Hello,
My question is how can I evaluate a discrete metaheuristic using a benchmack mathematical functions. in other words, the concerned metaheuristic handel a discrete solutions xi with D dimensions where each dimension can have discrete values (or either binary 0 or 1).
How can this metaheuritics be evaluated using the existing mathematical functions that have a continous inputs ?
In the case of a binary discrete values is their a special functions for that ?
Thank you very much.
Best regards.
I think you need to use a discrete function. There are several benchmarks that deal with binary strings, ranging from simple (onemax, where the best solution is a binary vector of all '1's, and the fitness is the sum of all the '1's in a given candidate solution) to complex (NK-landscapes, https://en.wikipedia.org/wiki/NK_model).
thank you Aberto, can you give me more examples or even propose me a link of a recent article that uses this kind of funtion.
Onemax is relatively trivial; for NK-landscapse, you can find several references and the associated code here: http://www.cs.unibo.it/fioretti/CODE/NK/index.html
It may be useful to use "log sigmoid" function to convert continuous variable to discrete variable.
- Badges
- Science topic
- Similar topics
- Mathematics