Banded Matrices

Overview

A library providing C++ linear algebra operators (matmul, solve, ...) dedicated to banded matrices and a TensorFlow interface. This extends the set of existing TensorFlow operators which as of August 2021 only include banded_triangular_solve.

Details on the implemented operators may be found in Durrande et al.: "Banded Matrix Operators for Gaussian Markov Models in the Automatic Differentiation Era", and in Adam et al.: "Doubly Sparse Variational Gaussian Processes"

Installation

For users

To install the latest (stable) release of the toolbox from PyPI, use pip:

$ pip install banded_matrices

For contributors

This project uses Poetry to manage dependencies in a local virtual environment. To install Poetry, follow the instructions in the Poetry documentation.

To install this project in editable mode, run the commands below from the root directory of the banded_matrices repository.

poetry install

This command creates a virtual environment for this project in a hidden .venv directory under the root directory.

You must also run the poetry install command to install updated dependencies when the pyproject.toml file is updated, for example after a git pull.

The project currently supports Python 3.10, 3.11, and 3.12. If Poetry selects a different Python version, use the command below to point it at a supported interpreter before installing:

poetry env use python3.12 && poetry install

NOTE: Unlike most other Python packages, by installing the banded_matrices package from source you will trigger a compilation of the C++ TensorFlow ops library. This means that running poetry install can take a while - in the order of 5 minutes, depending on the machine you are installing onto.

Known issues

Source installs require a working C++ compiler and CMake, and the build links against the TensorFlow version installed in the project environment. If you are only using the package as a dependency, prefer installing a released wheel with pip.

Running the tests

Run these commands from the root directory of this repository. To run the full Python test suite, including pylint and Mypy, run:

poetry run task test

Alternatively, you can run just the unit tests, starting with the failing tests and exiting after the first test failure:

poetry run task quicktest

To run linting of the C++ code (using cpplint), run:

poetry run task cpplint

NOTE: Running the tests requires that the project virtual environment has been updated. See Installation.

The Secondmind Labs Community

Getting help

Bugs, feature requests, pain points, annoying design quirks, etc: Please use GitHub issues to flag up bugs/issues/pain points, suggest new features, and discuss anything else related to the use of banded_matrices that in some sense involves changing the banded_matrices code itself. We positively welcome comments or concerns about usability, and suggestions for changes at any level of design. We aim to respond to issues promptly, but if you believe we may have forgotten about an issue, please feel free to add another comment to remind us.

Slack workspace

We have a public Secondmind Labs slack workspace. Please use this invite link and join the #banded_matrices channel, whether you'd just like to ask short informal questions or want to be involved in the discussion and future development of banded_matrices.

Contributing

All constructive input is very much welcome. For detailed information, see the guidelines for contributors.

Maintainers

Banded_matrices was originally created at Secondmind Labs and is now maintained by (in alphabetical order) Vincent Adam, Artem Artemev. We are grateful to all contributors who have helped shape banded_matrices.

Banded_matrices is an open source project. If you have relevant skills and are interested in contributing then please do contact us (see "The Secondmind Labs Community" section above).

We are very grateful to our Secondmind Labs colleagues, maintainers of GPflow, GPflux, Trieste and Bellman, for their help with creating contributing guidelines, instructions for users and open-sourcing in general.