Automated multi-objective search over hybrid neural-operator and classical iterative solver configurations finds meta-solvers that solve benchmark time-dependent nonlinear PDEs up to 15 times faster than bare Krylov methods.
Finite element procedures
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
math.NA 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Automatic discovery of optimal meta-solvers for time-dependent nonlinear PDEs
Automated multi-objective search over hybrid neural-operator and classical iterative solver configurations finds meta-solvers that solve benchmark time-dependent nonlinear PDEs up to 15 times faster than bare Krylov methods.