
Re: Multiobjective optimization  gamultiobj  feasible point
Posted:
Apr 8, 2013 8:51 AM


On 4/8/2013 5:40 AM, Gabboc wrote: > "Torsten" wrote in message <kju1a1$nc6$1@newscl01ah.mathworks.com>... > >> TOT must be bigger than the sum of the lower bounds for the 10 variables >> for that a feasible point for your problem exists. >> Did you take this limitation into account ? >> > > In my optimization script lb is: > > lb = [1000 150 200 80 20 50 40 20 150 1000 500 70 60 100 100 20 16 > 3000 500 300 150 100 1 50 35 500 1000 27 500 600]; % Lower bound > > The sum of all elements of lb is 10339 and TOT=17000 > > Any ideas??
ga and gamultiobj attempt to find initial feasible points by, among other techniques, solving a linear program. If they say that there is no feasible point, then I believe that there is no feasible point.
You can check for yourself by using linprog to look for a feasible point. Take an objective function vector f as anything, say your lb vector, and solve x = linprog(f,A,b,Aeq,beq,lb,ub) This will tell you for sure if there is no feasible point.
Alan Weiss MATLAB mathematical toolbox documentation

