Open AdityaSavara opened 1 year ago
I think it should switch to totally random when dimension is greater than 3.
Evidently, Sobol indices in SciPy have had an upgrade at some point, and the options in it could be investigated. https://github.com/SALib/SALib/issues/463
In higher dimensions, the odds of overlapping points are low. More importantly, I found that the higher dimension sobol sampling has distinct patterns (see attachment). Heat_Scatter.zip