Pymor Workshop

Clone this repository to get access to the slides and practice problems:

$ git clone https://github.com/esm-tools/pymor_workshop.git
$ cd pymor_workshop
$ mkdir work

Make a folder to work in, for example:

$ cd work

During the live workshop, you'll be able to use the QOS --qos=training in DKRZ's HPC, to gain priority in the slurm queue. To benefit from this make sure you add this to your salloc or sbatch commands, or to the #SBATCH headers of the scripts.

---

Outline

1. Intro

• Workshop intro - Slides, Video
  • Aims
  • Outline
• Pymor - Slides, Video
  • What is Pymor?
  • Main Pymor features
• Demo: pymor cli, Video
• Time to resolve installation problems @pgierz
  • Have you installed Pymor inside a conda environment named pymor already?
  • Can you run pymor --help and other commands?
  • Maybe you need to upgrade (pip install --upgrade py-cmor[dev,fesom])

2. Pymor syntax

• Presentation - Slides, Video
• Exercise: Running a basic pymor process (text and solutions), Video

3. Features

3.1. Units (module_units and cmor_units)

• Presentation - Slides, Video
• Exercise: Unit conversion (text and solutions), Video

3.2. Time span for output files (file_timespan)

• Presentation - Slides, Video
• Exercise: File timespan (text and solutions), Video

3.3. Time averages

• Presentation - Slides, Video
• Exercise: Temporal frequency, Video

3.4. Adjusting time stamps of steps

• Presentation - Slides, Video
• Exercise: Adjust timestamp, Video

3.5. Control computing resources

• Presentation - Slides, Video
• Exercise: Costumizing your compute resources, Video

3.6. Pipelines and custom steps

• Presentation - Slides, Video
• Exercise: Add a custom step in the pipeline, Video

3.7. Combining model output variables

• Presentation - Slides
• Exercise: Using aux files

3.8. PyFESOM

• Presentation - Slides
• Exercise: Using pyfesom2 with pymor

4. Try Pymor with your data

5. Close the workshop

---

Install Pymor

As the data to run the exercises lives on Levante,(DKRZ), we need to setup the Pymor package on Levante.

The installation is done in 2 steps:

1. Create a virtual environment
2. Install Pymor

Create a virtual environment

Note: the exercises assume you have a pymor conda environment, so make sure when you create the environment that its name matches that.

module load python3
eval "$(conda shell.bash hook)"      # Init conda without the need to add extra lines to ~/.bashrc
conda create -n pymor python=3.10    # Create a conda environment named pymor

Install Pymor

conda activate pymor
pip install --isolated py-cmor[dev,fesom]

You can then check the Pymor works by running:

pymor --version
pymor --help

For additional information, check out the following links

• Creating Virtual environments on Levante: Python environment locations.
• Installing Pymor: Pymor documentation.

Pymor workshop repository

The Pymor workshop repository contains scripts to run the exercises.

work="/work/$(id -gn)/$USER"
echo "work directory: $work"
cd $work
git clone https://github.com/esm-tools/pymor_workshop.git
cd pymor_workshop

Get the example data

This workshop is designed for ease-of-use if you are working on DKRZ's Levante system. You can access all of the test data here:

/work/ab0995/a270243/pymor_workshop/exercises/data

If you want to be able to see the files as part of your own repository, you can make a link

pwd  # You should be in the root of the repository (`/work/$PROJECT/$USER/pymor_workshop`)
ln -sv /work/ab0995/a270243/pymor_workshop/exercises/data ./exercises/data

Alternatively, if you would like to use synthetic data made up of random numbers, but with realistic metadata, dimensions, and attributes, you can generate some. Make sure you are in the same Python environment as pymor is installed for this:

make fake-data
ln -sv ./fake-data ./exercises/data