PyQUBO++: Python Interface for QUBO++

PyQUBO++ is a Python wrapper for the QUBO++ library, allowing you to model and solve combinatorial optimization problems using QUBO/HUBO formulations directly from Python.

Note: PyQUBO++ is currently in alpha. The API may change without notice, and there may be bugs. Please report any issues to the author.

Features

Symbolic construction of QUBO/HUBO expressions in Python
Access to QUBO++ solvers (Easy Solver, Exhaustive Solver, ABS3)
Familiar Python syntax with the full power of the QUBO++ engine

Supported Environment

Linux (Ubuntu 20.04 or later)
x86_64 or arm64 (aarch64) CPUs
Python 3.8 or later

Installation

PyQUBO++ is available on PyPI. We recommend using a Python virtual environment (venv) to install PyQUBO++. No sudo privileges are required.

$ python3 -m venv ~/qbpp-env
$ source ~/qbpp-env/bin/activate
$ pip install pyqbpp

After installation, activate your QUBO++ license. Set the QBPP_LICENSE_KEY environment variable to your license key and run qbpp-license -a. If QBPP_LICENSE_KEY is not set, an anonymous trial license will be activated.

$ export QBPP_LICENSE_KEY=[Your QUBO++ license key]
$ qbpp-license -a

Quick Example

The following program finds an 8×8 binary matrix where each row and each column contains exactly one 1 (a one-hot constraint).

from pyqbpp import EasySolver, var, sum, vector_sum

n = 8
x = var("x", n, n)

# Each row and column has exactly one 1
f = sum(vector_sum(x, 0) == 1) + sum(vector_sum(x, 1) == 1)

f.simplify_as_binary()

solver = EasySolver(f)
solver.target_energy(0)
sol = solver.search()

for i in range(n):
    print([sol(x[i][j]) for j in range(n)])

Example output:

[0, 0, 0, 0, 0, 1, 0, 0]
[1, 0, 0, 0, 0, 0, 0, 0]
[0, 0, 0, 1, 0, 0, 0, 0]
[0, 0, 1, 0, 0, 0, 0, 0]
[0, 0, 0, 0, 0, 0, 1, 0]
[0, 1, 0, 0, 0, 0, 0, 0]
[0, 0, 0, 0, 1, 0, 0, 0]
[0, 0, 0, 0, 0, 0, 0, 1]

Last updated: