mogp_emulator is a Python package for fitting Gaussian Process Emulators to computer simulation results.
The code contains routines for fitting GP emulators to simulation results with a single or multiple target
values, optimizing hyperparameter values, and making predictions on unseen data. The library also implements
experimental design, dimension reduction, and calibration tools to enable modellers to understand complex
computer simulations.
mogp_emulator requires Python version 3.6 or later. The code and all of its dependencies can be
installed via pip:
pip install mogp-emulatorOptionally, you may want to install some additional packages. In particular, matplotlib is useful for
visualising some of the results of the benchmarks. These can be found in the
requirements-optional.txt file in the main repository.
Additionally, packages used in testing and developing the code base are listed in requirements-dev.txt.
The documentation is available at readthedocs for the current
builds of the master and devel versions, plus any previous tagged releases. The documentation
is available there in HTML, PDF, or e-Pub format.
This package is under active development by the Research Engineering Group at the Alan Turing Institute as part of several projects on Uncertainty Quantification. Questions about the code or any feedback on the usability and features that you would find useful can be sent to Eric Daub <edaub@turing.ac.uk>. If you encounter any bugs or problems with installing the software, please see the Issues tab on the Github page, and if the issue is not present, create a new one.
If you find this software useful, please consider taking part in its development! We aim to make this a welcoming, collaborative, open-source project where users of any background or skill levels feel that they can make valuable contributions. Please see the Code of Conduct for more on our expectations for participation in this project, and our Contributing Guidelines for how to get involved.