elizabeth Qian
Elizabeth Qian is an Assistant Professor at Georgia Tech jointly appointed in the School of Aerospace Engineering and the School of Computational Science and Engineering. Her interdisciplinary research develops new computational methods to enable engineering design and decision-making for complex systems, with special expertise in model reduction, scientific machine learning, and multifidelity methods. Recent awards include a 2024 Air Force Young Investigator award and a 2023 Hans Fischer visiting fellowship at the Technical University of Munich. Prior to joining Georgia Tech, she was a von Karman Instructor at Caltech in the Department of Computing and Mathematical Sciences. She earned her SB, SM, and PhD degrees from MIT.
Multifidelity linear regression for scientific machine learning from scarce data
Machine learning (ML) methods have garnered significant interest as potential methods for learning surrogate models for complex engineering systems for which traditional simulation is expensive. However, in many scientific and engineering settings, training data are scarce due to the cost of generating data from traditional high-fidelity simulations. ML models trained on scarce data have high variance and are sensitive to vagaries of the training data set. We propose a new multifidelity training approach for scientific machine learning that exploits the scientific context where data of varying fidelities and costs are available; for example high-fidelity data may be generated by an expensive fully resolved physics simulation whereas lower-fidelity data may arise from a cheaper model based on simplifying assumptions. We use the multifidelity data to define new multifidelity Monte Carlo estimators for the unknown parameters of linear regression models, and provide theoretical analyses that guarantee accuracy and improved robustness to small training budgets. Numerical results show that multifidelity learned models achieve order-of-magnitude lower expected error than standard training approaches when high-fidelity data are scarce.
Seminar date and time: February 14, 2025. 10 AM ET.
Zoom recording here.
Sebastien Andre-Sloan
Sebastien Andre-Sloan is a 1st year PhD student in A.I. at the University of Manchester, where he earned his B.S.c in Computer Science and Mathematics. His work focuses on scientific applications in deep-learning theory, especially involving PINNs or DeepONets. He has shown DeepONet applications in air-foil experiments and presented them at ASDAI 2024. His current work is on discovering bounds for PINNs. Sebastien is advised by Dr. Anirbit Mukherjee, Dr. Matthew Colbrook and Prof. Alex Frangi.
Bigger PINNs are Needed for Noisy PDE Training
Physics-Informed Neural Networks (PINNs) are increasingly used to solve various partial differential equations (PDEs), especially in high dimensions. In real-world applications, data samples are noisy, making it essential to understand the conditions under which a predictor can achieve a small empirical risk. In this work, we present a first-of-its-kind lower bound on the size of neural networks required for the supervised PINN empirical risk to fall below the variance of noisy supervision labels. Specifically, we show that to achieve low training error on Ns data points, the number of parameters d_N must satisfy d_N log d_N ≳ N_s. We show that a similar constraint applies to the fully unsupervised PINN setting when boundary labels are sampled noisily. Consequently, increasing the number of noisy supervision labels alone does not provide a “free lunch” in reducing empirical risk. We investigate PINNs applied to the Hamilton–Jacobi–Bellman (HJB) PDE as a case study to understand the role of the number of trainable parameters in minimizing the associated PINN empirical risk. Our findings lay the groundwork for a program on rigorously quantifying parameter requirements for effective PINN training under noisy conditions. This work was done in collaboration with Dr. Anirbit Mukherjee, University of Mancheseter and Dr. Matthew Colbrook, DAMTP, University of Cambridge
Seminar date and time: February 7, 2025. 10 AM ET.
Zoom recording here.
Dr. Shivam Barwey
Shivam Barwey is the AETS named fellow at Argonne National Laboratory. He received his PhD in Aerospace Engineering at the University of Michigan. His research interests are scientific machine learning, computational fluid dynamics (CFD), and high-performance computing (HPC), with a focus on numerical modeling and simulation of high-speed reacting flows.
Scalable and interpretable scientific machine learning for modeling complex fluid flows
Scientific machine learning (SciML) has emerged as a promising field for modeling complex fluid flows, particularly those encountered in advanced power generation and propulsion concepts. However, these and other aerospace applications impose stringent requirements on SciML models, necessitating solutions that address both complex physics and geometry. To ensure the effectiveness of new models for power and propulsion applications, machine learning methods must be (a) physically interpretable, (b) inherently compatible with complex data representations, and (c) scalable. In this talk, we introduce new SciML models for fluid flows based on interpretable geometric deep learning to address these requirements, and discuss how hardware-oriented machine learning strategies can enable transformative advancements in simulation capabilities for high-speed reacting flows.
Seminar date and time: January 24, 2025. 10 AM ET.
Zoom recording here.
