Skip to content
Snippets Groups Projects
Select Git revision
  • nifty5_to_nifty7
  • master default protected
  • coursematerial
  • addnoiseinference
  • 2019_aachen
5 results

nifty_tutorial

  • Clone with SSH
  • Clone with HTTPS
  • Installation

    There are two methods to install the environment which is needed for following the NIFTy tutorial. First, install it regularly into the home folder of your machine. Second, use a Docker container. The advantage of the second approach is that the chance that it runs on a variety of machines is relatively high. The downsides are: You need root access to build and start the Docker container (security implications), a full Docker installation (security implications) and you need to be familiar with Docker in order to get your results out of the container.

    Standard

    • Install git, python3 and python3-pip, python3-scipy, python3-matplotlib. On Debian-based systems possibly:
    # apt-get update
    # apt-get install git python3-pip python3-scipy python3-matplotlib 
    • Optionally, install our Fourier transform package ducc0:
    $ pip3 install ducc0
    • Clone this repository:
    $ git clone --recursive https://gitlab.mpcdf.mpg.de/ift/nifty_tutorial.git
    • Install nifty:
    $ cd nifty_tutorial/nifty
    $ pip3 install .

    Docker

    • To build the image, run the following command inside the nifty_tutorial folder.
    # docker build -t niftytutorial .
    • Start a container and mount your local folder /mnt into the folder /mnt in the container. Do this only if you know what you are doing!
    # docker run -v /mnt:/mnt -it niftytutorial

    Double-check the installation

    Run the scripts *_solution.py from the repository, check that they run through and look at the output plots.