FMINDEMO optimization cross-comparison
A systematic test of all optimization methods is performed using a set of test
problems. Starting configurations are chosen randomly.
The option flag may contain the following keywords:
'verbose' displays individual detailed optimization results.
'rand' adds a 10% gaussian random noise to all functions
The test can be repeated iteratively so that a Monte Carlo sampling of starting
parameters give a better stastistical estimate of each method efficiency.
Calling:
fmindemo(dimensionality=vector, option='verbose, rand', repetitions)
Example:
fmindemo([ 2 10 50],1) % detailed test with 2,5 and 10 parameters
fmindemo(2,[], 10) % 2 parameters optimization repeated 10 times
Contrib:
Test functions from Nikolaus Hansen, 2001-2007. e-mail: hansen@bionik.tu-berlin.de
(c) E.Farhi, ILL. License: EUPL.