Tom McClintock

Agreed with the above. This repo is important enough to have new stewardship.

I just wanted to chime in that this is possible by transforming the Gaussian posterior predictive distribution to a _truncated_ Gaussian PPD. In other words, one can train the GP...

@lleewwiiss good find. It turns out that because of the nice properties of Gaussians, you can construct even more sophisticated constrained GPs (like ones where the slope or convexity of...

Oh the ones in the paper? I do not. Or did you mean the one I commented about in November?

Ah I see. Not on hand, but I'll whip up a gist and ping you. `scipy.stats.truncnorm` is very unwieldy and not amenable to working with a GP since the `loc`...

@lleewwiiss here you go: https://github.com/tmcclintock/TruncatedGP It ended up being too long for a gist. I show the posteriors for the truncated and non-truncated GPs, as well as draws from both...

This is probably late but here is a minimal example for multidimensional data ```python """Minimal example showing multidimensional GP regression. """ import george import numpy as np x = np.linspace(0,...

This is being addressed in #112 .

Thanks @lgarrison I'll try to reproduce this in a clean way. Perhaps I misdiagnosed the problem. Until I do properly figure out the cause I'll just close the issue. Also,...

Hey @manodeep. I think that the error might be in the 2ptcf_jk package, since I have confirmed that I can run `DD()` on the catalog itself without issue. I'll post...