Professor Kai Fukami
Kai Fukami is an Associate Professor in the Department of Aerospace Engineering at Tohoku University, Japan. He received his B. Eng. (Mar 2018) and M. Eng. (Sep 2020) degrees from Keio University (Japan). He received Ph.D. at University of California, Los Angeles (UCLA). in Feb 2024. After spending ten months at UCLA as a postdoc research associate, he then joined the Tohoku University as an Associate Professor in Jan 2025. His research is focused on developing physics-inspired data-driven techniques for turbulent flow analyses, leveraging computational fluid dynamics, super/unsupervised machine learning, and complex network theory. He was selected as a 2022 Amazon Fellow by the UCLA-Amazon Science Hub for Humanity and Artificial Intelligence.
Taming highly unsteady flows with data-oriented approaches: progress and outlook
In this talk, we discuss how highly unsteady flows can be analyzed in a data-driven manner from the aspect of global field reconstruction, reduced-complexity modeling, and flow control. In particular, our focus is on flows exhibiting unsteadiness with larger amplitudes than those often examined in traditional fluid dynamics. We first perform global field reconstruction from sparse sensors through the lens of generalized super-resolution analysis. This talk covers not only fundamental applications of fluid-flow super resolution but also practical uses for moving sensor conditions and industrial turbulence. To perform flow control leveraging the reconstructed fields from sparse sensors, we then aim to construct a control strategy of flows in a low-order subspace identified by nonlinear machine-learning-based data compression. Although it is generally challenging to analyze the nonlinear, transient nature of highly unsteady flows with conventional linear techniques, we reveal that the underlying physics of a collection of time-varying vortical flows in a high-dimensional space can be expressed on a low-rank manifold leveraging the present data-driven compression. It is also demonstrated that efficient control strategies can be derived at a minimal cost with the assistance of phase-amplitude reduction on the discovered manifold. Toward the end of the talk, based on the current findings, we discuss how we should be mindful of preparing training data of turbulent flows for data-driven studies in a smart manner.
Seminar date and time: January 17, 2025. 10 AM ET.
Zoom recording available here.
Dr. Panos Stinis
Panos Stinis specializes in scientific computing with application interests in model reduction of complex systems, multiscale modeling, uncertainty quantification, and machine learning. He studied aeronautical engineering at the Technical University of Athens, Greece. He earned his PhD in applied mathematics in 2003, from Columbia University in New York, in model reduction. He began his career at Lawrence Berkeley National Laboratory and the Stanford Center for Turbulence Research, where he worked on applying model reduction methods to hyperbolic systems and in developing techniques for locating and tracking singularities of partial differential equations. In 2008, he became a faculty member at the Mathematics Department at the University of Minnesota, where he worked on renormalization, mesh refinement, particle filtering and optimization. He moved to the Pacific Northwest National Laboratory in 2014, where he is currently leading the Computational Mathematics group.
When big neural networks are not enough: physics, multifidelity and kernels
Modern machine learning has shown remarkable promise in multiple applications. However, brute force use of neural networks, even when they have huge numbers of trainable parameters, can fail to provide highly accurate predictions for problems in the physical sciences. We present a collection of ideas about how enforcing physics, exploiting multifidelity knowledge and the kernel representation of neural networks can lead to significant increase in efficiency and/or accuracy. Various examples are used to illustrate the ideas.
Seminar date and time: December 13, 2024. 10 AM ET.
Recording available here.
Dr. Alec Linot
Alec Linot works as a postdoctoral researcher with Professor Kunihiko (Sam) Taira in the Mechanical and Aerospace Engineering Department at UCLA. Prior to working with Prof. Taira, he received a BS in Chemical Engineering from Kansas State University, and a Ph.D in Chemical and Biological Engineering from the University of Wisconsin–Madison with Michael D. Graham. His research focuses on the low-order modeling and control of complex fluid flows, through the development and use of methods in machine learning and stability analysis.
Hierarchical equivariant graph neural networks for forecasting collective motion in vortex clusters and microswimmers
Data-driven forecasting of collective dynamics is a challenging problem because short- and long-range interactions combine in complex ways to influence global system properties. Graph neural networks (GNNs) are a powerful class of methods that can be used for modeling collective motion, but GNNs struggle when graphs become too large and their effectiveness at capturing the aforementioned global properties remains unclear. Here we show that by constructing hierarchical and equivariant GNNs, we accurately predict local and global behavior in systems with collective motion. In particular, we validate this approach on simulated datasets of clusters of point vortices and of microswimmers. For the point vortices, we define a local graph of vortices within a cluster and a global graph of interactions between clusters. This approach drastically improves the tracking capabilities over a fully connected graph. Then, by incorporating equivariance to rotations and translations, this method successfully conserves the Hamiltonian for long periods. For the microswimmers, we define a local graph around each microswimmer and a global graph that groups the long-range interactions of many microswimmers. We also find that this approach – along with accounting for equivariance – allows us to both accurately predict short-time dynamics, and predict long-time statistics. Notably, this method effectively predicts the phase transition from aggregation to swirling in this microswimmer system.
