Machine learning active-nematic hydrodynamics

TL;DR

This paper demonstrates that neural networks can extract and predict hydrodynamic parameters and dynamics of active nematic systems directly from experimental image data, enabling AI-driven analysis of complex biological and physical systems.

Contribution

It introduces neural network methods to determine spatially varying hydrodynamic parameters and forecast chaotic dynamics from experimental images in active matter.

Findings

Neural networks successfully extract activity and elastic constants from experiments.

The approach predicts the evolution of active nematic systems from image sequences.

Method applies to microtubule-kinesin and actin-myosin experiments.

Abstract

Hydrodynamic theories effectively describe many-body systems out of equilibrium in terms of a few macroscopic parameters. However, such hydrodynamic parameters are difficult to derive from microscopics. Seldom is this challenge more apparent than in active matter where the energy cascade mechanisms responsible for autonomous large-scale dynamics are poorly understood. Here, we use active nematics to demonstrate that neural networks can extract the spatio-temporal variation of hydrodynamic parameters directly from experiments. Our algorithms analyze microtubule-kinesin and actin-myosin experiments as computer vision problems. Unlike existing methods, neural networks can determine how multiple parameters such as activity and elastic constants vary with ATP and motor concentration. In addition, we can forecast the evolution of these chaotic many-body systems solely from image-sequences of their past by combining autoencoder and recurrent networks with residual architecture. Our study paves the way for artificial-intelligence characterization and control of coupled chaotic fields in diverse physical and biological systems even when no knowledge of the underlying dynamics exists.