To maximize a likelihood function that has local maxima, what alternative maximization methods are available in Stata? Are there genetic algorithm options or something like simulated annealing for this purpose?
Related to this, what type of an algorithm does ml search use? Random search or hill climbing? If I specify a large enough number for "ml search, r(#)" and let it search for days, would it catch problematic regions for my problem? (I have about 10 parameters, so this may be a long shot)