Installation

qpax runs on CPU out of the box and on NVIDIA GPUs through JAX's CUDA wheels. Pick the variant that matches your hardware, or install from source if you plan to read or modify the code.

Quick install

Install the latest release from PyPI with pip:

CPUNVIDIA GPU (CUDA 12)NVIDIA GPU (CUDA 13)

pip install qpax

pip install "qpax[cuda12]"

pip install "qpax[cuda13]"

The GPU variants pull in the matching jax[cuda12] / jax[cuda13] extras. You only need one — match it to the CUDA toolkit on your machine.

Installing from source

Clone the repository and install in editable mode:

git clone https://github.com/qpax-solver/qpax
cd qpax

Then pick the variant you want:

MinimalDevelopmentNVIDIA GPU (CUDA 12)NVIDIA GPU (CUDA 13)

pip install -e .

Includes test, linting, and documentation dependencies.

pip install -e ".[dev]"

pip install -e ".[cuda12]"

pip install -e ".[cuda13]"

Verify installation

Run the test suite from the repository root:

python -m pytest test

A clean run prints a summary line ending in passed. If you only want a quick sanity check, this single import is enough:

import qpax
import jax.numpy as jnp

Q = jnp.eye(2)
q = jnp.array([1.0, -1.0])
A = jnp.zeros((0, 2)); b = jnp.zeros(0)
G = -jnp.eye(2);       h = jnp.zeros(2)

x, *_ = qpax.solve_qp(Q, q, A, b, G, h)
print(x)