English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Tools annual

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Recent advances in deep learning for physics have focused on discoveringshared representations of target systems by incorporating physics priors orinductive biases into neural networks. While effective, these methods arelimited to the system domain, where the type of system remains consistent andthus cannot ensure the adaptation to new, or unseen physical systems governedby different laws. For instance, a neural network trained on a mass-springsystem cannot guarantee accurate predictions for the behavior of a two-bodysystem or any other system with different physical laws. In this work, we takea significant leap forward by targeting cross domain generalization within thefield of Hamiltonian dynamics. We model our system with a graph neural networkand employ a meta learning algorithm to enable the model to gain experienceover a distribution of tasks and make it adapt to new physics. Our approachaims to learn a unified Hamiltonian representation that is generalizable acrossmultiple system domains, thereby overcoming the limitations of system-specificmodels. Our results demonstrate that the meta-trained model not only adaptseffectively to new systems but also captures a generalized Hamiltonianrepresentation that is consistent across different physical domains. Overall,through the use of meta learning, we offer a framework that achieves crossdomain generalization, providing a step towards a unified model forunderstanding a wide array of dynamical systems via deep learning.

Towards Cross Domain Generalization of Hamiltonian Representation via Meta Learning