#49 Results from MOEA solution

[Anonymous]

Hi, I successfully extended the MOEA AbstractProblem to solve my multi-objective problem. But the solutions were still different for very iteration that after the constraints were satisfied (see below outputs). I don't know what is the reason for that? I think the solution should be converged into one stable final solution, am I wrong?

In addtion, when I tried a mini problem with 2 decision variables, 3 objectives and 1 constraints, the results were a set of pareto solutions(about 6 solutions for each iteration). But when the number of decesion variables and parameters increased to about 400, there was only 1 solution outputed rather than a set of pareto solutions. I am wondering what is the problem for that?

Outputs:
Iteration: 70,000
Solution 1:
Obj 1: 23099.925807989108 Obj 2: 3973.561007339157 Obj 3: 1880.9047673527652
Var 0:6 Var 1:0 Var 2:3 Var 3:1 Var 4:5 Var 5:9 Var 6:12 Var 7:1 Var 8:0 Var 9:21
Con 1: 0.0

Iteration: 80,000
Solution 1:
Obj 1: 31320.8336711078 Obj 2: 5503.980317019027 Obj 3: 2240.6296175049433
Var 0:0 Var 1:1 Var 2:2 Var 3:0 Var 4:28 Var 5:6 Var 6:13 Var 7:55 Var 8:2 Var 9:13
Con 1: 0.0

Iteration: 90,000
Solution 1:
Obj 1: 31906.44272446188 Obj 2: 4971.337898616112 Obj 3: 2490.4302888334696
Var 0:20 Var 1:2 Var 2:23 Var 3:24 Var 4:0 Var 5:19 Var 6:15 Var 7:0 Var 8:14 Var 9:1
Con 1: 0.0

[Anonymous]

Even after the constraints are satisfied, MOEAs will still optimize the objectives. On hard problems, MOEAs tend to converge near the optimal solution. The magnitude of changes should decrease as the solutions approach the optimum.

Increasing the number of decision variables will make a problem harder to solve, in general. It just happens that given your time constraints (the maximum number of function evaluations), the MOEA is only able to find one solution. Running longer may allow it to discover more than one. Also check to see if the solution is feasible or not. Adding more decision variables may make it harder to find feasible solutions.