getting_started.md

Getting Started

Install fireatlas as a Package

We use pyproject.toml with pip to "deploy" fireatlas locally inside your environment:

$ git clone <this-repo> fireatlas

$ cd fireatlas/

# note that '-e' allows you to edit your code and all changes will be available without reinstall
$ pip install -e .

# or if you need testing infrastructure too 
$ pip install -e '.[test]'

Running the Algorithm

There are two different environments where you can run all algorithm steps: JupyterHub and MAAP-DPS

JupyterHub:

1. If you have accesss to JupyterHub then you can interactively walk through the notebooks in the /notebooks directory (see Notebook Docs)

2. Or if you'd rather just run one script that coordinates all the work for you in the right order with dask parallelism then you can run fireatlas/FireRunDaskCoordinator.py

DPS:

DPS is an async task runner. For more info read the docs

1. You can run individual steps asynchronously on DPS via the notebook at /maap_runtime/dps-interactive-coordinator.ipynb

2. Or you can kick off all steps in a single DPS job via the last step in the notebook at /maap_runtime/dps-coordinator.ipynb