A High-Order Spectral-IMEX Framework for Neural Field Equations with Space-Dependent Delays
This paper develops and analyzes a high-order numerical framework for neural field equations with spatially dependent time delays. The spatial discretization uses Chebyshev spectral collocation, while time integration employs third-order (ARK3) and fourth-order (ARK4) additive Runge–Kutta implicit–explicit schemes. Fourth-order Hermite interpolation is used to approximate delayed states. Under the stated regularity assumptions, we derive spatial spectral error estimates together with third- and fourth-order temporal convergence estimates for the fully discrete schemes. Numerical experiments support the predicted convergence orders and compare computational cost with a second-order CNAB2 scheme. For the reported two-dimensional work–accuracy test at a target error of 10 - 11 {10^{-11}} , ARK4 attains an approximately 420-fold speedup on the specified GPU platform. These measured efficiency gains are dependent on the implementation, hardware, and timing protocol. The proposed framework provides a high-order approach for numerical simulations of delayed neural field dynamics.
Authors
- Fangfang Qin (ORCID: https://orcid.org/0009-0006-6628-6736)
- Bingchi Liu (ORCID: https://orcid.org/0009-0002-7132-4275)
- Jian Huang
- Yanping Chen
Institutions
- Nanjing University of Posts and Telecommunications (CN)
- Xiangtan University (CN)
Publication Details
- Journal
- Computational Methods in Applied Mathematics
- Published
- 2026-11-20
- DOI
- https://doi.org/10.1515/cmam-2026-0168
- Primary Topic
- Numerical Solvers for Ordinary Differential Equations
- Type
- article
- Field-Weighted Citation Impact
- 0.00