FilterPy

A collection of Bayesian filtering methods in Python using Numpy and Scipy

FilterPy allows users to filter and track state space models using various Bayesian inference methods.

Installing FilterPy

To install FilterPy, follow these steps:

pip install pip@git+https://github.com/mjcarter95/FilterPy.git

Using FilterPy

A number of example problems are provided in the examples folder.

Example: Linear Gaussian State Space Model

  # Instantiate the measurement and observation models
  transition_model = model.TransitionModel(F, Q)
  measurement_model = model.MeasurementModel(H, R)

  # Simulate the state and observation sequences
  x_true, y = lgssm.simulate_data(T, transition_model, measurement_model)

  # Instantiate the Kalman filter
  kf = BasicKalmanFilter(transition_model, measurement_model)

  # Initialise the state and state covariance
  x_hat = np.zeros((T, 1))
  P = np.zeros((T, 1))

  # Set the initial state and state covariance
  x_hat[0] = np.random.multivariate_normal(np.zeros(1), np.eye(1))
  P[0] = np.array([[0.5]]])

  # Run the Kalman filter
  for t in range(1, T):
      x_pred, P_pred = kf.predict(x_hat[t-1], P[t-1])
      x_hat[t], P[t] = kf.update(x_pred, P_pred, y[t])

Contributing to FilterPy

To contribute to FilterPy, follow these steps:

1. Fork this repository.
2. Create a branch: git checkout -b <branch_name>.
3. Make your changes and commit them: git commit -m '<commit_message>'
4. Push to the original branch: git push origin <project_name>/<location>
5. Create the pull request.

Alternatively see the GitHub documentation on creating a pull request.

Contact

If you want to contact me you can reach me at <m (dot) j (dot) carter (at) liverpool (dot) ac (dot) uk>.

Citation

We appreciate citations as they let us discover what people have been doing with the software.

To cite FilterPy in publications use:

Carter, M. (2023). FilterPy (1.0.0). https://github.com/mjcarter95/FilterPy

Or use the following BibTeX entry:

@misc{filterpy,
  title = {FilterPy (1.0.0)},
  author = {Carter, Matthew},
  year = {2023},
  month = May,
  howpublished = {GitHub},
  url = {https://github.com/mjcarter95/FilterPy}

}

Acknowledgements

The University of Liverpool and IBM Research for co-funding the EPSRC CDT in Distributed Algorithms doctoral project titled Uncertain Heterogeneous Algorithmic Teamwork