Stochastic algorithms are frequently employed for complex optimzation
problems, with many local minima.
These include simulated annealing, Tabu search and genetic algorithms.
My query is not about optimization algorithms but on the statistical
analysis of the results.
If the same algorithm is run on the same problem several times, this
will generate a sample of values of the objective function.
Is anyone on allstat aware of any work on statistical modelling of the
probablity distribution of these values.
I suspect that the distribution is unusual in being bounded below (at
the global mimimum) and having finite density there.
Does anyone on Allstat know of any work in this area? If there are
established methods then I would like to use them rather than invent my
own.
Replies to me please, and I will summarize to the list.
Thanks and best wishes
Tim Auton
--
T R Auton PhD MSc C.Math
Head of Biomedical Statistics
Protherics Molecular Design Ltd
Beechfield House
Lyme Green Business Park
Macclesfield
Cheshire SK11 0JL
UK
email: [log in to unmask]
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|