Green's function-based time stepping for the Kuramoto-Sivashinsky initial-boundary value problem
For researchers studying pattern formation in dissipative systems with fixed boundaries, this work provides a numerically stable alternative to ill-conditioned polynomial-based methods.
The paper presents a Green's function-based time-stepping method for the Kuramoto-Sivashinsky equation with fixed boundary conditions, eliminating numerical differentiation and linear solving. The method is validated over five decades of viscosity, demonstrating accurate and efficient solutions.
Both theoretical and numerical studies of the Kuramoto-Sivashinsky equation have mostly considered periodic boundary conditions. In this setting, the Fourier decomposition of the solution is central to theoretical ideas, such as renormalization group arguments, as well as to numerical solution, allowing for the construction of accurate and efficient time-steppers using standard pseudo-spectral methods. In contrast, fixed boundary conditions induce boundary layers and necessitate the use of non-uniform grids, usually generated by orthogonal polynomials. On such bases, numerical differentiation is ill-conditioned and can potentially lead to a catastrophic blow-up of round-off error. In this paper, we use ideas recently explored by Viswanath (J. Comput. Phys. 251(2013), pp. 414-431) to completely eliminate numerical differentiation and linear solving from the time-stepping algorithm. We use the Green's function-based method to investigate elements of the Kuramoto-Sivashinsky dynamics over a range of five decades of the viscosity.