Universality classes for self-similarity of noiseless multi-dimensional Burgers turbulence and interface growth

Abstract

The present work is devoted to the evolution of random solutions of the unforced Burgers and KPZ equations in d-dimensions in the limit of vanishing viscosity. We consider a cellular model and as initial condition assign a value for the velocity potential chosen independently within each cell. We show that the asymptotic behavior of the turbulence at large times is determined by the tail of the initial potential probability distribution function. Three classes of initial distribution leading to self-similar evolution are identified: (a) distributions with a power-law tail, (b) compactly supported potential, (c) stretched exponential tails. In class (c) we find that the mean potential (mean height of the surface) increases logarithmically with time and the 'turbulence energy' E(t) (mean square gradient of the surface) decays as 1/t times a logarithmic correction. In classes (a) and (b) we find that the changes in the mean potential and energy have a power-law time dependence. In class (c) the roughness of the surface, measured by its mean--square gradient, may either decrease or increase with time. We discuss also the influence of finite viscosity and long range correlation on the late stage evolution of the Burgers turbulence