Previous
abstract 
Next
abstract
We discuss the general problem of finding multiple planets around pulsars through analysis of pulsar timing data. Fitting a full Keplerian orbit requires a search through four non-linear parameters for each planet. This problem is especially difficult when there is a large range of planetary masses and orbital periods.
As one means for attacking the search problem, we have considered genetic algorithms, which are a general method for optimization that make use of biological-like genetic concepts like ``survival of the fittest,'' mutation, and chromosome exchange. Through these means, the algorithm searches parameter space in the same way that life finds optimal niches in the biological environment: through incremental rewarding of successful genetic variations.
We show examples of the genetic algorithm as applied to simulated pulsar data and we compare its performance with alternative methods such as grid searches, nonlinear least squares fitting, the simplex method, hillclimbing, and simulated annealing. We also show preliminary application to real pulsar data.
