Surrogates model breaks downstream

[vikram-s-narayan]

It looks like moving from QuasiMonteCarlo@0.2.16 to QuasiMonteCarlo@0.2.19 breaks a surrogate model called GEKPLS downstream in Surrogates.jl.

This gist may help pinpoint the issue.

[vikram-s-narayan]

Additional info:

It looks like GoldenSample is generating a different set of points with version 0.2.19.

QuasiMonteCarlo@0.2.19 includes the lower bound as the first sample when we use GoldenSample(). This was not the case with QuasiMonteCarlo@0.2.16 which maintained an offset. For example, if we do:

using QuasiMonteCarlo
lb = [0.125, 5.0, 5.0]
ub = [1.0, 10.0, 10.0]
n_test = 5
transpose(QuasiMonteCarlo.sample(n_test, lb, ub, GoldenSample()))

We get the following results:

With QuasiMonteCarlo@0.2.16:

 0.192681  5.96681  6.29918
 0.697863  9.43361  5.09836
 0.328044  7.90042  8.89754
 0.833226  6.36723  7.69671
 0.463407  9.83403  6.49589

With QuasiMonteCarlo@0.2.19 (note that the first record is the same as the lower bound):

 0.125     5.0      5.0
 0.841776  8.35522  7.7485
 0.683552  6.71044  5.497
 0.525328  5.06565  8.24551
 0.367104  8.42087  5.99401

[vikram-s-narayan]

This issue may partly be caused by what appears to be a minor indexing issue in Kronecker.jl:

return @. mod(generator * (0:(n - 1))' + g.origin, 1)

should be

return @. mod(generator * (1:(n))' + g.origin, 1)

Also, the source article suggests that a seed of 0.5 works better. This suggestion had been implemented in prior versions but seems to have been removed in 0.2.19?

[ChrisRackauckas]

@ParadaCarleton can you take a look at this?

[ParadaCarleton]

@ParadaCarleton can you take a look at this?

Will make a PR today correcting these issues as part of a general refactor (breaking everything into files, fixing names, and some other stuff).