Stochastic model of the Rayleigh-Taylor turbulent mixing

We propose a stochastic model, which describes the random character of the Rayleigh-Taylor turbulent mixing. Fluctuations of the rate of momentum loss are accounted for through multiplicative noise with uniform and log-normal distributions. We show that the mixing growth-rate (so-called ``alpha,'' alpha = h/gt$^2$) is extremely sensitive to the stochastic effects and long tails of the distributions. The ratio between the rates of momentum loss and momentum gain as well as the rates of energy dissipation and energy gain is the statistic invariant and a robust parameter to diagnose for either sustained or time-dependent acceleration.