"Bullseye!" is a new algorithm for computing the Gaussian Variational Approximation of a target distribution. Its strong point lies in the fact that it can easily be parallelized and distributed.
These instructions will get you a copy of the project up and running on your local machine for development and testing purposes. See deployment for notes on how to deploy the project on a live system.
Bullseye! is now available as a PyPI package:
pip install bullseye_method
or clone the repository :
git clone https://github.com/Whenti/bullseye
or download and extract the zip into your project folder.
To see if everything is working properly, you can already run the algorithm on a multilogit model with artificially generated data.
from Bullseye.Tests import simple_test
simple_test()import Bullseye
from Bullseye import generate_multilogit
theta_0, x_array, y_array = generate_multilogit(d = 10, n = 10**3, k = 5)
bull = Bullseye.Graph()
bull.feed_with(x_array,y_array)
bull.set_predefined_model("multilogit")
bull.set_predefined_prior("normal_iid")
bull.init_with(mu_0 = 0, cov_0 = 1)
bull.set_options(local_std_trick = True, s=5)
bull.build()
bull.run()- Quentin Lévêque Whenti
See also the list of contributors who participated in this project. Hopefully, there will be more.
This project licensed under the GPL3 License - see the LICENSE.txt file for details.