Latent‐space Dynamics for Reduced Deformable Simulation
Authors:
- 1 May 2019
- journal article
- research article
- Published by Wiley in Computer Graphics Forum
- Vol. 38 (2) , 379-391
Abstract
We propose the first reduced model simulation framework for deformable solid dynamics using autoencoder neural networks. We provide a data‐driven approach to generating nonlinear reduced spaces for deformation dynamics. In contrast to previous methods using machine learning which accelerate simulation by approximating the time‐stepping function, we solve the true equations of motion in the latent‐space using a variational formulation of implicit integration. Our approach produces drastically smaller reduced spaces than conventional linear model reduction, improving performance and robustness. Furthermore, our method works well with existing force‐approximation cubature methods.Keywords
This publication has 48 references indexed in Scilit:
- Solid simulation with oriented particlesACM Transactions on Graphics, 2011
- Real-time large-deformation substructuringACM Transactions on Graphics, 2011
- Example-based elastic materialsACM Transactions on Graphics, 2011
- Skipping steps in deformable simulation with online model reductionACM Transactions on Graphics, 2009
- Numerical coarsening of inhomogeneous elastic materialsACM Transactions on Graphics, 2009
- Optimizing cubature for efficient integration of subspace deformationsACM Transactions on Graphics, 2008
- Precomputed acoustic transferACM Transactions on Graphics, 2006
- Model reduction for real-time fluidsACM Transactions on Graphics, 2006
- Dimensional model reduction in non‐linear finite element dynamics of solids and structuresInternational Journal for Numerical Methods in Engineering, 2001
- Good vibrations: modal dynamics for graphics and animationACM SIGGRAPH Computer Graphics, 1989