Neural network method for time-space fractional-order elliptic partial differential equations
Physica Scripta · 05.06.2026 · 101 (22) · 225217
DOI: 10.1088/1402-4896/ae73fa ↗Kurz gefasst
Fachartikel in der Zeitschrift Physica Scripta, online erschienen am 05.06.2026. Verfasst von Mahmut Modanli und 1 weiteren Autor:innen. Die Arbeit nennt 28 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
In this article, the application of neural network-based methods for solving time-space fractional-order elliptic partial differential equations is presented. In order to solve the time-space fractional elliptic partial differential equation as accurately as possible, a non-traditional numerical approach, namely the neural network method (NNM), is proposed. Neural networks have been actively used in recent years to solve partial differential equations and to approximate their solutions. In most cases, these networks provide an approximate function representing the solution to the differential equation. The neural network learns the fractional orders of the derivatives, thereby increasing the accuracy of the solution. In computing fractional derivatives, neural networks have more flexible and powerful modeling capabilities compared to traditional methods. The proposed approach employs shifted Legendre polynomials and Caputo derivatives. The NNM is applied to the elliptic partial differential equation for fractional-order values of α = 1.5 , 1.75 and 1.9 . The Adam optimization algorithm is used in training the deep network due to its efficiency. The neural network solutions corresponding to each fractional-order value are compared with the analytical solution. The results are supported by tables, error metrics, and graphical illustrations (line plots and contour plots), showing good agreement with the analytical solutions. In the proposed NNM, it is observed that as the value of α ∈ ( 1 , 2 ] approaches 2, the approximate solution converges to the analytical solution.
Zitieren
APA
Modanli, M., & Coşkun, S. (2026). Neural network method for time-space fractional-order elliptic partial differential equations. Physica Scripta, 101(22), 225217. https://doi.org/10.1088/1402-4896/ae73fa
Vancouver
Modanli M, Coşkun S. Neural network method for time-space fractional-order elliptic partial differential equations. Physica Scripta. 2026;101(22):225217. doi:10.1088/1402-4896/ae73fa
BibTeX
@article{Modanli2026,
author = {Modanli, Mahmut and Coşkun, Sümeyye},
title = {Neural network method for time-space fractional-order elliptic partial differential equations},
journal = {Physica Scripta},
year = {2026},
volume = {101},
number = {22},
pages = {225217},
doi = {10.1088/1402-4896/ae73fa}
}
EUVIMED