This documentation is written with sphinx and markdown. To use markdown files in sphinx we use the myst parser.
Attention!
what wording to use?
- backend vs module
- interfaces like transpose, conversion, etc under linop or under matrix?
user guide structure:
user-guide/ contains top level .md files that are the entry point to a certain concept.
This top level file should contain the most important bits of information.
Details should be left to files under concept/file.md.
For example vectors.md mentions the creation function with arrays, but how to use this to get user data into vectors is done in vectors/user-data.md
reading order:
- getting-started: very high-level overview, introduces backends, example
- build-install: getting ginkgo running
- executors: introduce executor concept + specific executors
- vectors: creating and handling of (dense) vectors
- matrices: general creating and handling (apply) of matrices, link to matrix formats
- linear-operators: generalize matrices, polymorphic object stuff
- linear-solvers: iterative and direct solvers, preconditioners as links
- distributed: larger category, recaps vectors, linops, solvers for distributed settings
- object-configuration: description of config format
- logging: adding loggers to objects
- arrays: description of ginkgo arrays, is already referenced earlier
- extensions
- mixed-precision
.. toctree:: :maxdepth: 2 :hidden: :caption: User Guide: user-guide/faq user-guide/getting-started user-guide/build-install user-guide/executors user-guide/vectors user-guide/matrices user-guide/linear-operators user-guide/linear-solvers user-guide/distributed user-guide/object-configuration user-guide/logging Arrays <user-guide/arrays> user-guide/extensions user-guide/mixed-precision
.. toctree:: :maxdepth: 2 :hidden: :caption: Architecture: architecture/overview architecture/polymorphism architecture/adding-kernels architecture/lifetime