Physics-enriched neural solvers for transient ice-flow simulation
It is shown that supplying the network with inexpensive input fields derived from low-order ice-flow balances improves this online solver, and suggests that enriching a neural solver's inputs with reduced-order physics can make repeated higher-order solves much cheaper, with no training data and no offline training.
Thomas Gregov, S. Rosier, Brandon D. Finley et al.
· 0 citations