Installation

GBFKIT supports the idea that all software should be trivially installable and require minimal technical expertise from the end user.

Operating system requirements

The following operating systems are supported:

  • Linux: All Linux distributions released after year 2010
  • macOS: 10.14 (Mojave) - 10.15 (Catalina)

Note

While the software should, in principle, work on Microsoft Windows, there is no official support for this operating system yet. If you are a Windows user and you want to use GBKFIT, please give us a shout.

Python environment requirements

GBKFIT requires Python 3.7 or later.

Dealing with old Python environments

In case your Python version is older than the minimum required, and for some reason you cannot update it, you could try running the commands below.

On Linux:

mkdir gbkfit && cd gbkfit
wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh
chmod +x Miniconda3-latest-Linux-x86_64.sh
./Miniconda3-latest-Linux-x86_64.sh -b -s -p ./miniconda
./miniconda/bin/python -m venv venv

On macOS:

mkdir gbkfit && cd gbkfit
curl -O -L https://repo.anaconda.com/miniconda/Miniconda3-latest-MacOSX-x86_64.sh
chmod +x Miniconda3-latest-MacOSX-x86_64.sh
./Miniconda3-latest-MacOSX-x86_64.sh -b -s -p ./miniconda
./miniconda/bin/python -m venv venv

The above lines will create an isolated Python environment (venv) which will not affect your system’s Python environment and can be used to install and run GBKFIT.

To activate the environment, run:

. venv/bin/activate

To deactivate the environment, run:

deactivate

Attention

The above steps are just a suggestion and not required. It is up to the user to choose how to update their Python version. Furthermore, the above snippets use Minicoda for convenience. Miniconda is not a GBKFIT requirement.

Dependencies

All required run-time dependencies are installed automatically during GBKFIT’s installation.

Optional Dependencies

The following optional run-time dependencies can be installed by the user in order to activate additional functionality:

  • TODO

Installing from PyPI

Use the steps below to install GBKFIT from the Python Package Index (PyPI).

  • Upgrade pip to the latest version (optional step, but highly recommended):

    pip install pip -U

  • Install GBKFIT:

    pip install --index-url https://test.pypi.org/simple/ --extra-index-url https://pypi.org/simple --no-cache-dir gbkfit

Attention

Because GBKFIT is currently in Alpha development phase, it is located on the Test instance of PyPI. Once the Alpha phase is complete, the software will be moved to the production version of PyPI and will be installable using: pip install gbkfit.

Installing from source

Installing GBKFIT from source requires additional technical expertise and you should have a good reason for preferring it over installing from PyPI. The two most common reasons are:

  • GBKFIT is not available on PyPI for your platform
  • You want to compile GBKFIT with a specific compiler

To install GBKFIT from source you will need:

  • A C++ 14 capable compiler. Any recent version of GCC, Clang, ICC, or PGI will do.
  • The FFTW3 library. This should be available through your OS package manager. Alternatively, it can be obtained from here.
  • An OpenMP library. This usually comes with your compiler and you do not have to install anything. However, this is not always the case. For example, when compiling with Apple Clang compiler, you may have to install the libomp library (available through Homebrew and MacPorts).

Once all required dependencies are installed, run:

git clone --recurse-submodules --remote-submodules https://github.com/bek0s/gbkfit.git

Before compiling the source code, we need to specify what hardware support we want to compile GBKFIT with.

To enable multi-core CPU support, run:

export GBKFIT_BUILD_HOST=1

To enable CUDA GPU support, run:

export GBKFIT_BUILD_CUDA=1

Attention

Support for CUDA GPUs is not fully functional yet. Do not enable it.

To compile and install your local copy of GBKFIT, run:

pip install ./gbkfit

Congratulations! Now it is time to model some galaxies!