Building From Source

JAX & GPU

PyAutoLens runs significantly faster on GPUs — often 50x or more compared to CPUs.

This acceleration is achieved through [JAX](https://docs.jax.dev/en/latest/notebooks/thinking_in_jax.html), which provides GPU and TPU support.

When you install PyAutoLens (see instructions below), JAX will also be installed. However, the default installation may not include GPU support.

To ensure GPU acceleration, it is recommended that you install JAX with GPU support before installing PyAutoLens, by following the official [JAX installation guide](https://jax.readthedocs.io/en/latest/installation.html).

If you install PyAutoLens without a proper GPU setup, a warning will be displayed.

Packages

Building from source means that you clone (or fork) the PyAutoLens GitHub repository and run PyAutoLens from there. Unlike conda and pip this provides a build of the source code that you can edit and change, to contribute the development PyAutoLens or experiment with yourself!

A large amount of PyAutoLens functionality is contained in its parent projects:

PyAutoFit https://github.com/PyAutoLabs/PyAutoFit

PyAutoArray https://github.com/PyAutoLabs/PyAutoArray

PyAutoGalaxy https://github.com/PyAutoLabs/PyAutoGalaxy

If you wish to build from source all code you may need to build from source these 3 additional projects. We include below instructions for building just PyAutoLens from source or building all projects.

Building Only PyAutoLens

We upgrade pip to ensure certain libraries install:

pip install --upgrade pip

First, clone (or fork) the PyAutoLens GitHub repository:

git clone https://github.com/PyAutoLabs/PyAutoLens

Next, install the PyAuto parent projects via pip:

pip install autoconf
pip install autofit
pip install autoarray
pip install autogalaxy

Next, install the PyAutoLens dependencies via pip:

pip install -r PyAutoLens/requirements.txt

Next, install the optional dependency numba via pip (see this link for a description of numba):

pip install numba

For unit tests to pass you will also need the following optional requirements:

pip install pynufft

If you are using a conda environment, add the source repository as follows:

[NOTE: Certain versions of conda use the command conda develop (without a dash) instead of those shown below.]

conda-develop PyAutoLens

Alternatively, if you are using a Python environment include the PyAutoLens source repository in your PYTHONPATH (noting that you must replace the text /path/to with the path to the PyAutoLens directory on your computer):

export PYTHONPATH=$PYTHONPATH:/path/to/PyAutoLens

Finally, check the PyAutoLens unit tests run and pass (you may need to install pytest via pip install pytest):

cd /path/to/PyAutoLens
python3 -m pytest

Building All Projects

We upgrade pip to ensure certain libraries install:

pip install --upgrade pip

First, clone (or fork) all 4 GitHub repositories:

git clone https://github.com/PyAutoLabs/PyAutoFit
git clone https://github.com/PyAutoLabs/PyAutoArray
git clone https://github.com/PyAutoLabs/PyAutoGalaxy
git clone https://github.com/PyAutoLabs/PyAutoLens

Next, install PyAutoConf via pip:

pip install autoconf

Next, install the source build dependencies of each project via pip:

pip install -r PyAutoFit/requirements.txt
pip install -r PyAutoArray/requirements.txt
pip install -r PyAutoGalaxy/requirements.txt
pip install -r PyAutoLens/requirements.txt

Next, install the optional dependency numba via pip (see this link for a description of numba):

pip install numba

For unit tests to pass you will also need the following optional requirements:

pip install -r PyAutoArray/optional_requirements.txt

If you are using a conda environment, add each source repository as follows:

[NOTE: Certain versions of conda use the command conda develop (without a dash) instead of those shown below.]

conda-develop PyAutoFit
conda-develop PyAutoArray
conda-develop PyAutoGalaxy
conda-develop PyAutoLens

Alternatively, if you are using a Python environment include each source repository in your PYTHONPATH (noting that you must replace the text /path/to with the path to the each directory on your computer):

export PYTHONPATH=$PYTHONPATH:/path/to/PyAutoFit
export PYTHONPATH=$PYTHONPATH:/path/to/PyAutoArray
export PYTHONPATH=$PYTHONPATH:/path/to/PyAutoGalaxy
export PYTHONPATH=$PYTHONPATH:/path/to/PyAutoLens

Finally, check the unit tests run and pass for each project (you may need to install pytest via pip install pytest):

cd /path/to/PyAutoFit
python3 -m pytest
cd ../PyAutoArray
python3 -m pytest
cd ../PyAutoGalaxy
python3 -m pytest
cd ../PyAutoLens
python3 -m pytest