Ronnie Mainieri

Trace Formulas for Stochastic Evolution Operators: Weak Noise Perturbation Theory

Periodic orbit theory is all effective tool for the analysis of classical and quantum chaotic systems. In this paper we extend this approach to stochastic systems, in particular to mappings with additive noise. The theory is cast in the standard field-theoretic formalism and weak noise perturbation theory written in terms of Feynman diagrams. The result is a stochastic analog of the next-to-leading ħ corrections to the Gutzwiller trace formula, with long-time averages calculated from periodic orbits of the deterministic system. The perturbative corrections are computed analytically and tested numerically on a simple 1-dimensional system.

Cycle expansion for the Lyapunov exponent of a product of random matrices

Using cycle expansion for the thermodynamic zeta function, a formula is derived for the Lyapunov exponent of a product of random hyperbolic matrices chosen from a discrete set. This allows for an accurate numerical solution of the Ising model in one dimension with quenched disorder. The formula is compared with weak disorder expansions and with the microcanonical approximation and shown to apply to matrices with degenerate eigenvalues.

On the equality of Hausdorff and box counting dimensions

By viewing the covers of a fractal as a statistical mechanical system, the exact capacity of a multifractal is computed. The procedure can be extended to any multifractal described by a scaling function to show why the capacity and Hausdorff dimension are expected to be equal.

Phase shift in experimental trajectory scaling functions

Abstract For one-dimensional maps the trajectory scaling function is invariant under coordinate transformations and can be used to compute any ergodic average. It is the most stringent test between theory and experiment, but so far it has proven difficult to extract from experimental data. It is shown that the main difficulty is a dephasing of the experimental orbit which can be corrected by reconstructing the dynamics from several time series. From the reconstructed dynamics the scaling function can be accurately extracted.

The convergence of chaotic integrals

We review the convergence of chaotic integrals computed by Monte Carlo simulation, the trace method, dynamical zeta function, and Fredholm determinant on a simple one-dimensional example: the parabola repeller. There is a dramatic difference in convergence between these approaches. The convergence of the Monte Carlo method follows an inverse power law, whereas the trace method and dynamical zeta function converge exponentially, and the Fredholm determinant converges faster than any exponential. (c) 1997 American Institute of Physics.

Cycle expansions with pruned orbits have branch points

Abstract Cycle expansions are an efficient scheme for computing the properties of chaotic systems. When enumerating the orbits for a cycle expansion not all orbits that one would expect at first are present — some are pruned. This pruning leads to convergence difficulties when computing properties of chaotic systems. In numerical schemes, I show that pruning reduces the number of reliable eigenvalues when diagonalizing quantum mechanical operators, and that pruning slows down the convergence rate of cycle expansion calculations. I then exactly solve a diffusion model that displays chaos and show that its cycle expansion develops a branch point.

Dynamic scaling function at the quasiperiodic transition to chaos

Abstract We obtain a five-step approximation to the quasiperiodic dynamic scaling function for experimental Rayleigh-Benard convection data. When errors are taken into account in the experiment, the f(α) spectrum of scalings is equivalent to just two of these five scales. To overcome this limitation, we develop a robust technique for extracting the scaling function from experimental data by reconstructing the dynamics of the experiment.