Probing optimisation in physics-informed neural networks

Nayara Fonseca, Veronica Guidetti, Will Trojak

Keywords: optimiser, physics-informed neural networks, training trajectory curvature

TL;DR: We introduce a new method to study optimisation in neural networks, which provides training trajectory curvature data at low computational cost.

Abstract: A novel comparison is presented of the effect of optimiser choice on the accuracy of physics-informed neural networks (PINNs). To give insight into why some optimisers are better, a new approach is proposed that tracks the training trajectory curvature and can be evaluated on the fly at a low computational cost. The linear advection equation is studied for several advective velocities, and we show that the optimiser choice substantially impacts PINNs model performance and accuracy. Furthermore, using the curvature measure, we found a negative correlation between the convergence error and the curvature in the optimiser local reference frame. It is concluded that, in this case, larger local curvature values result in better solutions. Consequently, optimisation of PINNs is made more difficult as minima are in highly curved regions.