The SIAM International Meshing Roundtable Workshop 2024 is pleased to announce the following plenary talks to appear at this year’s conference.
- Dr. Steve Karman, Oak Ridge National Laboratory
Point Creation for Anisotropic Adaptive Mesh Generation - Prof. Mark S. Shephard, Rensselaer Polytechnic Institute
Unstructured Mesh Tools to Support Massively Parallel DOE Simulation Codes - Prof. Yusu Wang, University of California, San Diego
Graph learning models: theoretical understanding, limitations and enhancements
Dr. Steve Karman
Point Creation for Anisotropic Adaptive Mesh Generation
Abstract: Traditional unstructured mesh generation processes create points as the mesh evolves. The placement of the points is carefully chosen to improve mesh quality or resolve boundary layers. Point creation can be performed separate from the tessellation process. I will present an alternate approach that uses physics-based particle interactions to distribute points. These points have size and shape in the form of Riemannian Metric Tensors that can be manipulated to resolve boundary layers or adapt to Flowfield solutions. The tensors represent the size field for the mesh. An advancing front connection scheme generates triangles that enforce this size field. Several two-dimensional examples will be used to illustrate the quality and smoothness of the resulting mesh. Adaptive mesh refinement will be shown for inviscid and viscous applications.
Biography: Steve Karman is a Modeling & Simulation Engineer at Oak Ridge National Laboratory (ORNL) where he develops high order and adaptive mesh generation tools. He received his B.S. and M.Eng. in Aerospace Engineering from Texas A&M University and his Ph.D. in Aerospace Engineering from the University of Texas at Arlington.
Dr. Karman spent 20 years at General Dynamics/Lockheed Martin in the CFD group where he worked on structured and unstructured flow solvers. Steve transitioned to academia, working for 11 years at the University of Tennessee at Chattanooga where he led the development of mesh generation tools and taught graduate courses in mesh generation. The next 7 years he worked at Pointwise (Cadence Design Systems) performing applied research in mesh smoothing and high order mesh generation. In 2021 Dr. Karman joined the Multiphysics Modeling and Flows Group at ORNL to develop mesh generation tools to support their advanced, high order FEM analysis tools.
Dr. Karman is an Associate Fellow of American Institute of Aeronautics and Astronautics (AIAA). He has served on multiple technical committees with the AIAA including Applied Aerodynamics, Fluid Dynamics and Meshing, Visualization and Computational Environments (serving as chair). He received a best paper award from AIAA in 2006 for “Unstructured Viscous Layer Insertion Using Linear Elastic Smoothing”. Dr. Karman also served on the NATO/RTO technical team AVT-113 studying viscous vortical flows over F-16XL. And he has served on the Organizing Committee and the Steering Committee for International Meshing Roundtable.
Prof. Mark S. Shephard
Unstructured Mesh Tools to Support Massively Parallel DOE Simulation Codes
Abstract: The Department of Energy Office of Science Advance Scientific Computing Research supports the development of massively parallel simulation codes that address key areas of scientific investigation. As the ability to account for additional physics and geometric complexity has increased, a number of the simulation codes have moved to the use of unstructured mesh methods. This presentation will discuss unstructured mesh technologies being developed to support the needs of multiple fusion energy system and ice sheet modeling codes. The tools to be discussed include:
- Geometry clean-up and combination tools to construct analysis geometries of the desired level of fidelity.
- Procedures to generate and adapt meshes meeting specific simulation code requirements.
- An infrastructure for massively parallel particle in cell simulations on unstructured meshes that has been used for fusion plasma and impurity transport simulations, as well as material point methods used in ice sheet modeling.
- An infrastructure for the in-memory coupling of massively parallel simulation codes that employ unstructured and structured mesh methods.
Biography: Mark S. Shephard is the Samuel A. and Elisabeth C. Johnson, Jr. Professor of Engineering at Rensselaer Polytechnic Institute. He is the director of Rensselaer’s Scientific Computation Research Center. Dr. Shephard has made contributions to the areas of automatic mesh generation, automated and adaptive analysis methods, and parallel adaptive simulation technologies. He is a fellow and past president of the US Association for Computational Mechanics, and was recipient of the 1997 USACM Computational and Applied Sciences Award, and the 2011 John von Neumann metal; a fellow of the International Association for Computational Mechanics; and a fellow of ASME. Dr. Shephard was a co-founder of Simmetrix Inc., a computer-aided engineering company dedicated to producing the technologies and associated software components to enable simulation-based engineering.
Prof. Yusu Wang
Graph learning models: theoretical understanding, limitations and enhancements
Abstract: Graph data is ubiquitous in many application domains. The rapid advancements in machine learning also lead to many new graph learning frameworks, such as message passing (graph) neural networks (MPNNs), graph transformers and higher order variants. In this talk, I will describe some of our recent journey in attempting to provide better (theoretical) understanding of these graph learning models (e.g, their represetnation power and limitations in capturing long range interactions in graphs), the pros and cons of different models, and ways to further enhance them in practice. This talk is based on multiple joint work with various collaborators, whom I will mention in the talk.
Biography: Yusu Wang is currently Professor in the Halicioglu Data Science Institute at University of California, San Diego, where she also serves as Director for the NSF National AI Institute TILOS. She obtained her PhD degree from Duke University in 2004, and from 2004-2005, she was a post-doctoral fellow at Stanford University. Yusu Wang primarily works in the fields of Computational geometry, Computational and applied topology, and appliations to data analysis and machine learning. She received DOE Early Career Principal Investigator Award in 2006, and NSF Career Award in 2008. She currently serves on the Computational Geometry Steering Committee, as well as the AATRN Advisory Committee. She is also an Associate Editor for SIAM Journal on Computing (SICOMP) and Journal of Computational Geometry (JoCG).
