Latin Hypercube Sampling just creating a grid?

[SandorAlbert]

Hi, thank you so much for this package. I just noticed that:

opt = ProcessOptimizer.Optimizer([(-5, 5), (-5, 5)], n_initial_points=50, lhs=True)

just creates an evenly spaced grid instead of a latin hypercube - is this a bug or was this intended? At some point the sample point also overlap. Does this have to do with the boundaries being interpreted as integers?

Thank you in advance and many regards :)

*Edit. Woops, yes.

SPACE = Space([Real(-5,5), Real(-5,5)])

and

opt = ProcessOptimizer.Optimizer(SPACE, n_initial_points=50, lhs=True)

solves this issue. Sorry!

[SRFU-NN]

If the upper an lower bound of a Dimension in Space is given as integers, an Integer dimension is initialized. It sounds like that was what happened?

If either is bound is a float, a you will get a Real Dimension instead. So a simpler way to solve your problem would be

opt = ProcessOptimizer.Optimizer([(-5.0, 5.0), (-5.0, 5.0)], n_initial_points=50, lhs=True)