elcorto/gp_playground

Explore selected topics related to Gaussian processes

gp_playground's Issues

`Gpy` is not built using `cython`

This a minor thing but still worth documenting.

We removed cython from reqs since it gets pulled in by Gpy anyway as install_requires. However the wheel doesn't get built using cython. Therefore we use Gpy using the numpy fallback, even if we install cython first. Must be a Gpy issue.

Here we create a venv, install cython, Gpy and then the rest in order. cython is not used.

$ fd -g "GPy*.whl" ~/.cache/pip
/home/elcorto/.cache/pip/wheels/27/bd/9f/82ab4216eae088cba864ca0dc1d75699bd4bf6823790fb2f77/GPy-1.10.0-py3-none-any.whl

$ fd -g "GPy*.whl" ~/.cache/pip | xargs rm
$ mkvirtualenv gp_play
(gp_play) $ pip install --extra-index-url https://download.pytorch.org/whl/cpu torch
(gp_play) $ pip install cython
(gp_play) $ pip install Gpy
Collecting Gpy
  Using cached GPy-1.10.0.tar.gz (959 kB)
  Preparing metadata (setup.py) ... done
Collecting numpy>=1.7
  Using cached numpy-1.24.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (17.3 MB)
Collecting six
  Using cached six-1.16.0-py2.py3-none-any.whl (11 kB)
Collecting paramz>=0.9.0
  Using cached paramz-0.9.5-py3-none-any.whl
Requirement already satisfied: cython>=0.29 in /home/elcorto/.virtualenvs/gp_play/lib/python3.10/site-packages (from Gpy) (0.29.33)
Collecting scipy>=1.3.0
  Using cached scipy-1.10.0-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (34.4 MB)
Collecting decorator>=4.0.10
  Using cached decorator-5.1.1-py3-none-any.whl (9.1 kB)
Building wheels for collected packages: Gpy
  Building wheel for Gpy (setup.py) ... done
  Created wheel for Gpy: filename=GPy-1.10.0-py3-none-any.whl size=1089372 sha256=8f010617e946ba7bdb211d3e78dd37e2e2be165a377d355cefea5f78dbfdefb8
  Stored in directory: /home/elcorto/.cache/pip/wheels/27/bd/9f/82ab4216eae088cba864ca0dc1d75699bd4bf6823790fb2f77
Successfully built Gpy
Installing collected packages: six, numpy, decorator, scipy, paramz, Gpy
Successfully installed Gpy-1.10.0 decorator-5.1.1 numpy-1.24.1 paramz-0.9.5 scipy-1.10.0 six-1.16.0

(gp_play) $ pip install -r requirements.txt

(gp_play) $ python book/content/gp_pred_comp/notebook_comp.py
No GPU/TPU found, falling back to CPU. (Set TF_CPP_MIN_LOG_LEVEL=0 and rerun for more info.)
warning in stationary: failed to import cython module: falling back to numpy
warning in coregionalize: failed to import cython module: falling back to numpy
warning in choleskies: failed to import cython module: falling back to numpy

Uninstall, rm wheel and install again builds it using cython. Strange.

(gp_play) $ pip uninstall Gpy
(gp_play) $ fd -g "GPy*.whl" ~/.cache/pip | xargs rm
(gp_play) $ pip install Gpy
Collecting Gpy
  Using cached GPy-1.10.0.tar.gz (959 kB)
  Preparing metadata (setup.py) ... done
Requirement already satisfied: numpy>=1.7 in /home/elcorto/.virtualenvs/gp_play/lib/python3.10/site-packages (from Gpy) (1.24.1)
Requirement already satisfied: six in /home/elcorto/.virtualenvs/gp_play/lib/python3.10/site-packages (from Gpy) (1.16.0)
Requirement already satisfied: paramz>=0.9.0 in /home/elcorto/.virtualenvs/gp_play/lib/python3.10/site-packages (from Gpy) (0.9.5)
Requirement already satisfied: cython>=0.29 in /home/elcorto/.virtualenvs/gp_play/lib/python3.10/site-packages (from Gpy) (0.29.33)
Requirement already satisfied: scipy>=1.3.0 in /home/elcorto/.virtualenvs/gp_play/lib/python3.10/site-packages (from Gpy) (1.10.0)
Requirement already satisfied: decorator>=4.0.10 in /home/elcorto/.virtualenvs/gp_play/lib/python3.10/site-packages (from paramz>=0.9.0->Gpy) (5.1.1)
Building wheels for collected packages: Gpy
  Building wheel for Gpy (setup.py) ... done
  Created wheel for Gpy: filename=GPy-1.10.0-cp310-cp310-linux_x86_64.whl size=3229998 sha256=5043579109b96992f294d5bdc524a8f51419bb83ba99dffe9de3863fafc87bb9
  Stored in directory: /home/elcorto/.cache/pip/wheels/27/bd/9f/82ab4216eae088cba864ca0dc1d75699bd4bf6823790fb2f77
Successfully built Gpy
Installing collected packages: Gpy
Successfully installed Gpy-1.10.0

(gp_play) $ python book/content/gp_pred_comp/notebook_comp.py
No GPU/TPU found, falling back to CPU. (Set TF_CPP_MIN_LOG_LEVEL=0 and rerun for more info.)

Use `gpytorch.settings` as context managers

Seems like gpytorch.settings need to be used as a context manager in order to get applied.