Estimating limits on resolution and predictability in computations of hydrodynamically unstable flows (Q1096489)

Unstable hydrodynamic flows containing different materials such as Rayleigh-Taylor instability rapidly evolve to a chaotic state with limited predictability and a distribution of materials having a fractal dimension. Here it is shown how Lagrange points, either existing as part of the hydrodynamic solution method or added as passive tracers, can be used to test for self-similarity and to compute the dimension and spatial resolution scaling for the Rayleigh-Taylor instability mixing region. The efficiency of the integral or point counting methods of estimating dimension is discussed and it is shown how it can be increased by order of magnitude factors using random sampling. The problem of how to compute the rate of loss of memory of initial conditions in the transient case is discussed and a quantity similar to a largest Lyapunov exponent is defined and used to determine the predictability time for the Rayleigh- Taylor problem.