The GPdoemd package has been tested and validated on OSX and Ubuntu.
No guarantees are provided that GPdoemd works on Windows-based systems.
Python 3.4+
- numpy >= 1.7
- scipy >= 0.17
- GPy
- gp_grief (forked repository): GP-GRIEF surrogate models
We recommend installing GPdoemd in a virtual environment.
Open the command prompt and navigate to your preferred installation folder.
To set up a new virtual environment called myenv (example name), run the following
python -m venv myenv
After the virtual environment has been created, activate it as follows
myenv\Scripts\activate
It is recommended that you update the pip installation in the virtual environment
python -m pip install --upgrade pip
First install all required packages in the virtual environment.
The required packages are listed above and in the file requirements.txt.
pip install numpy scipy six paramz matplotlib
Try running
pip install GPy
If you are lucky, this works. In our experience, the last command will fail - GPy will need to be built from source. Download the files from the GPy GitHub repository. Enter the downloaded GPy folder and run
python setup.py develop
If the setup is successful, skip to the installation of GPdoemd. If the previous command generates cython-related errors, try to comment out the lines in setup.py related to ext_mods, and add
ext_mods = []
After this, try running python setup.py develop
again.
To install GPdoemd, run the following in the virtual environment
pip install git+https://github.com/cog-imperial/GPdoemd
Alternatively, if you do not have git installed, download the files from the GitHub repository and install using python setup.py develop
.
The GPdoemd package can be uninstalled by running
pip uninstall GPdoemd
Alternatively, the folder containing the virtual environment can be deleted.