Developer documentation: Document GPy architecture

[bobturneruk]

The GPy documentation could perhaps be improved by more information on how the various parts of GPy fit together and operate in order to do their jobs. This might include:

• Concept of operation - how does GPy implement Gaussian Processes?
• How are the key features of the package organised in code?
• How do the classes relate to each other? Diagrams?
• What are the important classes?

I will self assign this issue and begin work on it on a fork (https://github.com/RSE-Sheffield/GPy), and ultimately make a pull request as per https://github.com/SheffieldML/GPy#contributing-to-gpy. Clearly, whatever I produce must be acceptable to the approvers on this project if it's to be incorporated (which I hope it can be) so I'd value any input and feedback from this direction.

To introduce myself to the community - I'm a Research Software Engineer working on a project for @maalvarezl. The documentation I'm proposing is necessary for our project, so I'll be putting it together anyway. It might as well (hopefully) be of some benefit to the broader community.

Obviously, if anyone has any prior work on this, I'd love to see it!

[lawrennd]

Fantastic, would love to have you contribute to this!

[maalvarezl]

Thanks, Bob.

[zhenwendai]

Cool! It is a great idea. Documentation has been a weakness of GPy. If you run into problems, you can email me or Max (@mzwiessele).

[bobturneruk]

Great. I've set up a work in progress readthedocs https://rse-sheffield-gpy.readthedocs.io/en/docs/. This is my WIP branch https://github.com/RSE-Sheffield/GPy/tree/docs