Skip to content

Latest commit

 

History

History
47 lines (37 loc) · 3.41 KB

setup_environment.md

File metadata and controls

47 lines (37 loc) · 3.41 KB
Raw

Set up the conda environment for STEMMUS-SCOPE

In this guide, you will learn how to configure the conda environment in your jupyter-workspace on SURF research cloud for running the STEMMUS-SCOPE model in the notebook.

This guide assumes you already have access to SRC as a user and the workspace created by the administrator of EcoExtreML project group.

In order to run STEMMUS-SCOPE on SRC, we need to configure the conda environment and download the executable file.

Managing environments and kernels on this Jupyter-workspace using conda

In our jupyter-workspace, miniconda is pre-installed at the location /etc/miniconda. We will use it to create virtual python environments and install pystemmusscope.

To set up the conda environment:

  • Start a "Terminal" tab in the Jupyter Lab launcher and type:

    /etc/miniconda/bin/conda init

  • Close the terminal tab and start a new one. You will see that the terminal prompt has changed to something like:

    (base) metheuser@mywsp:

    (When you log in to your jupyter-workspace for the first time, a notebook called KERNEL-README.ipynb will be created by default and it is accessbile from files panel. In this notebook, you can find the guidelines about how to manage environments and kernels on this Jupyter-workspace.)

  • Since this miniconda tool is shared by all the users on this jupyter-workspace, you should not use/modify this base environment. Instead, you can create and activate another pystemmusscope conda environment using the environment.yml file. Let's first clone the STEMMUS_SCOPE_Processing repo to our workspace using HTTPS (not via SSH, otherwise you need to configure your ssh key):

    git clone https://github.com/EcoExtreML/STEMMUS_SCOPE_Processing.git

  • Now we can create the pystemmusscope conda environment using the environment.yml file:

    cd STEMMUS_SCOPE_Processing
conda env create -f environment.yml

  • When the installation process is finished, we can acvtivate the pystemmusscope conda environment:

    conda activate pystemmusscope

  • Then we can create a kernel from it (make sure that the modified conda-environment is still the active one):

    python3 -m ipykernel install --user --name pystemmusscope --display-name "pystemmusscope"

Run STEMMUS-SCOPE model using the executable file

Now we can run the STEMMUS-SCOPE model using the executable file. Open the notebook run_model_in_notebook.ipynb from the path /STEMMUS_SCOPE_Processing/docs/notebooks in your jupyter-workspace and use the pystemmusscope kernel.

Before executing the notebook, remember to modify the config file and create your work directory. If you want to put the work directory on the shared storage (/data/volume_2/), remember to create your own folder to avoid conflicts. Note that the executable file and all the related forcing files are stored at /data/volume_2/.

Note: if this message "error while loading shared libraries" occurs when running the notebook in your jupyter-workspace after configuring the environment properly, please execute the cell for setting the LD_LIBRARY_PATH.