# Is there any stochastic method to approximate the likelihood?

I am looking for a list of stochastic algorithms to approximate likelihoods.

-
Can you be more specific? –  cardinal Nov 23 '11 at 22:12
If likelihood analytically very expensive or not able to calculate then the likelihood-free (approximate Bayesian computation (en.wikipedia.org/wiki/Approximate_Bayesian_computation) ) methods are available. I am looking for an alternative to such algorithm in domain of stochastic algorithms. –  Biostat Nov 23 '11 at 22:39
What do you mean by "stochastic algorithms"? ABC is a stochastic algorithm since it relies on simulation. –  Xi'an Nov 24 '11 at 8:51
You could try Googling 'simulated maximum likelihood'. I don't know anything about it myself beyond the name. –  onestop Nov 24 '11 at 10:22
Indirect inference and ABC are very similar (see for instance the incoming <a href="rss.org.uk/uploadedfiles/userfiles/files/… Paper in JRSS B</a>). They both use simulation. So I am still unclear as to what you are precisely looking for. A way to maximise the likelihood rather than approximating the whole function? This would be the domain of simulated annealing methods. –  Xi'an Nov 25 '11 at 8:08