tddlog for Python

Exact probabilistic inference from Python.

tddlog computes exact probabilities for ProbLog programs, LPADs, Dice programs, Circulette netlists and Bayesian and Markov networks. This package calls the same Rust implementation as the command line.

Install

pip install tddlog

Wheels cover CPython 3.10 and later on Linux and macOS (x86-64 and ARM64) and Windows x86-64. Elsewhere pip builds from source (Build from source, below).

Getting started

import tddlog

solution = tddlog.solve("""
    0.1::burglary.
    0.2::earthquake.
    0.7::hears_alarm(john).
    0.4::hears_alarm(mary).
    alarm :- burglary.
    alarm :- earthquake.
    calls(X) :- alarm, hears_alarm(X).
    evidence(calls(john), true).
    query(burglary).
    query(earthquake).
    query(calls(mary)).
""", evidence_probability=True)

for query, probability in solution.answers.items():
    print(f"{query}: {probability:.6f}")
print(f"P(e) = {solution.evidence_probability:.6f}")

prints

burglary: 0.357143
earthquake: 0.714286
calls(mary): 0.400000
P(e) = 0.196000

• solution.answers maps each query to its probability given the evidence, in the program's order. It is None where the evidence is inconsistent.
• format says what the text is: "problog" (the default, LPADs included), "dice", "circulette", or a network's "bif", "hugin", "xdsl" or "uai".
• evidence is the text of a network's evidence file (Ace's .inst or UAI's .evid).
• arguments is a list of strings that take the program's places $1, $2, ...
• vtree, evidence_probability and queries are the command line's --vtree, --evidence-probability and --no-queries.
• An error raises tddlog.ParseError, GroundingError or CompilationError, with the command line's message.
• solve releases the interpreter lock, so threads can solve several programs at once.

A network with evidence, in a checkout of the repository:

network = open("examples/car.net").read()
observed = open("examples/car.inst").read()
solution = tddlog.solve(network, format="hugin", evidence=observed)

The API reference documents each function and exception, and bindings/python/examples/ holds this program and one built from Python data. The guide describes the language, and its worked examples open in the playground. The website has the benchmark results.

Build from source

pip builds the package from the repository, for a commit after the release. This needs Python 3.10 or later, Rust 1.88 or newer (rustup installs it), a C++20 compiler and make. The guide's getting started says what to do if the build fails.

python3 -m venv .venv
. .venv/bin/activate
python -m pip install "git+https://github.com/Tractables/tddlog#subdirectory=bindings/python"

From a checkout, copy the license and the notices in first:

git clone --depth 1 https://github.com/Tractables/tddlog
cd tddlog
mkdir -p bindings/python/notices
cp LICENSE NOTICE THIRD-PARTY.md bindings/python/notices/
python -m pip install ./bindings/python

To work on the binding, build the extension in place and run its tests:

python -m pip install 'maturin>=1.9,<2' pytest
maturin develop --release --manifest-path bindings/python/Cargo.toml
python -m pytest bindings/python/tests

Acknowledgements

This material is based upon work supported by the Defense Advanced Research Projects Agency (DARPA) under Agreement No. HR0011269E224.