The optimal n-body method for stability studies of galaxies

The stability of a galaxy model is most easily assessed through N-body simulation. Particle-mesh codes have been widely used for this purpose, since they enable the largest numbers of particles to be employed. We show that the functional expansion technique, originally proposed by Clutton-Brock for other simulation problems, is in fact superior for stability work. For simulations of linear evolution it is not much slower than grid methods using the same number of particles, and reproduces analytical results with much greater accuracy. This success rests on its ability to represent global modes with a modest number of basis functions; grid methods may be more effective for other applications, however. Our conclusions are based on implementations of functional expansion and grid algorithms for disk galaxies.