Skip to content

Further reading

This concludes the tutorial.

The language documentation is a good next step to learn more about the Birch language. Perusing the standard library documentation will also give you a better idea of the features available, such as supported probability distributions.

For further reading on the design philosophy behind Birch, see:

For details of the delayed sampling heuristic that forms an important part of Birch, see:

For further examples, see:

  • Birch.Example for simple examples, including those in this tutorial.

  • VectorBorneDisease for some simple epidemiological models assembled from multiple classes.

  • MultiObjectTracking for a multiple object tracking example. This is particularly interesting to motivate universal probabilistic programming, as the model does not have fixed dimension (the number of objects is unknown), and the inference method consists of multiple Kalman filters nested within a single particle filter. Further details are in the paper Murray & Schön (2018) above.