Code for the article : "Kernel Stein Discrepancy Descent"

Compat

This package has been developed and tested with python3.7. It is therefore not guaranteed to work with earlier versions of python.

Install the repository on your machine

This package can easily be installed using pip, with the following command:

pip install -e .

This will install the package and all its dependencies, listed in requirements.txt.

Reproducing the figures of the paper

To run the ICA experiment, run

python examples/ica.py

To display the effects of different initializations, run

python examples/plot_init.py

To display the effects of several different initializations with different mixtures of gaussians, run

python examples/plot_minima.py

To compare the running times of algorithms, run

python examples/plot_quantitative_expe_gaussian.py

where you can change the number of trials n_expe and the number of particles n_samples

To obtain the trajectory of different algorithms on a simple Gaussian density, run

python examples/plot_simple_gaussian.py

To see the behavior of the algorithm on skewed mixture of Gaussian run

python examples/plot_skewed.py

To see the effect of the annealing policy, run

python examples/plot_temperature.py