A CNN-LSTM model recovers the symbolic Hamiltonian from an image of its vector field on the plane, with 85 to 88 percent accuracy on a closed vocabulary of polynomial and trigonometric Hamiltonians.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
SymFlux: deep symbolic regression of Hamiltonian vector fields
A CNN-LSTM model recovers the symbolic Hamiltonian from an image of its vector field on the plane, with 85 to 88 percent accuracy on a closed vocabulary of polynomial and trigonometric Hamiltonians.