On the convergence of Lawson methods for semilinear stiff problems (Q777507)

The convergence behaviour of Lawson-type exponential integrators is studied for semilinear problems. The convergence proof is based on the variation-of-constants formula, whereas the nonlinearity is expanded along with the flow of the homogeneous problem. The expansion of the exact solution is carried out in terms of elementary integrals, and of that of the numerical solution in terms of elementary quadrature rules. It is shown that the non-stiff order conditions together with some assumptions on the exact solution give full order of convergence. The authors prove that a Lawson method converges with order \(p\) if the order of the underlying Runge-Kutta method is at least \(p\) and the solution satisfies appropriate regularity assumptions. The regularity assumptions are related to the corresponding conditions for splitting methods. These conditions are studied for methods of orders one and two, respectively, and is worked out for the nonlinear Schrödinger equation. The error analysis reveals a different behaviour between the first-order Lawson method and the exponential Euler method, as observed in numerical experiments for the linear Schrödinger equation with periodic boundary conditions.