Record best solution in each GA generation running with UseParallel=true

Hello,

Is there a way to record the best solution at each generation running with parallelization? (that is, with UseParallel=1)

I used outputFcn_global (from File Exchange), but it uses a global variable to record the info. In each generation, I would like to use the output function to just record the best solution (e.g. 1x1000 array) in the population. Since my objective function is a combination of several things, I need the best solution of each generation to then calculate the value of each term in the fitness function separately, and see how the objectives evolve over the generations.

Is there a way to do that without global variables?

Answer 1

[Star Strider]

See: How to save data from Genetic Algorithm in case MATLAB crashes? That will likely do what you want. Define it as the output function 'OutputFcn',@SaveOut.

[Mitsu]

Finally got back to work on this, sorry for not upvoting earlier.

Thank you! I misunderstood what the output function does, and thought it was not compatible with parallelizing the optimizer. Since the output function is called only once during each iteration, it works fine!

[Star Strider]

As always, my pleasure!