Simulated annealing for fitting linear combinations of Gaussians to data (Q5928221)

In this interesting paper the author presents a number of techniques for accelerating convergence when using simulated annealing for fitting linear combinations of Gaussians to data. The author's suggestions 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. A well known fact is demonstrated again, i.e. that providing a good initial parameter estimate is essential to finding a good fit in a reasonable amount of time. The presented approach seems to be promissing.