Meta problem opt transformation

[AVCZ]

No description provided.

[AVCZ]

After all I found only three out of the eight meta-problems currently implemented suitable for minima updates at creation from minima of the original problems so that the transformation would be valid in the general case. The scaled meta-problem is also a candidate, but at this point I don't know what values the scale can attain (or which values the scaled class expects). Any comments are welcome.

[darioizzo]

test on best_f failed .... any idea?

[krzysztof]

I think the signature for the transformation function should be either transform_x(const std::vector<d_v>&) or transform_best_x(void), as the signature either implies we can transform arbitrary vector (argument), or we transform only the "best" one, which is already member of the problem class.

[AVCZ]

@kiryx Yeah, it makes sense since we can transform every vector, did the refactoring. @jdiez17 Thanks, that's a good eye of yours.

[darioizzo]

still failing the tests

[AVCZ]

@darioizzo Will look into it, unfortunately I have to leave now, but will get back to it today.

[jdiez17]

@AVCZ I believe the unit test that fails with this PR can be fixed by rebasing master (I included some changes to best_solutions_test.cpp) that might be relevant.

[AVCZ]

@kiryx Yeah, I will break up the test into functions. I also considered once again the transform_best_x vs. transform_x naming and think that we should actualy return the transformed vector(s) from the function (now that it's called transform_x) so that we don't have to call set_best_x(...) inside this function and whenever we decide to use this function for transforming arbitrary vectors, we don't actualy change the minima (with incorrect values).
@darioizzo Also looking at the tests now. I think that the change didn't actually break any of the original code, the tests just weren't written with this in mind and I forgot to tune them where appropriate. (But looking at the conversation in the pagmo thread I guess it's clear now).

[darioizzo]

I would like to see if I can merge this in master ... can you fix the auto merge?

[landscape-bot]

Code quality remained the same when pulling 70b7b76 on AVCZ:meta-problem_opt_transformation into 13bcce4 on esa:master.