Taylor-Fourier integration (2406.03124v1)

Abstract: In this paper we introduce an algorithm which provides approximate solutions to semi-linear ordinary differential equations with highly oscillatory solutions that after an appropriate change of variables can be written as a non-autonomous system with $(2\pi/\omega)$-periodic dependence on $t$. The proposed approximate solutions are written in closed form as functions $X(t,\omega\, t)$ where $X(t,\theta)$ is, (i) a truncated Fourier series in $\theta$ for fixed $t$, and (ii) a truncated Taylor series in $t$ for fixed $\theta$ (that is the reason for the name of the proposed integrators). Such approximations are intended to be uniformly accurate in $\omega$ (in the sense that their accuracy is not deteriorated as $\omega\to \infty$). This feature implies that Taylor-Fourier approximations become more efficient than the application of standard numerical integrators for sufficiently high basic frequency $\omega$. The main goal of the paper is to propose a procedure to efficiently compute such approximations by combining power series arithmetic techniques and the FFT algorithm. We present numerical experiments that demonstrate the effectiveness of our approximation method through its application to well-known problems of interest.