Installation¶

Linux¶

Conda¶

To install the library from conda-forge, simply run:

conda install -c conda-forge qpsolvers

PyPI¶

First, install the pip package manager, for example on a recent Debian-based distribution with Python 3:

sudo apt install python3-dev

You can then install the library by:

pip install qpsolvers

Add the --user parameter for a user-only installation.

Windows¶

Anaconda¶

First, install the Visual C++ Build Tools
Install your Python environment, for instance Anaconda
Install the library from conda-forge, for instance in a terminal opened from the Anaconda Navigator:

conda install -c conda-forge qpsolvers

Microsoft Visual Studio¶

Open Microsoft Visual Studio
Create a new project:
- Select a new “Python Application” project template
- Click “Next”
- Give a name to your project
- Click “Create”
Go to Tools → Python → Python Environments:
- To the left of the “Python Environments” tab that opens, select a Python version >= 3.8
- Click on “Packages (PyPI)”
- In the search box, type “qpsolvers”
- Below the search box, click on “Run command: pip install qpsolvers”
- A window pops up asking for administrator privileges: grant them
- Check the text messages in the “Output” pane at the bottom of the window
Go to the main code tab (it should be your project name followed by the “.py” extension)
Copy the example code from the README and paste it there
Click on the “Run” icon in the toolbar to execute this program

At this point a python.exe window should open with the following output:

QP solution: x = [0.30769231, -0.69230769, 1.38461538]
Press any key to continue . . .

Solvers¶

Open source solvers¶

To install at once all open source QP solvers available from the Python Package Index, run the pip command as follows:

pip install "qpsolvers[open_source_solvers]"

You can also install a subset of QP solvers of your liking, for instance:

pip install qpsolvers[clarabel,daqp,proxqp,scs]

Gurobi¶

Gurobi comes with a one-line pip installation where you can fetch the solver directly from the company servers:

python -m pip install -i https://pypi.gurobi.com gurobipy

This version comes with limitations. For instance, trying to solve a problem with 200 optimization variables fails with the following warning:

Warning: Model too large for size-limited license; visit https://www.gurobi.com/free-trial for a full license

COPT¶

COPT comes with an installation doc at COPT installation where you can install by pip:

python -m pip install coptpy

This version comes with limitations. For instance, trying to solve a problem with 200 optimization variables fails with the following warning:

No license found. Starting COPT with size limitations for non-commercial use
Please apply for a license from www.shanshu.ai/copt

HiGHS¶

The simplest way to install HiGHS is:

pip install highspy

If this solution doesn’t work for you, follow the Python installation instructions from the README.

PDHCG¶

You can install the GPU-accelerated PDHCG solver directly from PyPI:

pip install pdhcg

Note that PDHCG requires an NVIDIA GPU and CUDA 12.0+. If your system has multiple CUDA versions or the installation fails to find the compiler, you must explicitly point to your modern CUDA compiler using environment variables before installing:

export CUDACXX=/your/path/to/nvcc
export SKBUILD_CMAKE_ARGS="-DCMAKE_CUDA_COMPILER=/your/path/to/nvcc"
pip install pdhcg

quadprog¶

You can install the quadprog solver from PyPI:

pip install quadprog

This package comes with wheels to avoid recompiling the solver from source.

qpOASES¶

The simplest way to install qpOASES is via conda-forge:

conda install -c conda-forge qpoases

You can also check out the official qpOASES installation page for the latest release.