Feras Saad

Remember to handle the built-in scopes/blocks such as "all", "one", etc.

Because dimensionality change of conditional model, my concern is that the "trivial" Gibbs sampler is not going to be reversible and we need RJ-MCMC algortihm for which interface design is...

This is actually a research project.

Wrong commit msg.

Revisiting now that we are going to try and use the richness of data type modeling for important experiment. Simple univariate 2-sample test comparing the posterior predictive to true generator...

Test implemented in [disabled_test_simulate_univariate.py](https://github.com/probcomp/gpmcc/blob/master/tests/disabled_test_simulate_univariate.py). Uses two-sample tests for simulations from the posterior predictive and true generators. Disabled for now until I debug the tests more carefully.

Hi Lucas, Thanks for the detailed report; the Venture build is known to be hard to run from scratch. It can be installed directly on ubuntu 16.04, but also via...

@LSaldyt Okay, if you want a conda install see: https://github.com/probcomp/iventure/blob/master/docs/conda.md For a pip-based install, see: https://github.com/probcomp/iventure/blob/master/docs/virtualenvs.md The above instructions are tested assuming a fresh system of Ubuntu 16.04.

Hi @fritzo, thanks for writing us about Tree-Cat. I looked through the model and find it very interesting and creative. > Does cgpm have a standard engine interface that I...

The Engine class in general can be much improved, thank you for these suggestions.