edit
GPy
GPy is the group’s main Python Gaussian process framework. Related packages and materials sit alongside it.
| Package |
Language |
Description |
| GPy |
Python |
Gaussian process framework — GPs, GP-LVM, Bayesian GP-LVM, and related models from the group and collaborators. |
| GPyOpt |
Python |
Bayesian optimisation built on GPy. |
| deepGPy |
Python |
Deep Gaussian processes using GPy. |
| gpy-gallery |
Notebooks |
Notebook gallery for the GPy software. |
Other software
| Package |
Language |
Description |
| lynguine |
Python |
Data-oriented architecture framework. |
| lamd |
Python |
Scripts for turning markdown into talks. |
| mlai |
Python |
Software for lectures on machine learning. |
| tig-code |
Python |
Code for the inaccessible game. |
| vibesafe |
Python |
Practices for improving quality and manageability of LLM co-created codebases. |
| fynesse |
Python |
Framework for data science analysis. |
| data-readiness |
— |
Suggestions and materials around data readiness levels. |
| referia |
Python |
Reviewing framework (originally for REF-style paper review). |
Before the move to Python, the shared research codebase was GPmat. The packages below are historical toolboxes and paper-code releases that build on that lineage (or parallel implementations in C++ and R).
| Package |
Language |
Description |
| GPmat |
MATLAB |
Core MATLAB toolbox for IVM, GPs, GP-LVM, and related models — the pre-Python group codebase. |
| GPc |
C++ |
IVM, GP, and GP-LVM software in C++. |
| gprege |
R |
Gaussian process ranking and estimation of gene-expression time series. |
| multigp |
MATLAB |
Multi-output Gaussian processes and latent force models, including sparse approximations. |
| vargplvm |
MATLAB |
Variational Bayesian GP-LVM. |
| deepGP |
MATLAB |
Deep Gaussian processes in MATLAB. |
| fgplvm |
MATLAB |
Faster GP-LVM software. |
| mocap |
MATLAB |
Tools for processing motion-capture files. |
Reproducible research
Since 2001 we have released code with (or before) paper submission. For why that matters in computational science, see Donoho et al. on WaveLab and Jon Claerbout’s notes on reproducible research.