Installation instructions

These are the instructions for building QPALM from source. If you just want to install the pre-built binaries, please see the main page.

Prerequisites

You'll need a C compiler, CMake, and Make or Ninja. To compile the Matlab interface, you need Matlab and MEX as well. For the Python interface, you need the Python development files, which are not installed by default on some systems. For example, on Ubuntu, install the python3-dev package.

Download

Download QPALM from https://github.com/kul-optec/QPALM:

git clone https://github.com/kul-optec/QPALM.git --single-branch --depth 1 --recursive
cd QPALM

If you already cloned it without the --recursive argument, use:

git submodule update --init --recursive

Matlab installation

Build

Open a terminal inside of the QPALM repository, and configure and build the project:

cmake -B build -S QPALM -D CMAKE_BUILD_TYPE=Release -D QPALM_WITH_MEX=On -D CMAKE_POSITION_INDEPENDENT_CODE=On

cmake --build build --config Release -j

Install

On Linux, Matlab automatically adds ~/Documents/MATLAB to the path, so it's easiest install QPALM there:

cmake --install build --config Release --component mex_interface --prefix ~/Documents/MATLAB

Uninstall

To uninstall QPALM, simply remove the +qpalm folder from where you installed it:

rm -r "~/Documents/MATLAB/+qpalm"

Python installation

QPALM is available on PyPI, with binaries for Windows (x86_64), macOS (x86_64) and Linux (x86_64 and arm64). To install it, use:

python3 -m pip install qpalm

To build the Python bindings yourself, create a virtual environment:

# Linux/macOS
python3 -m venv .venv
. ./.venv/bin/activate

# Windows
python3 -m venv .venv
./.venv/Scripts/Activate.ps1

Then install Eigen:

# Linux/macOS
./scripts/install-eigen.sh

# Windows
./scripts/install-eigen.ps1

And finally, install the package:

pip install .

Julia installation

First, build the binary JLL package using BinaryBuilder.jl, and deploy it locally (to your ~/.julia/dev folder):

julia QPALM/interfaces/julia/build_tarballs.jl $(gcc -print-multiarch) --deploy=local
julia -e "using Pkg; Pkg.develop(path=\"$HOME/.julia/dev/QPALM_jll\")"

Then install the development version of QPALM.jl as well:

cd # wherever you want to download the source code
git clone https://github.com/kul-optec/QPALM.jl.git
julia -e "using Pkg; Pkg.develop(path=\"$PWD/QPALM.jl\")"

You can run the tests using:

julia QPALM.jl/test/runtests.jl

Usually, you'll use BinaryBuilder.jl instead of compiling it manually, but if you need to, you can:

cmake -B build -S QPALM \
    -D CMAKE_BUILD_TYPE=Release \
    -D QPALM_WITH_JULIA=On \
    -D CMAKE_POSITION_INDEPENDENT_CODE=On
cmake --build build --config Release -j
cmake --install build --config Release --component julia_modules --prefix $PWD/staging-julia
# staging-julia/lib/libqpalm_jll.so now exists

C/C++/Fortran installation

To install the C and C++ libraries and headers, simply follow the standard CMake configure, build, install instructions:

cmake -B build -S QPALM \
    -D CMAKE_BUILD_TYPE=Release \
    -D QPALM_WITH_CXX=On \
    -D CMAKE_POSITION_INDEPENDENT_CODE=On
cmake --build build --config Release -j
cmake --install build --config Release --prefix /usr/local
cmake --install build/LADEL --prefix /usr/local

If you just need the shared libraries, you can use:

cmake -B build -S QPALM \
    -D CMAKE_BUILD_TYPE=Release \
    -D QPALM_WITH_CXX=On \
    -D CMAKE_POSITION_INDEPENDENT_CODE=On \
    -D BUILD_SHARED_LIBS=On
cmake --build build --config Release -j
cmake --install build --config Release --component shlib --prefix /usr/local
cmake --install build/LADEL --component shlib --prefix /usr/local

To build the Fortran interface as well, simply add the option -D QPALM_WITH_FORTRAN=On to the first CMake command.

For the C++ interface, you'll need to have the Eigen library installed. You can find installation instructions in the scripts provided by the QPALM repository (scripts/install-eigen.{sh,ps1}), or install it through your system's package manager, for example:

sudo apt install libeigen3-dev