Research Communication | Open Access
Volume 2021 | Communication ID 11
Volume 2021 | Communication ID 11
| A new approach to build linear schemes for Keller-Segel models |
Mohammed Benzakour Amine
Received |
Accepted |
Published |
January 08, 2021 |
January 31, 2021 |
March 15, 2021 |
Abstract: Solving nonlinear systems of partial differential equations with fully implicit schemes requires the numerical resolution of a large nonlinear algebra system, which is too expensive. Standard linearized semi-implicit schemes are more advantageous in terms of computational cost, but the main drawback of such schemes is their lack of accuracy. An algorithm based on a single-layer neural network is developed to build more accurate linearized implicit schemes for Keller-Segel chemotaxis systems: the selected steps training algorithm. The proposed schemes use also a finite volume spatial ...