Installing tofu

As of March 2021, tofu is still under development. Therefore, installation instructions are provided officialy for Linux and ITER users. However, instructions for other platforms are also provided since they have been known to work out for some users. If installation fails for your use-case, please open an issue over at Github. so that we can try adressing it.

Note

If you encounter problems during the installation, check our list of known bugs or open an issue..

Linux

To install and use tofu on Linux, we recommend to proceed in two steps: install the Python package manager conda and then install tofu. We recommend Miniconda (light version of the Anaconda Python distribution for data science, but you can also work with pip or another Python package manager of your choice).

$ conda install -c tofuproject tofu

  • Check that tofu works by printing its version number:

$ python -c "import tofu; print(tofu.__version__)"

Now you can follow a tutorial.

Mac OS X

See Installing tofu as a developer.

Additional caveat: if you are using a version of gcc < 8 be sure to turn off all parallelizations since there is a known bug with cython.

Windows

See Installing tofu as a developer.

Additional caveat: you may need to open an Anaconda prompt (usually found by pressing the Windows key) to run the commands described in the linked section.

Using tofu on a ITER cluster

If you have an ITER account, you can use tofu directly from ITER Computing Cluster. No need to install tofu !

  • Ask for access to ITER servers, if you don’t have them already.

  • For information about the cluster, see this link.

  • Open a new terminal and connect to server (see link above)

  • Create a new file in your $HOME directory, you can name it load_tofu_modules.sh

  • Open it and add the following lines:

module refresh
source /etc/profile.d/modules.sh # make sure you have the module environment
module purge # unload any previously loaded modules
module load IMAS/3.24.0-4.1.5 # for IMAS data base
module load IPython/6.3.1-intel-2018a-Python-3.6.4 # for iPython
module load PySide2/5.12.0-intel-2018a-Python-3.6.4
module load ToFu/1.4.0-intel-2018a-Python-3.6.4 # Load tofu :)

  • Convert it to an executable, from the terminal: chmod +x ~/load_tofu_modules.sh

  • Execute it: ./load_tofu_modules.sh

  • If you are going to use tofu often, you might want to add the execution of the script to your .bash_profile (or .bashrc file):

echo './load_tofu_modules.sh' >> .bash_profile

You are all set, open a Python/IPython console and try importing tofu.

$ python
In [1]: import tofu as tf

You can now follow a tutorial.

Installing tofu as a developer

To install tofu as a developer, we recommend using the conda ecosystem (Miniconda in particular):

$ conda create -n tofu3 python=3.6 scipy numpy cython git
$ conda activate tofu3

  • Move to where you would like to install your local copy of ToFu $ cd some_path

  • $ git clone https://github.com/ToFuProject/tofu.git (make sure you remember the path where you are installing, if you want to install it into your home repository, just make sure to $ cd ~ before the git clone...)

  • Move to the “cloned” tofu directory that has been created by the git clone command: cd ~/tofu

  • Switch to the git branch you will be working on. If you are just starting you probably want to start from the latest develop branch: git checkout devel. If you are not familiar with git take a look at this tutorial (long) or this short one

  • Run pip install -e .[dev]. This will install dependencies, compile the tofu cython extensions and install it into your conda environment while you can still modify the source files in the current repository.`

  • Make sure tofu tests are running by typing pytest tofu/tests