I want to see if we can create a loop that will generate several starting points without having to run the code several times and obtain possible local optima and ultimately reach out to the global optima by comparing the very obtained local optima.
Dear Ahmad,
Your question is not very clear. Please what exactly do you want?
GA usually starts off with a random initial population, is this what you mean?
What exactly is a single matlab run? single iteration of the GA?
My regards,
Abubakar
Abubakar Bala thanks for coming in. What I meant was whether there is a modified version of GA that will make the algorithm start with several initial populations and find out what the optimal solution is to make sure it doesn't get stuck in local optima just because it has started on one single set of initial population. By single matlab run I mean I wanna run the code once but I want the code to include several initial populations.
Dear Ahmad,
I do not see any advantage to doing this. Or is there? Why not just increase the population size (say 50 individuals) and have more iterations. What problem are you trying to solve?
Regards,
Abubakar
Abubakar Bala The problem is that the I need to make sure multiple runs are carried out to bypass all the local optima, rather than increasing the population size. The issue is that my bound (lower and upper) is too extensive and I have many decision variables that will either make the model get stuck in the local optima or will make it terribly time-consuming. Increasing the population size will help me with finding the solutions but will take days or weeks to respond in light of extensive lower and upper bound, and too many decision variables. But several runs on different initial populations will increase the chance of reaching out to the optimal solution, I guess.
- Badges
- Science topic