🚀 go-pugleaf

In article <1992Oct3.210337.21498@news2.cis.umn.edu> charlie@umnstat.stat.umn.edu (Charles Geyer) writes:

> It makes no sense to compare Gibbs sampling versus simulated annealing, 
> for example.

Correct, but not for the reasons you perhaps mean. Simulated annealing
is a technique that can be used in conjunction with Gibbs sampling 
(or other Markov chain methods, such as the Metropolis algorithm) 
in order to reach the final equilibrium distribution more quickly.

Most of the annealing literature is devoted to the case of annealing down
to a "temperature" of zero, where the distribution being sampled from
is concentrated at the points where the global minimum is realized, but
the technique is equally applicable when the goal is to sample from the
distribution at some non-zero temperature. 

Gibbs sampling and simulated annealing have been used in this fashion with
the "Boltzmann machine" neural network of Ackley, Hinton, and Sejnowski
(Ccognitive Science, vol. 9, pp. 147-169, 1985), and for many simulations 
of physical systems.

    Radford Neal