1) INSTALL
The implementation was tested with python 3.8.4 and tensorflow 2.4 but should also work with similar version.
You can run "pip install -r requirements.txt" to install the necessary packages (virtualenv is recommended)

2) Usage
run python run_paper_experiments.py --help for possible arguments. Specifying the Experiment and Algorithm should automatically select the correct hyperparameters.

3) Code Structure
- distributions/: Contains implementations of GMM, Gaussians and Categorical
- experiments/: Contains target density functions (under target_lnpdfs) and run-scripts for the individual experiments (under VIPS)
- optimization/: Contains implementations of the actual algorithms. VOGN, VON and GM are implemented in VonVIPS (which extends VIPS),
                gMORE and MORE are implemented in VIPS
- plotting/: code for visualizing the planar robot experiment
- rec/: uninteresting
- recording/: uninteresting
- util/ main contains operations on Gaussians (in distributions.py)
