Presentation

<
GMT: A Geometric Multigrid Transformer Solver for Microstructure Homogenization
SessionNumerical Methods for Geometry Processing
DescriptionGMT is a Geometric Multigrid Transformer that significantly accelerates microstructure homogenization. Embedding Transformers into the multigrid hierarchy to capture multi-scale physical interactions delivers a 160x speedup over SOTA GPU solvers while maintaining engineering-grade accuracy at high resolutions. GMT enables real-time material screening, multi-scale simulations, and rapid inverse design of metamaterials.
Authors
Yu Xing
Shandong University
Yang Liu
Microsoft Research Asia
Tianyang Xue
The University of Hong Kong
Lin Lu
Shandong University
Event Type
Technical Paper
TimeWednesday, 22 July 20264:15pm - 4:25pm PDT
LocationRoom 403 B
Digital Library PDF
Session Time & Location
Sunday, 19 July 20266:00pm - 8:45pm PDTHall K
Wednesday, 22 July 20263:45pm - 5:35pm PDTRoom 403 B
Keywords
Animation
Artificial Intelligence/Machine Learning
Fabrication
Simulation
Registration Categories
Full Conference Supporter
Full Conference
Next PresentationNext PresentationRobust Biharmonic Skinning Using Geometric Fields
Similar Presentations
Shellular Metamaterial Design via Compact Electric Potential Parametrization
A Unified Homogenization Framework for Straight- and Curved-Crease Origami Materials
Fast VEM Fluid Simulation