A physics-informed neural network for solving multi-term space-time fractional differential equations
Physica Scripta · 26.06.2026 · 101 (25) · 256004
DOI: 10.1088/1402-4896/ae7672 ↗Kurz gefasst
Fachartikel in der Zeitschrift Physica Scripta, online erschienen am 19.06.2026. Verfasst von Aditi Sharma und 3 weiteren Autor:innen. Die Arbeit nennt 54 Literaturangaben. Gefördert von: Ministry of Education, Government of India, Korea Environment Industry & Technology Institut, ANRF.
Abstract (Originalsprache)
In this study, we propose an attention-based physics-informed neural networks (PINNs) framework for solving multi-term Riesz–Caputo fractional advection-diffusion equations. The considered model generalizes the classical advection diffusion equation by replacing the first-order temporal derivative with the Caputo fractional derivative of order α ∈ ( 0 , 1 ] , and the first and second-order spatial derivatives with the Riesz fractional derivatives of orders γ 1 ∈ ( 0 , 1 ) and γ 2 ∈ ( 1 , 2 ] , respectively. Owing to the presence of multi-term fractional derivatives, conventional numerical methods often encounter challenges related to stability, computational complexity, and approximation accuracy. To address these difficulties, we propose a PINNs framework incorporating attention mechanisms together with hard- and soft-constrained formulations for enforcing initial and boundary conditions. The fractional derivatives are discretized using the L1 scheme for temporal terms and the Grunwald–Letnikov and shifted Grunwald–Letnikov formulations for spatial terms. A systematic comparison is performed among different variants of PINNs like vanilla, self-attention, ResNet, and the proposed location-attention under both hard and soft constraint settings. The numerical results demonstrate that the proposed attention-based framework significantly improves parameter efficiency and solution accuracy while maintaining stable convergence behavior. In particular, the location-based attention mechanism effectively captures the spatially varying and nonlocal dynamics inherent in fractional differential equations. The proposed framework achieves competitive performance compared with existing architectures and conventional numerical approaches, demonstrating its effectiveness for solving complex fractional advection-diffusion problems.
Förderung
- Ministry of Education, Government of India
- Korea Environment Industry & Technology Institut
- ANRF
- Korea Ministry of Climate, Energy, Environmen
Zitieren
APA
Sharma, A., Hwang, J., Jung, D., & Yadav, N. (2026). A physics-informed neural network for solving multi-term space-time fractional differential equations. Physica Scripta, 101(25), 256004. https://doi.org/10.1088/1402-4896/ae7672
Vancouver
Sharma A, Hwang J, Jung D, Yadav N. A physics-informed neural network for solving multi-term space-time fractional differential equations. Physica Scripta. 2026;101(25):256004. doi:10.1088/1402-4896/ae7672
BibTeX
@article{Sharma2026,
author = {Sharma, Aditi and Hwang, Jeonghwan and Jung, Donghwi and Yadav, Neha},
title = {A physics-informed neural network for solving multi-term space-time fractional differential equations},
journal = {Physica Scripta},
year = {2026},
volume = {101},
number = {25},
pages = {256004},
doi = {10.1088/1402-4896/ae7672}
}
EUVIMED