Pareto optimization

[sqbl]

Investigate whether optimizer.py line 1085 should take the transformed x as argument

[sqbl]

Investigate further, if it nneds to be changed (which I believe it does), then consider a solution like:

from ProcessOptimizer.learning import GaussianProcessRegressor
if isinstance(opt.base_estimator_, GaussianProcessRegressor):
    ...fit(...transform(xx)...)
else:
   ...fit(xx)...

[sqbl]

Specifically, line 1083 could become something like this:
for i in range(self.n_objectives): if isinstance(self.base_estimator_, GaussianProcessRegressor): xx_tranformed = self.space.transform(xx) F[i] = self.models[-1][i].predict(xx_tranformed)[0] else: F[i] = self.models[-1][i].predict(xx)[0]

After trying this shortly, it appears to give some issues. with the transformation

[sqbl]

There seems to be concensus that pareto is working. I must have missed the line of code that ensures this