PyQBPP: Python Interface for QUBO++

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

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
• CUDA-enabled NVIDIA GPUs
• Python 3.8 or later

Installation

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

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

The wheel contains only the Python code. The shared libraries (about 330 MB) are downloaded automatically from GitHub Releases the first time you run a PyQBPP program, verified against the SHA-256 recorded in the wheel, and kept in ~/.cache/pyqbpp/ (set PYQBPP_CACHE_DIR to change it). To download them in advance, run qbpp-license:

$ qbpp-license

For machines without internet access, see the installation guide.

Without a license key, the number of binary variables is limited to 100.

To remove this limit, register from the User Portal to obtain a trial license key.

Once you have a license key, activate it as follows:

$ qbpp-license -k YOUR-LICENSE-KEY -a

Quick Example

The following program solves a simple QUBO problem by expanding and minimizing the expression f = (a + 2b + 3c - 4)^2.

import pyqbpp as qbpp

a = qbpp.var("a")
b = qbpp.var("b")
c = qbpp.var("c")
f = qbpp.sqr(a + 2 * b + 3 * c - 4)
f = qbpp.simplify_as_binary(f)
print("f =", f)

solver = qbpp.EasySolver(f)
sol = solver.search(time_limit=10, target_energy=0)
print("sol =", sol)

Output:

f = 16 -7*a -12*b -15*c +4*a*b +6*a*c +12*b*c
sol = Sol(energy=0, {a: 1, b: 0, c: 1})

Documentation

https://qubo-plus.github.io/python/