A Python wrapper for the MiniZinc tool pipeline.
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.
docs docs: update docs Jun 3, 2018
examples/knapsack examples: move knapsack into knapsack01 Nov 18, 2017
CHANGES.rst CHANGES: add 0.16.0 changes Nov 18, 2017
README.rst README: fix readme Nov 18, 2017



PyMzn is a Python library that wraps and enhances the MiniZinc tools for constraint programming. PyMzn is built on top of the minizinc toolkit and provides a number of off-the-shelf functions to readily solve problems encoded with the MiniZinc language and parse the solutions into Python objects.


First, we need to define a constraint program via MiniZinc. Here is a simple 0-1 knapsack problem encoded with MiniZinc:

%% knapsack01.mzn %%
int: n;                     % number of objects
set of int: OBJ = 1..n;
int: capacity;              % the capacity of the knapsack
array[OBJ] of int: profit;  % the profit of each object
array[OBJ] of int: size;    % the size of each object

var set of OBJ: x;
constraint sum(i in x)(size[i]) <= capacity;
var int: obj = sum(i in x)(profit[i]);
solve maximize obj;

%% knapsack01.dzn %%
n = 5;
profit = [10, 3, 9, 4, 8];
size = [14, 4, 10, 6, 9];

You can solve the above problem using the pymzn.minizinc function:

import pymzn
s = pymzn.minizinc('knapsack01.mzn', 'knapsack01.dzn', data={'capacity': 20})

The result will be:

[{'x': {3, 5}}]

The returned object is a lazy solution stream, which can however be directly referenced as a list. The minizinc function takes care of flattening the MiniZinc model, launching the solver, and parsing the solutions into Python dictionaries.

PyMzn is also able to:

  • Convert python objects to dzn format and back;
  • Interface with many different solvers;
  • Perform dynamic modelling through a Python interface or by embedding code from the Jinja2 templating language;
  • Safely parallelize several instances of the same problem;
  • and more ...

For a follow-up of the previous example, read the Quick Start guide.

For more information on the PyMzn classes and functions refer to the reference manual.


PyMzn can be installed via Pip:

pip install pymzn

or from the source code available on GitHub:

python setup.py install

Currently, PyMzn is developed and maintained in Python 3.5 with a porting to Python 2.7 at every release (the python2 branch does not always contain the most recent changes).


PyMzn requires the MiniZinc toolkit to be installed on your machine, along with at least one solver. The easiest way to install MiniZinc is to download the MiniZincIDE package, which contains both the MiniZinc binaries and several solvers. After downloading the package, make sure the executables are visible to PyMzn by either setting the PATH environment variable or by configuring it using the pymzn.config module.

For more details take a look at the Install section in the documentation.


If you find a bug or think of a useful feature, please submit an issue on the GitHub page of PyMzn.

Pull requests are very welcome too. If you are interested in contributing to the PyMzn source code, read about its implementation details.


Paolo Dragone, PhD student at the University of Trento (Italy).