Possible integration of MCCGA in Metaheuristics.jl about metaheuristics.jl HOT 9 CLOSED

sample(probvector) performs rand() call for each single bit, so in a p-variable problem it performs p x 32 x rand() calls in a map().

floats() is a little bit costy as it combines 32bits to build up an integer then later reinterpretes it as Float.

imo, in initial state building, it is not the most costy part when it is compared to generating the initial probvector itself.

from metaheuristics.jl.

Thanks for your response. MCCGA is now implemented within the Metaheuristics.jl package (master branch).
It would be nice if the docstring about MCCGA is updated.

• This description is wrong (my mistake).
• Bibliographic references can be inserted here.
• The docstring for this function could be improved

Example

using Metaheuristics

# MCCA requires Optim (unless you put MCCA(use_local_search=false))
import Optim 

# problem
f, bounds, solutions = Metaheuristics.TestProblems.rastrigin();

result = optimize(f, bounds, MCCGA())

Output:

+=========== RESULT ==========+
  iteration: 31419
    minimum: 0
  minimizer: [3.9383281247463824e-29, -1.60938227312533e-36, 5.963404220446944e-28, 6.760617690476724e-38, -1.3224051368625615e-32, 9.76955021280433e-39, -3.574582341995119e-38, 1.6852928556302238e-25, 1.776057958933574e-36, 4.273668128645299e-34]
    f calls: 63576
 total time: 3.3656 s
+============================+

from metaheuristics.jl.

nice job! now Metaheuristics.jl includes more methods. Thank you so much for your nice attitude, effort, and help.

from metaheuristics.jl.

You are welcome! If you have more recommendations or any questions about this package, please let me know. I'm open to collaborations on topics in optimization and AI.

Please, feel free to close this issue at any time.

from metaheuristics.jl.

Hi! The MCCGA approach can be implemented into the Metaheuristics package (MHP). I'd like MHP has as many optimizers and features as possible. Therefore, MCCGA is completely suitable to be included in MHP.

Some comments before integration:

• I noted you are using Optim in your dependencies, which can be included in MHP via Requires.jl.
• If you want to carry out the integration, I'd suggest you read "Create Your Own Metaheuristic".
• I can perform the integration, or help you including the tests, documentation, or solving any issue you have related to the implementation.

I am also open to further collaborations.

from metaheuristics.jl.

Based on your response via DM, I started the implementation in develop branch.

At this point, MCCGA is not working correctly in MHP due to I just copy, paste and adapt your code.

I have a question for you in the following line:

Thanks in advance.

from metaheuristics.jl.

that was super fast, thank you!

from metaheuristics.jl.

Just a quick and dirty PR, #14, that adds the example below. I will enrich the documentation over time.

By the way, it seems MCCGA minimizes Rastrigin nicely :)

from metaheuristics.jl.

Thank you for contributing. I just merged the changes.

MCCGA is an interesting and original GA, I'm excited it is now integrated into Metaheuristics.jl.

from metaheuristics.jl.