Abstract
This paper addresses numerical computation of time-periodic diffusion equations with fractional Laplacian. Time-periodic differential equations present fundamental challenges for numerical computation because we have to consider all the discrete solutions once in all instead of one by one. An idea based on the diagonalization technique is proposed, which yields a direct parallel-in-time computation for all the discrete solutions. The major computation cost is therefore reduced to solve a series of independent linear algebraic systems with complex coefficients, for which we apply a multigrid method using the damped Richardson iteration as the smoother. Such a linear solver possesses mesh-independent convergence factor, and we make an optimization for the damping parameter to minimize such a constant convergence factor. Numerical results are provided to support our theoretical analysis.
Original language | English |
---|---|
Article number | e2178 |
Journal | Numerical Linear Algebra with Applications |
Volume | 25 |
Issue number | 5 |
DOIs | |
Publication status | Published - Oct 2018 |
Externally published | Yes |
Keywords
- convergence analysis
- diagonalization technique
- fractional diffusion equations
- multigrid method
- parameter optimization
- time-periodic condition