Code covered by the BSD License  

Highlights from
Pareto Set

image thumbnail
from Pareto Set by Yi Cao
find the pareto set from n points with k objectives

membership=paretoset(X)
function membership=paretoset(X)

% PARETOSET  To get the Pareto set from a given set of points.
% synopsis:           membership =paretoset (objectiveMatrix)
% where:
%   objectiveMatrix: [number of points X number of objectives] array
%   membership:      [number of points X 1] logical vector to indicate if ith
%                    point belongs to the Pareto set (true) or not (false).
%
% by Yi Cao, Cranfield University, 02 June 2007
% Revised by Yi Cao on 17 October 2007
% Version 3, 21 October 2007, new sorting scheme to improve speed.
% Bugfix, 25 July 2008, devided by zero error is fixed.
%
% Examples: see paretoset_examples
%

m=size(X,1);
Xmin=min(X);
X1=X-Xmin(ones(m,1),:);     %make sure X1>=0;
Xmean=mean(X1); 
%sort X1 so that dominated points can be removed quickly
[x,checklist]=sort(max(X1./(Xmean(ones(m,1),:)+1)));
Y=X(checklist,:);                  
membership=false(m,1);
while numel(checklist)>1
    k=checklist(1);
    [membership(k),checklist,Y]=paretosub(Y,checklist);
end
membership(checklist)=true;

function [ispareto,nondominated,X]=paretosub(X,checklist)

Z=X-X(ones(size(X,1),1),:);
nondominated=any(Z<0,2);                    %retain nondominated points from the check list                         
ispareto=all(any(Z(nondominated,:)>0,2));   %check if current point belongs to pareto set    
X=X(nondominated,:);
nondominated=checklist(nondominated);

Contact us at files@mathworks.com