Schematic illustration of the SINDy algorithm. A numerical derivative of the data from different GRAINS simulation are taken and put into a large data matrix $\stackrel{.}{X}$ (left square). A function library, Θ, is created and a linear regression problem is solved to obtain the coefficient matrix Ξ, including a sparsity regularisation term (top right square). The found equations of motion are then analysed using dynamical systems theory (lower right square).