I have been testing both the CMA-ES and Bayesian optimisers, so I only have information on these two. I have run each algorithm for thousands of iterations and have noticed that some iterations, albeit rarely, suggest NaN for values of all variable. All of the variables are bounded by choco.uniform(0, X), with X an integer, and I was wondering whether this was intentional?
I have been testing both the CMA-ES and Bayesian optimisers, so I only have information on these two. I have run each algorithm for thousands of iterations and have noticed that some iterations, albeit rarely, suggest NaN for values of all variable. All of the variables are bounded by choco.uniform(0, X), with X an integer, and I was wondering whether this was intentional?