Seminar date and time: December 6, 2024. 10 AM ET.
Zoom recording here.
Professor Bernat Font
Dr. Bernat Font is an Assistant Professor at the Faculty of Mechanical Engineering in TU Delft. He obtained his PhD from the University of Southampton (UK) in the topic of dimensional reduction and turbulence modelling of flow past slender geometries. As a postdoctoral researcher at the Barcelona Supercomputing Center (Spain), he specialized in turbulence modelling for high-order methods and active flow control using reinforcement learning. Dr. Font's main research interests are in the combination of numerical methods and data-driven models, high-performance computing, and optimization.
Perspective and applications of data-informed computational fluid dynamics
The widespread adoption of data-driven modeling tools and advances in hardware architectures across the scientific computing community have also shaped the future of CFD solvers. In this context, we will review the opportunities and challenges arising from the use of ML models in scale-resolving turbulent flow simulations. As examples, the cases of data-driven turbulence modelling, and active flow control using reinforcement learning will be discussed. We will also present our Julia GPU-accelerated CFD solver, WaterLily, and discuss future research directions.
Seminar date and time: November 29, 2024. 10 AM ET.
See recording here
Dr. Ali Siahkoohi
Ali Siahkoohi is a Simons Postdoctoral Fellow in the Department of Computational Applied Mathematics & Operations Research at Rice University, jointly hosted by Dr. Maarten V. de Hoop and Dr. Richard G. Baraniuk. He received his Ph.D. in Computational Science and Engineering from Georgia Institute of Technology in 2022. His research lies at the intersection of computational science and AI, focusing on designing scalable methods for quantifying uncertainty in AI models, with a broader goal of enhancing AI reliability.
Mitigating biases in self-consuming generative models
We highlight the risks of the current industrial AI practices involving training large-scale generative models on vast amounts of data scraped from the internet. This process unwittingly leads to training newer models on increasing amounts of AI-synthesized data that is rapidly proliferating online, a phenomenon we refer to as ``model autophagy'' (self-consuming models). We show that without a sufficient influx of fresh, real data at each stage of an autophagous loop, future generative models will inevitably suffer a decline in either quality (precision) or diversity (recall). To mitigate this issue and inspired by fixed-point optimization, we introduce a penalty to the loss function of generative models that minimizes discrepancies between the model's weights when trained on real versus synthetic data. Since computing this penalty would require training a new generative model at each iteration, we propose a permutation-invariant hypernetwork to make evaluating the penalty tractable by dynamically mapping data batches to model weights. This ensures scalability and seamless integration of the penalty term into existing generative modeling paradigms, mitigating biases associated with model autophagy. Additionally, this penalty improves the representation of minority classes in imbalanced datasets, which is a key step toward enhancing fairness in generative models.
Seminar date and time: November 22, 2024. 10 AM ET.
Recording available here
Vivek Oomen
Vivek Oomen is a 4th year PhD student at the School of Engineering at Brown University, advised by Prof. George Em Karniadakis. Before joining Brown, he completed his undergrad in Mechanical Engineering with an integrated Master's in Data Science from the Indian Institute of Technology Madras.
Improving Spectral Bias in Neural Operators with Diffusion Models
We integrate neural operators with diffusion models to address the spectral limitations of neural operators in surrogate modeling of turbulent flows. While neural operators offer computational efficiency, they exhibit deficiencies in capturing high-frequency flow dynamics, resulting in overly smooth approximations. To overcome this, we condition diffusion models on neural operators to enhance the resolution of turbulent structures. Our approach is validated for different neural operators on diverse datasets, including a high Reynolds number jet flow simulation and experimental Schlieren velocimetry. The proposed method significantly improves the alignment of predicted energy spectra with true distributions compared to neural operators alone. Additionally, proper orthogonal decomposition analysis demonstrates enhanced spectral fidelity in space-time. This work establishes a new paradigm for combining generative models with neural operators to advance surrogate modeling of turbulent systems, and it can be used in other scientific applications that involve microstructure and high-frequency content.
See our project page: https://vivekoommen.github.io/NO_DM/
Seminar date and time: November 15, 2024. 10 AM ET.
Seminar recording here.
Professor Somdatta Goswami
Dr. Somdatta Goswami is an Assistant Professor in the Department of Civil and Systems Engineering at Johns Hopkins University, with a joint appointment in the Department of Applied Mathematics and Statistics. Her research focuses on advancing methods to address long-time horizon challenges and the complexities of coupling scales within multiscale, multiphysics material modeling. Her group is actively engaged in developing AI-accelerated numerical simulations to significantly enhance both the efficiency and accuracy of these processes.
