Improved linear response for stochastically driven systems (Q693186)

The paper deals with the problem of numerical instabilities for long response times of algorithms predicting the average response of nonlinear chaotic systems with forcing and dissipation to small external perturbation, being caused by the presence of positive Lyapunov exponents. The author uses the short time linear response formula in his algorithm, and shows that it is superior to the classical formula with Gaussian approximation for both additive and multiplicative forcing for short and moderate response times. He finally blends it with the classical formula to eliminate numerical instability also for long response times.