Path sampling of recurrent neural networks by incorporating known physics

TL;DR

This paper introduces a path sampling method based on Maximum Caliber to incorporate prior physical knowledge into recurrent neural networks, enhancing their modeling of dynamical systems across various scientific domains.

Contribution

The authors present a novel approach to embed thermodynamic and kinetic constraints into recurrent neural networks using Maximum Caliber, applicable to diverse physical and social science time series.

Findings

Successfully applied to molecular dynamics and quantum system simulations.

Enables inclusion of prior knowledge into RNNs without retraining from scratch.

Generalizable to other generative models and time series data.

Abstract

Recurrent neural networks have seen widespread use in modeling dynamical systems in varied domains such as weather prediction, text prediction and several others. Often one wishes to supplement the experimentally observed dynamics with prior knowledge or intuition about the system. While the recurrent nature of these networks allows them to model arbitrarily long memories in the time series used in training, it makes it harder to impose prior knowledge or intuition through generic constraints. In this work, we present a path sampling approach based on principle of Maximum Caliber that allows us to include generic thermodynamic or kinetic constraints into recurrent neural networks. We show the method here for a widely used type of recurrent neural network known as long short-term memory network in the context of supplementing time series collected from different application domains. These include classical Molecular Dynamics of a protein and Monte Carlo simulations of an open quantum system continuously losing photons to the environment and displaying Rabi oscillations. Our method can be easily generalized to other generative artificial intelligence models and to generic time series in different areas of physical and social sciences, where one wishes to supplement limited data with intuition or theory based corrections.