PyRecEst Documentation

PyRecEst is a Python library for recursive Bayesian estimation on Euclidean spaces and manifolds. The library includes distributions, filters, trackers, smoothers, samplers, evaluation helpers, and a backend abstraction for NumPy, PyTorch, and JAX.

These Markdown pages are the starting point for project documentation. They are kept lightweight so they can be read directly on GitHub while the project grows toward a generated documentation site.

Start Here

  • Getting started: install PyRecEst, run the examples, choose a backend, and set up a development checkout.
  • API overview: understand the main packages and where common public classes live.
  • Task tutorials: work through common distribution, filtering, tracking, and evaluation tasks.
  • Backend compatibility: choose between NumPy, PyTorch, and JAX and understand known support gaps.
  • Shapes and conventions: learn the expected state, measurement, covariance, batch, and manifold-coordinate shapes.
  • Examples: browse executable scripts that demonstrate basic workflows.
  • API reference: generated reference pages for the main public packages.

Current Documentation Scope

The README gives the shortest install and Kalman filter quickstart. The pages in this directory add more orientation, but the tests are still the most complete source of usage coverage for many advanced distributions, filters, trackers, smoothers, samplers, and evaluation helpers.

Good next documentation additions would be:

  • deeper convention notes for grids, state-space subdivisions, and advanced tracker outputs;
  • API-specific backend support tables for advanced distributions, trackers, evaluators, and utilities.