GitHub - wrongu/sampling-dynamics: Source code for research project on mixing times in a sampling model of perception

Overview of codebase:

graphs.py: basic Graph and DiGraph classes
graphical_models.py: currently supports on Bayesian Networks with Discrete variables. The BayesNet class is a subclass of DiGraph
visualize.py: using networkx_interface.py and the graphviz module, makes diagrams of BayesNets
sampling.py: defines gibbs sampling functions
counting.py: useful combinatorics functions and some sampling procedures
util.py: other helper functions (e.g. load_or_run() which wraps precomputing/saving/loading numpy arrays)
models.py: functions to create special cases of Graphs or BayesNets
generators.py: create evidence-generators for sampling

Because of the way packages are set up here, from the project directory run python -m scripts.script_name without .py. For example:

$ python -m scripts.make_mixing_time_movie

Name		Name	Last commit message	Last commit date
Latest commit History 96 Commits
plots		plots
saved_data		saved_data
scripts		scripts
.gitignore		.gitignore
README.md		README.md
__init__.py		__init__.py
counting.py		counting.py
generators.py		generators.py
graphical_models.py		graphical_models.py
graphs.py		graphs.py
models.py		models.py
networkx_interface.py		networkx_interface.py
requirements.txt		requirements.txt
sampling.py		sampling.py
util.py		util.py
visualize.py		visualize.py