The Random Phase and Amplitude Formalism (RPA) has significantly extended the scope of weak turbulence studies. Because RPA does not assume any proximity to the Gaussianity in the wavenumber space, it can predict, for example, how the fluctuation of
the complex amplitude of each wave mode grows through nonlinear interactions with other modes, and how it approaches the Gaussianity. Thus, RPA has a great potential capability, but its validity has been assessed neither numerically nor experimentally. We compare the theoretical predictions given by RPA with the results of direct numerical simulation (DNS) for a three-wave Hamiltonian system, thereby assess the validity of RPA. The predictions of RPA agree quite well with the results of DNS in all the aspects of statistical characteristics of mode amplitudes studied here.
