Phase flow and statistical structure of Galton-board systems

Abstract

Galton boards, found in museum exhibits devoted to science and technology, are often used to demonstrate visually the ubiquity of so-called ‘‘laws of probability’’ via an experimental realization of normal distributions. A detailed theoretical study of Galton-board phase-space dynamics and statistical behavior is presented. The study is based on a simple inelastic-collision model employing a particle fall- ing through a spatially periodic lattice of rigid, convex scatterers. We show that such systems exhibit indeterminate behavior through the presence of strange attractors or strange repellers in phase space; nevertheless, we also show that these systems exhibit regular and predictable behavior under specific circumstances. Phase-space strange attractors, periodic attractors, and strange repellers are present in numerical simulations, confirming results anticipated from geometric analysis. The system’s geometry (dictated by lattice geometry and density as well as the direction of gravity) is observed to play a dominant role in stability, phase-flow topology, and statistical observations. Smale horseshoes appear to exist in the low-lattice-density limit and may exist in other regimes. These horseshoes are generated by homoclinic orbits whose existence is dictated by system characteristics. The horseshoes lead directly to deterministic chaos in the system. Strong evidence exists for ergodicity in all attractors. Phase-space complexities are manifested at all observed levels, particularly statistical ones. Consequently, statistical observations are critically dependent upon system details. Under well-defined circumstances, these observations display behavior which does not constitute a realization of the ‘‘laws of probability.’’