Simulated annealing for fitting linear combinations of Gaussians to data

It is difficult to find a good fit of a combination of Gaussians to arbitra ry empirical data. The surface defined by the objective function contains m any local minima, which trap gradient descent algorithms and cause stochast ic methods to tarry unreasonably in the vicinity. A number of techniques fo r accelerating convergence when using simulated annealing are presented. Th ese are tested on a sample of known Gaussian combinations and are compared for accuracy and resource consumption. A single 'best' set of techniques is found which gives good results on the test samples and on empirical data.