I haven't read the paper, so maybe completely irrelevant, but isn't there an analytical solution for a system of N coupled oscillators?