For support contact Tonn Rüter

Python is a high-level, interpreted programming language known for its readability and versatility. In REMIND, Python is essential for coupling to specialized models, such as those used for climate assessment reporting. These models often require advanced data processing and integration with other tools, which Python in conjecture with R reticulate facilitates efficiently.

Python support is enabled by default in REMIND and is necessary to couple to specialized models, e.g. for climate assessment reporting. This can be verified in the default.cfg file where cfg$pythonEnabled is set to "on". Users need to set the cfg$pythonPath variable to point to an appropriate Python environment. On the PIK cluster, the conda environment can be found in /p/projects/rd3mod/python/environments/scm_magicc7_hpc/. It is best practice to clone this default environment into one's home directory when using the PIK cluster, see make commands below. Further management of the Python environment, such as updating Python packages environments, is a responsibility of the user.

REMIND scripts do not alter the Python environment, ensuring that the environment remains stable and predictable throughout the usage of REMIND. For repeatability a lockfile is generated for each REMIND run that archives the state of the Python environment. This ensures that the exact environment used for a specific run can be recreated if needed to ensure repeatability. In case a virtual Python environment like conda or venv is used, the user needs to make sure to activate it before starting a REMIND run. After starting a REMIND run, checks ensure that all necessary dependencies are available.

REMIND relies on the conda run subcommand to execute Python scripts. A recent conda installation is therefore required to make use of all REMIND capabilities. For more details on installing conda, refer to the Installing conda section.

The Makefile provided by REMIND includes targets to create or clone a Python environment:

• make clone-conda: Clones an existing conda environment. Note: Users of the PIK cluster should activate the default environment at (see below for more information) then use make conda-clone to create their own version
• make create-conda: Creates a new conda environment based on the py_requirements.txt file.

To interface with other models or libraries, it is often necessary to call Python code because the model itself is written in Python or bindings are available in Python. REMIND has Python support via the R library reticulate. One can, for example, execute an R script from GAMS, then use reticulate::import to import Python libraries. For more information on using reticulate, refer to the Documentation.

When using REMIND, you might encounter warnings about the inability to verify the version of certain Python packages installed from specific repositories. These warnings occur because the installed Python packages do not retain information about their origin, only their version number. This version number is meaningless if the package was installed from a dedicated repository. Therefore, when using the PIK cluster default REMIND Python environment, you can safely disregard these warnings. They are simply a result of the package management system's inability to verify the origin of the installed packages.

conda is an open-source package and environment management system for Windows, macOS, and Linux. It simplifies installing and managing Python dependencies and environments, making it ideal for running REMIND. For users running REMIND on their desktop, any Python installation will do; however, we recommend using conda for its ease of managing dependencies and environments.

All necessary software is available on the cluster. Calling piamenv::condaInit(how = "pik-cluster") in e.g. a coupling script ensures that modules are loaded at the appropriate time, piamenv::condaRun will run a Python command in a specified conda environment.

You can load the PIK HPC conda module manually with

module load anaconda/2025.10

and activate a conda environment with

source activate path/to/env

before starting a run, but please note that as of REMIND version 3.5.0 you don't have to.

Note This is not applicable to REMIND users on the PIK HPC

1. Download the conda installer:

   • Recommended: For Miniconda (a minimal installer for conda): Miniconda Distribution
   • Optional: For Anaconda (includes conda and many scientific packages): Anaconda Distribution

2. Run the installer:

   • On Windows, double-click the .exe file
   • On macOS and Linux, open a terminal and run:

     bash path/to/installer.sh

3. Follow the prompts to complete the installation.

4. Verify the installation by opening a terminal or command prompt and running:

   conda --version

To create a new conda environment from the config/py_requirements.txt file that comes with REMIND, you can use the make create-conda target in the Makefile. This will create a new conda environment with Python 3.11 and the packages specified in the requirements file.

• Create a new environment with the default name remind in the default directory ($HOME/.conda/envs):

  make create-conda

• Create a new environment with a specified name:

  make create-conda ENV=my_new_env

• Create a new environment in a specified directory:

  make create-conda DEST=/path/to/directory

• Create a new environment with a specified name in a specified directory:

  make create-conda ENV=my_new_env DEST=/path/to/directory

To clone a conda environment, you can use the make clone-conda target in the Makefile. This will clone the specified conda environment or the active environment to a new environment in the user's home directory or a specified destination.

• Clone the active environment:

  make clone-conda

• Clone a specified environment:

  make clone-conda ENV=my_env

• Clone a specified environment to a custom directory:

  make clone-conda ENV=my_env DEST=~/my_custom_directory

If you need to rename the cloned environment, follow these steps:

1. Activate the cloned environment:

   conda activate /path/to/cloned-env

2. Export the environment state to a YAML file:

   conda env export > cloned-env.yml

3. Create a new environment with the desired name from the YAML file:

   conda env create -f cloned-env.yml -n new-env-name

4. Remove the old cloned environment:

   conda remove --prefix /path/to/cloned-env --all

These steps ensure that the environment is properly renamed without breaking conda's environment management