README.md

gpytorch-astro

GPyTorch Gaussian Process framework applied to astronomy time series datasets

Running the demos

There is a conda environment file in this repository that contains all the dependencies needed for running the GPyTorch developer version, even remaking the documentation. I installed the developer versions of GPyTorch and lightkurve in case we eventually make Pull Requests to add any cool demos we make here.

You can replicate the conda environment in one line:

conda env create -f environment_MKL.yml
conda activate gpytorch_MKL

But you won't be done yet! You'll have to install the developer versions of:

• lightkurve
• GPyTorch

for reasons outlined above.

Alternatively you can just add those two packages to the environment_MKL.yml file:

- lightkurve
# ...
- pip
  - gpytorch

and then

conda env update -f environment_MKL.yml

ps, I'm not actually sure if the MKL support comes out of the box, you used to have to install from source.