PRIMER ON SOLVING DIFFERENTIAL EQUATIONS USING MACHINE LEARNING TECHNIQUES

Tamil Arasan Bakthavatchalam, Selvakumar Murugan, Suriyadeepan Ramamoorthy, Malaikannan Sankarasubbu, Radha Ramaswamy, Vijayalakshmi Sethuraman, Boris A. Malomed

Research output: Contribution to journalArticlepeer-review

Abstract

Machine Learning (ML) has shown a substantial impact on computational sciences in recent years. The adaptation of ML techniques to deal with various systems in physical sciences has gained ground in addition to the existing numerical methods. In this work, we introduce the readers to machine learning with special reference to Artificial Neural Networks (ANNs) that can solve ordinary differential equations (ODEs) and partial differential equations (PDEs) including those which are subject to specific symmetries. This paper will be helpful for graduate and undergraduate students as an introductory material to early career researchers interested in applying ML techniques to solve problems in computational sciences. In particular, we choose elementary differential equations that describe systems from various fields of science to illustrate the proficiency of ANNs to capture the regularities that underlie such systems in the hope of adding ML techniques to the physicists’ toolbelt.

Original languageEnglish
Article number113
JournalRomanian Reports in Physics
Volume74
Issue number2
StatePublished - 2022

Keywords

  • Deep Learning
  • Differential Equations
  • Differential Programming
  • Machine Learning
  • Mathematical Modelling
  • Neural Networks
  • Nonlinear Dynamics
  • Numerical Methods

Fingerprint

Dive into the research topics of 'PRIMER ON SOLVING DIFFERENTIAL EQUATIONS USING MACHINE LEARNING TECHNIQUES'. Together they form a unique fingerprint.

Cite this