GeoHNN combines SPD-constrained inertia matrices with a biorthogonal autoencoder to learn reduced Hamiltonian dynamics, reporting lower energy drift and trajectory error on oscillator, pendulum, two-body, and cloth systems.
Model reduction on manifolds: A differential geometric framework, 2024
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
GeoHNNs: Geometric Hamiltonian Neural Networks
GeoHNN combines SPD-constrained inertia matrices with a biorthogonal autoencoder to learn reduced Hamiltonian dynamics, reporting lower energy drift and trajectory error on oscillator, pendulum, two-body, and cloth systems.