Quickstart

This page covers installation and a first ADELM run.

Clone the repository

git clone https://git.bgc-jena.mpg.de/ml4hes/adelm.git
cd adelm

Prepare the environment

For GPU-enabled servers, use the provided Conda environment:

conda env create -f gpu_environment.yml
conda activate adelm_gpu

For a lightweight CPU setup, install the Python dependencies directly:

pip install -r requirements.txt

Prepare a configuration file

All ADELM runs are driven by a single YAML configuration file. The project root contains a commented template config.yaml that you can copy and adapt:

cp config.yaml my_experiment.yaml
# edit my_experiment.yaml with your data paths and settings

A minimal config requires:

• the shared model, data, and parameterization blocks

• one workflow-specific block: site_simulation, site_learning, or grid_simulation

Required fields vary by workflow and input data layout. See Configuration for the field-by-field reference and Workflow for runnable examples.

Note

Entry scripts live in scripts/; reusable YAML examples live in examples/. Select a script first, then point it at the example config closest to the experiment.

Run a workflow

For a first site-scale forward run:

python scripts/site_simulation.py --config examples/example_site_simulation_default_parameters.yaml

For a first site-learning run:

python scripts/site_learning.py --config examples/example_single_learning_config.yaml

For a first gridded forward run:

python scripts/grid_simulation.py --config examples/example_grid_simulation_default_parameters.yaml

See Workflow for the available workflows, example configs, and launch patterns.