probml / pyprobml

Python code for "Probabilistic Machine learning" book by Kevin Murphy
MIT License
6.53k stars 1.53k forks source link

apply GTSAM to LGSSM and compare to JSL code #754

Closed murphyk closed 2 years ago

murphyk commented 2 years ago

The GTSAM library is a C++ library with Python wrappers for performing MAP estimaiton and (marginal) posterior inference in Gaussian graphical models. If the model is linear, the inference is exact; if it is nonlinear, it is approximate. The methods generalizes Kalman smoothing to arbitrary graphs.

The goal of this exercise is to reproduce the results of JSL kalman smoother and Kalman sampler using GTSAM applied to some toy problem.

dellaert commented 2 years ago

Hey, here is code for Kalman smoothers:

And here is a cool pic from the second one:

image