Venture

Venture is an interactive, Turing-complete probabilistic programming platform that aims to be sufficiently expressive, extensible, and efficient for general-purpose use.

http://probcomp.csail.mit.edu/venture/

Venture is rapidly-evolving, alpha quality research software. The key ideas behind its design and implementation have yet to be published. We are making Venture available at this early stage primarily to facilitate collaboration and support the emerging probabilistic programming community.

Installation

This release is for early adopter types who are willing to put up with much of the pain that a more mature software package would not impose. In particular, documentation is sparse and the user interface is unforgiving. Often, the only way to learn what's going on will be to ask us or to read the source code.

Ubuntu

sudo apt-get install -y libboost-all-dev libgsl0-dev ccache
sudo apt-get install -y python-matplotlib  # If you don't already have it
sudo apt-get install -y python-scipy       # If you don't already have it
pip install venture-0.4.2.tar.gz

OSX

We do not officially support installing Venture directly on OSX. Installation with pip should work, but we are not routinely testing it:

SKIP_PUMA_BACKEND=1 pip install venture-0.4.2.tar.gz

If this doesn't work or you want Puma, you could

• Run Venture in a Docker container (see our testing Docker file for inspiration), or

• Run Venture in a full Ubuntu virtual machine, or

• Have a look at how someone managed to get Venture running on a mac in September 2014

The Puma backend is an optional faster VentureScript engine written in C++.

Windows

We do not officially support installing Venture directly on Windows. Your best bet is likely to be running Venture in a Docker container or a virtual machine with a guest Ubuntu.

Checking that your installation was successful

For a quick check that should catch most installation problems, unpack the source distribution and run

./tool/check_capabilities.sh

in it, or

SKIP_PUMA_BACKEND=1 ./tool/check_capabilities.sh

if you didn't install the Puma backend.

For a more thorough check, you can run our development test suite.

This requires installing the test dependencies:

pip intall --find-links /path/to/tarball/directory venture[tests]

and then

./sanity_tests.sh

or

./sanity_tests.sh lite

if you didn't install the Puma backend.

Getting Started

• Interactive Venture console:

  venture
  

• Run a VentureScript program:

  venture -f <file>.vnts
  

• Venture as a library in Python:

  python -i -c 'from venture import shortcuts; ripl = shortcuts.Puma().make_ripl()'
  

  Using Venture as a library allows you to drive it programmatically.

• You can find several examples in the examples/ directory.

• There is a reference manual

Developing Venture

You can build a local version of the refence manual from the refman/ directory. This requires Sphinx; see instructions in the README there.

The interesting parts of the code are:

• The frontend stack (including SIVM, RIPL, server, and Python client) in python/.
• The pure-Python, clearer, normative Lite backend in backend/lite/.
• The C++, faster Puma backend (plus a thin Python driver) in backend/puma/.
• The test suite lives under test/.
• The actual entry points are in script/, notably script/venture.
• Advanced example programs live in examples/.
• There are some developer tools available in tool/.
• The Javascript client and web demos are in demos/jsripl (which corresponds to the erstwile VentureJSRIPL repository).
• There are language-level benchmarks (and correctness tests) in the VentureBenchmarksAndTests repository, but they may have bit rotted by now.