Single-layer Laguerre neural network model for solving Lane-Emden-Fowler type equations

Creative Commons License

Kurmanç S. N., ÖZAVŞAR M.

Mathematical Communications, vol.30, no.2, pp.243-258, 2025 (SCI-Expanded, Scopus) identifier identifier

  • Publication Type: Article / Article
  • Volume: 30 Issue: 2
  • Publication Date: 2025
  • Doi Number: 10.64785/mc.30.2.8
  • Journal Name: Mathematical Communications
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Academic Search Premier, zbMATH, Directory of Open Access Journals
  • Page Numbers: pp.243-258
  • Keywords: Adam Optimization, Chebyshev neural network, Laguerre neural network, Lane-Emden–Fowler equation, Legendre neural network, single-layer functional link artificial neural network
  • Yıldız Technical University Affiliated: Yes

Abstract

In this study, a single-layer functional link artificial neural network (FLANN) model based on Laguerre polynomials, referred to as the Laguerre Neural Network (LgNN), is utilized to solve second-order linear and non-linear equations of Lane-Emden-Fowler type. Using this model, in which the hidden layers are replaced with Laguerre polynomials, we initially expand the input patterns corresponding to a given set of nonlinear Lane-Emden-Fowler type equations. Subsequently, the network parameters are adjusted through an unsupervised error backpropagation algorithm that employs Adam optimization. Consequently, we compare the LgNN results with those obtained by other FLANN-based models, namely the Chebyshev Neural Network (ChNN) and the Legendre Neural Network (LeNN), by solving some initial value problems of Lane-Emden–Fowler type equations.