Data-driven model reconstruction for nonlinear wave dynamics

The use of machine learning to predict wave dynamics is a topic of growing interest, but commonly used deep-learning approaches suffer from a lack of interpretability of the trained models. Here, we present an interpretable machine learning framework for analyzing the nonlinear evolution dynamics of...

Full description

Saved in:
Bibliographic Details
Main Authors: Ekaterina Smolina, Lev Smirnov, Daniel Leykam, Franco Nori, Daria Smirnova
Format: Article
Language:English
Published: American Physical Society 2025-06-01
Series:Physical Review Research
Online Access:http://doi.org/10.1103/2jh8-p5y2
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849425080592891904
author Ekaterina Smolina
Lev Smirnov
Daniel Leykam
Franco Nori
Daria Smirnova
author_facet Ekaterina Smolina
Lev Smirnov
Daniel Leykam
Franco Nori
Daria Smirnova
author_sort Ekaterina Smolina
collection DOAJ
description The use of machine learning to predict wave dynamics is a topic of growing interest, but commonly used deep-learning approaches suffer from a lack of interpretability of the trained models. Here, we present an interpretable machine learning framework for analyzing the nonlinear evolution dynamics of optical wave packets in complex wave media. We use sparse regression to reduce microscopic discrete lattice models to simpler effective continuum models, which can accurately describe the dynamics of the wave packet envelope. We apply our approach to valley-Hall domain walls in honeycomb photonic lattices of laser-written waveguides with Kerr-type nonlinearity and different boundary shapes. The reconstructed equations accurately reproduce the linear dispersion and nonlinear effects, including self-steepening and self-focusing. This scheme is proven free of the a priori limitations imposed by the underlying hierarchy of scales traditionally employed in asymptotic analytical methods. It represents a powerful interpretable machine learning technique of interest for advancing design capabilities in photonics and framing the complex interaction-driven dynamics in various topological materials.
format Article
id doaj-art-07eca312858843aeb680a6dca8d3b98a
institution Kabale University
issn 2643-1564
language English
publishDate 2025-06-01
publisher American Physical Society
record_format Article
series Physical Review Research
spelling doaj-art-07eca312858843aeb680a6dca8d3b98a2025-08-20T03:29:53ZengAmerican Physical SocietyPhysical Review Research2643-15642025-06-017202331410.1103/2jh8-p5y2Data-driven model reconstruction for nonlinear wave dynamicsEkaterina SmolinaLev SmirnovDaniel LeykamFranco NoriDaria SmirnovaThe use of machine learning to predict wave dynamics is a topic of growing interest, but commonly used deep-learning approaches suffer from a lack of interpretability of the trained models. Here, we present an interpretable machine learning framework for analyzing the nonlinear evolution dynamics of optical wave packets in complex wave media. We use sparse regression to reduce microscopic discrete lattice models to simpler effective continuum models, which can accurately describe the dynamics of the wave packet envelope. We apply our approach to valley-Hall domain walls in honeycomb photonic lattices of laser-written waveguides with Kerr-type nonlinearity and different boundary shapes. The reconstructed equations accurately reproduce the linear dispersion and nonlinear effects, including self-steepening and self-focusing. This scheme is proven free of the a priori limitations imposed by the underlying hierarchy of scales traditionally employed in asymptotic analytical methods. It represents a powerful interpretable machine learning technique of interest for advancing design capabilities in photonics and framing the complex interaction-driven dynamics in various topological materials.http://doi.org/10.1103/2jh8-p5y2
spellingShingle Ekaterina Smolina
Lev Smirnov
Daniel Leykam
Franco Nori
Daria Smirnova
Data-driven model reconstruction for nonlinear wave dynamics
Physical Review Research
title Data-driven model reconstruction for nonlinear wave dynamics
title_full Data-driven model reconstruction for nonlinear wave dynamics
title_fullStr Data-driven model reconstruction for nonlinear wave dynamics
title_full_unstemmed Data-driven model reconstruction for nonlinear wave dynamics
title_short Data-driven model reconstruction for nonlinear wave dynamics
title_sort data driven model reconstruction for nonlinear wave dynamics
url http://doi.org/10.1103/2jh8-p5y2
work_keys_str_mv AT ekaterinasmolina datadrivenmodelreconstructionfornonlinearwavedynamics
AT levsmirnov datadrivenmodelreconstructionfornonlinearwavedynamics
AT danielleykam datadrivenmodelreconstructionfornonlinearwavedynamics
AT franconori datadrivenmodelreconstructionfornonlinearwavedynamics
AT dariasmirnova datadrivenmodelreconstructionfornonlinearwavedynamics