Sampler drawing samples with p=0

[charlieprior]

I am trying to use lintsampler to draw samples from a 2D pdf generated from an interpolated dataset (using scipy.interpolate.RegularGridInterpolate). However some of the samples generated, when evaluated using the original pdf, have probability 0. Should there be some kind of check for this?

[aneeshnaik]

Hi @charlieprior. This is weird, we haven't seen this kind of behaviour from lintsampler before. Off the top of my head this could be a resolution issue (i.e., if the true PDF has 'holes' of zero probability but these holes aren't resolved in the lintsampler grid). Would you mind sharing some code so we can see what exactly you're doing and work out what might be going wrong?