2008-04-02 12:36:34 +00:00

36 lines
1.3 KiB
Matlab

function int = postProcess(int, steps, sigmaClust, varargin)
% Do post processing of the last optimization results and export
% the solution set to the JEInterface.
% int=postProcess(int, steps, sigmaClust [, nBest])
% Either of the parameters steps and sigmaClust may be 0 to leave out
% hill-climbing or clustering step. Setting both to 0 is not meaningful.
% The optional nBest parameter gives an upper bound to the number of
% returned solutions, which are sorted by fitness. Leaving this parameter
% out means that all solutions suggested by the optimizer (after
% clustering if activated) are returned, the number of which is usually,
% but not for all optimizers (e.g., not for CBN), limited to the population size.
%
% Arguments:
% int: the JEInterface instance
% steps: number of hill climbing steps to perform or 0
% sigmaClust: paramter for the density based clustering to perform or 0,
% relative to the problem range
% nBest: (optional) maximum number of solutions to return
% stepSize: (optional) step size of the stochastic hill climber.
if (int.finished == 0)
error('please wait for the current run to finish');
end
int.finished = 0;
if (nargin > 3) && (isnumeric(varargin{1}))
nBest = varargin{1};
else
nBest = -1;
end
int=runEvalLoopJE(int, 2, -1, '', steps, sigmaClust, nBest);