Differentiable Physics-informed Graph Networks

TL;DR

This paper introduces Differentiable Physics-informed Graph Networks (DPGN), a novel architecture that embeds implicit physics knowledge into neural networks to improve climate prediction and other applications.

Contribution

The paper presents a new architecture, DPGN, that incorporates domain expert physics knowledge into neural networks via latent space, enhancing predictive performance.

Findings

Climate prediction accuracy is significantly improved.

DPGN effectively supports inductive learning.

The model enables accurate multistep predictions.

Abstract

While physics conveys knowledge of nature built from an interplay between observations and theory, it has been considered less importantly in deep neural networks. Especially, there are few works leveraging physics behaviors when the knowledge is given less explicitly. In this work, we propose a novel architecture called Differentiable Physics-informed Graph Networks (DPGN) to incorporate implicit physics knowledge which is given from domain experts by informing it in latent space. Using the concept of DPGN, we demonstrate that climate prediction tasks are significantly improved. Besides the experiment results, we validate the effectiveness of the proposed module and provide further applications of DPGN, such as inductive learning and multistep predictions.