Extracting dynamical structure from unstable periodic orbits.

The topological recurrence algorithm provides a fast and robust method for detecting the presence of unstable periodic orbits (UPO's) in short, noisy experimental data files. We present here a technique for improving this method by using a matrix fitting algorithm to extract dynamical information about the system from these UPO's. This method greatly increases the sensitivity of the algorithm, and also provides a method for identifying false positive results.