Fill in all x_best and f_best and c_best (also for meta-problems)

[darioizzo]

Most problems in PaGMO do not have the best_x, best_f, best_c defined, even though the solution is known.

Meta problems, at creation, should check if best_x, best_f, best_c are defined and, if so, define them for the meta problem too.

[krzysztof]

To clarify: best_x, best_f and best_c in problem class are the best known solutions (from definition and literature). The task here is to supplement those as in https://github.com/esa/pagmo/blob/master/src/problem/cec2006.cpp#L435

[AVCZ]

I've chosen one function just to start with and (hopefully) added the desired functionality. Sending a PR. As far as the "Meta problems" are concerned, I think it's now clear what the goal is, but I'm still looking for the most appropriate solution. So far I've considered only the simple cases (e.g., rotation, translation) and was thinking whether it is even necessary to first check whether best_x, best_f, best_c are defined or not? In case they are not defined, it would be basically just a no-op, but probably I'm missing some corner cases...

[AnirudhGP]

Hi I'd like to work on this bug. Could I get some pointers as to how I should go about fixing this? Thanks

[shivin9]

Hi, I want to work on this bug and was wondering whether you want an implementation as done in Ackley.cpp (best_x is initialized with the optimum value here 0 in the constructor) along with the necessary changes in the problem's description in doxygen. Thanks!

[darioizzo]

For simple cases, yes. That is how it should be done. More difficult cases, meta-problems for example, will require extra care.