The policy of the Summit is that the participants leave their polished talks
and reticence behind in favor of burning questions, open problems, new ideas,
and readiness to interact.
Topics for the 2025 meeting
- AMG for anisotropic problems, in particular for problems with high contrast
- Coarse-grid subspaces for multilevel machine learning, a simplified perspective targeting feasability
- FAS coarse-grid equations for machine learning, in particular for stochastic optimization
- Use of mass matrices in strength of connection for AMG
- AMG for systems of PDEs
- Use of Green's functions in strength of connection in AMG
- Regularization in machine learning
- H-Curl and H-Div AMG solvers
- Computation / use of Gauss-Newton in machine learning
- Graph Laplacians, graph neural networks, and how to carry out multilevel neural network training
- Elimination-based AMG, AMG from new approximate factorizations
- Machine learning: weights vs objective
- Diagonal rescaling to recover classic constant-like near nullspace
- AMG for Helmholtz
- Multilevel space-time discretizations, as it relates to parallel-in-time
- Tensor solvers
- Paradiag parallel-in-time solvers, what they do
|