Physics-informed operator learning on latent spaces
Deep operator network (DeepONet) has demonstrated significant potential in solving partial differential equations (PDEs) by leveraging neural networks to learn mappings between function spaces. However, their performance deteriorates as the system size and complexity increase. Recent advancements with Latent DeepONet have shown promise in accelerating surrogate models for these complex systems by learning operators in low-dimensional latent spaces. Despite their potential, the Latent DeepONet architectures rely exclusively on data-driven training, necessitating large datasets and proving unsuitable for physics-informed training. To address these limitations, we introduce latent operator learning in a physics-informed framework, termed PI-Latent-DeepONet. Our method employs a two-stacked DeepONet framework: the first DeepONet learns the latent representations, while the second DeepONet recovers the solution in the original space. We demonstrate the effectiveness of the proposed framework in rapidly predicting solutions for high-dimensional PDEs.
Seminar date and time: November 8, 2024. 10 AM ET.
Recording available here.
Professor Ricardo Vinuesa
Dr. Ricardo Vinuesa is an Associate Professor at the Department of Engineering Mechanics, KTH Royal Institute of Technology in Stockholm. He is also Lead Faculty at the KTH Climate Action Centre. He studied Mechanical Engineering at the Polytechnic University of Valencia (Spain), and he received his PhD in Mechanical and Aerospace Engineering from the Illinois Institute of Technology in Chicago. His research combines numerical simulations and data-driven methods to understand, control and predict complex wall-bounded turbulent flows, such as the boundary layers developing around wings and urban environments. Dr. Vinuesa has received, among others, an ERC Consolidator Grant, the TSFP Kasagi Award, the MST Emerging Leaders Award, the Goran Gustafsson Award for Young Researchers, the IIT Outstanding Young Alumnus Award, the SARES Young Researcher Award and he leads several large Horizon Europe projects. He is also a member of the Young Academy of Science of Spain.
Coherent structures and reduced-order models for turbulence through deep learning
In this presentation we first use a framework for deep-learning explainability to identify the most important Reynolds-stress (Q) events in a turbulent channel (simulated via DNS) and a turbulent boundary layer (obtained experimentally). This objective way to assess importance reveals that the most important Q events are not the ones with the highest Reynolds shear stress. This framework is also used to identify completely new coherent structures, and we find that the most important coherent regions in the flow only have an overlap of 70% with the classical Q events. In the second part of the presentation we use beta variational autoencoders (beta VAEs) and transformers to build reduced-order models (ROMs) for turbulent flows. The beta VAEs enable a non-linear reduced-order representation of the flow while preserving disentanglement of the latent variables. The transformer yields excellent temporal predictions in the latent space, far superior than those possible with e.g. long short-term memory (LSTM) networks, thanks to the multi-scale characteristics captured by the attention mechanisms. A combination of explainability and autoencoders with causality, to further shed light on the most important flow mechanisms, is also presented.
Seminar date and time: November 1, 2024. 10 AM ET.
Dr. Troy Arcomano
Dr. Troy Arcomano is a postdoctoral fellow at Argonne National Lab working on machine learning applications for weather and climate in the environmental science division. During his time at ANL, he was the Argonne lead for several projects including a large collaboration to create a state-of-the-art foundation model for weather prediction. Troy received his PhD at Texas A&M University where he worked on developing machine learning applications for weather forecasting and investigated how machine learning could be used to improve climate models.
Email: tarcomano@anl.gov
The AI Revolution for Weather and Climate
Recently, advances in machine learning, hardware (e.g. GPUs/TPUs), and availability of high-quality data have set the stage for machine learning (ML) to tackle problems for weather and climate. This has led to a paradigm shift in operational weather forecasting, most evidently seen by the vast amount of resources being invested into AI models at the leading operational centers including NOAA, ECMWF, ECCC, and others. This has been motivated by the influx of deep learning-based models in the last 3 years for weather forecasting which have been demonstrated to have forecasting skill approaching or even exceeding the best available numerical weather prediction (NWP) models. These data-driven models include Graphcast, Fourcastnet, FuXi, Pangu-weather, ClimaX, Fengwu, and Stormer, each with vastly different training methods, machine learning architectures, and variables predicted.
In this seminar, we explore the rise of ML-based weather prediction and discuss research ongoing at Argonne that is helping revolutionize how weather and climate are researched using machine learning. Specifically, looking at: 1.) an Argonne-led project for medium-range weather forecasting called Stormer and 2.) one of the first systematic evaluations of machine learning-based emulators for climate research and several novel future research directions on creating the next generation of climate model emulators.
Seminar date and time: October 25, 2024. 10 AM ET.
Recording available here.