Machine learning for modelling band structure properties

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F68407700%3A21230%2F25%3A00387035" target="_blank" >RIV/68407700:21230/25:00387035 - isvavai.cz</a>

Výsledek na webu

<a href="http://dx.doi.org/10.61782/fa.2025.0539" target="_blank" >http://dx.doi.org/10.61782/fa.2025.0539</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.61782/fa.2025.0539" target="_blank" >10.61782/fa.2025.0539</a>

Jazyk výsledku

angličtina

Název v původním jazyce

Machine learning for modelling band structure properties

Popis výsledku v původním jazyce

Although there are various analytical approaches and numerical methods for solving sonic crystal problems, analytical expressions for modelling the band structure properties are limited to a few special cases. The access to a numerical model offers a solid foundation for data-driven discovery. In our approach, we employed the Webster equation for the unit cell and Floquet-Bloch theory for periodic structures, with the waveguide parametrized by cubic splines. To extract analytical formulae linking the waveguide geometry to the corresponding dispersion relation, we applied methods of physics-informed machine learning, such as coordinate transformation and symbolic regression. These results provide a deeper understanding of the underlying principles and serve as an efficient alternative to computationally demanding numerical optimization. Moving toward a Schroedinger-like equation and parametrization by Gaussian curvature allows for a more multiphysical approach, yet it also presents challenges related to the feasibility limits of the geometry.

Název v anglickém jazyce

Machine learning for modelling band structure properties

Popis výsledku anglicky

Although there are various analytical approaches and numerical methods for solving sonic crystal problems, analytical expressions for modelling the band structure properties are limited to a few special cases. The access to a numerical model offers a solid foundation for data-driven discovery. In our approach, we employed the Webster equation for the unit cell and Floquet-Bloch theory for periodic structures, with the waveguide parametrized by cubic splines. To extract analytical formulae linking the waveguide geometry to the corresponding dispersion relation, we applied methods of physics-informed machine learning, such as coordinate transformation and symbolic regression. These results provide a deeper understanding of the underlying principles and serve as an efficient alternative to computationally demanding numerical optimization. Moving toward a Schroedinger-like equation and parametrization by Gaussian curvature allows for a more multiphysical approach, yet it also presents challenges related to the feasibility limits of the geometry.

Druh

D - Stať ve sborníku

CEP obor

—

OECD FORD obor

10307 - Acoustics

Projekt

—

Návaznosti

S - Specificky vyzkum na vysokych skolach

Rok uplatnění

2025

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název statě ve sborníku

Proceedings of the 11th Convention of the European Acoustics Association Forum Acusticum / EuroNoise 2025

ISBN

978-84-87985-35-5

ISSN

3005-7124

e-ISSN

3005-7124

Počet stran výsledku

3

Strana od-do

5287-5289

Název nakladatele

European Acoustics Association

Místo vydání

Madrid

Místo konání akce

Malaga

Datum konání akce

23. 6. 2025

Typ akce podle státní příslušnosti

WRD - Celosvětová akce

Kód UT WoS článku

—