aiqbench analysis
This paper benchmarks how to learn port-Hamiltonian models, which describe physical systems through stored energy, dissipation and inputs, from data. It compares neural networks, analytic expressions and hybrids, fitted by either gradient-based optimisation or sparse regression, on simulated mechanical and electrical systems with and without noise. For ML readers, the useful signal is the accuracy-versus-compute trade-off between learned and symbolic components in physics-structured models, though the results are limited to simulated benchmarks.
Comparing model representations and fitting methods for port-Hamiltonian system identification
arXiv:2611.05031v1 Announce Type: cross Abstract: Port-Hamiltonian models describe physical systems by separating stored energy, dissipation and external inputs. Learning such models from data requires a choice of representation and fitting method. We compare using neural networks, analytic expressions and combinations of the two for representing the different components of the port-Hamiltonian system, and compare fitting the models with gradient-based optimisation or sparse regression. Testing on simulations of mechanical and electrical systems, with and without added noise, we present prediction accuracy and computational cost of the different approaches.
Abstract:Port-Hamiltonian models describe physical systems by separating stored energy, dissipation and external inputs. Learning such models from data requires a choice of representation and fitting method. We compare using neural networks, analytic expressions and combinations of the two for representing the different components of the port-Hamiltonian system, and compare fitting the models with gradient-based optimisation or sparse regression. Testing on simulations of mechanical and electrical systems, with and without added noise, we present prediction accuracy and computational cost of the different approaches.
Systems and Control (eess.SY); Machine Learning (cs.LG)
arXiv:2611.05031 [eess.SY]
(or arXiv:2611.05031v1 [eess.SY] for this version)
https://doi.org/10.48550/arXiv.2611.05031 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 180 words