Improving the performance of regularized Stokeslet simulations using Rosenbrock time integration methods (Q7420459)

We demonstrate the potential of linearly implicit Rosenbrock time integration methods to greatly reduce computational time required for regularized Stokeslet simulations, achieving speed-ups up to two thousand times over explicit Runge-Kutta methods by increasing stability and reducing stiffness in a classic model problem. We also present approximate scaling laws for this speedup as a function of key model parameters at the root of numerical stiffness.