search

guilgautier / DPPy

Python toolbox for sampling Determinantal Point Processes
https://dppy.readthedocs.io
MIT License
219 stars 52 forks source link

The option to sample more than once #4

Closed rbardenet closed 6 years ago

rbardenet commented 6 years ago

I think I'd only implement a sample method that can draw one sample, to keep things simple and avoid confusion for future users between the number of DPP samples and the cardinality of one sample.

Relatedly, you illustrate beta-ensembles by showing histograms that involve several samples, don't you? If I take GUE, you show agreement with the semi-circle law. But if you want to illustrate Wigner's theorem, putting several samples in the same histogram is a bit strange.

guilgautier commented 6 years ago

About the histograms, in fact when I was displaying the first plots in the notebook in order to play and test some modifications I was working with small number of points N/size and I was averaging Wigner's theorem to check if things were ok But you're right I should consider the limiting case N/size large one a single sample when using the hist() function.

About the potential confusion between number_of_samples and cardinality of one sample I don't think such confusion is possible since it is a parameter given by the user via N/size.

guilgautier commented 6 years ago

@rbardenet Histogram display is more accurate now in the Notebook