pythonabm 0.3.3

Framework for building Agent-Based Models (ABMs) in Python

PythonABM

PythonABM makes complex agent-based modeling (ABM) simulations in Python accessible by providing an efficient base framework for constructing ABMs. More information on PythonABM can be found below.

• Documentation: pythonabm.readthedocs.io
• PyPI: pypi.org/project/pythonabm

Installation

Assuming you have Python 3.7-3.10, the latest version of the PythonABM library can be installed using the following command.

pip install pythonabm

PythonABM can also be built from source once downloaded from GitHub.

python setup.py install

Running a simulation

Calling the start() method in the Simulation class will launch the ABM platform and run it as follows. (See the example.py script as a template for building a simulation.) A text-based UI will prompt for a name of the simulation and a corresponding mode (described below).

• 0: New simulation
• 1: Continue a previous simulation
• 2: Convert a previous simulation’s images to a video
• 3: Archive (ZIP) a previous simulation’s outputs

To avoid the text-based UI, the name and mode can be passed at the command line by using flags (without the parentheses). Note: the simulation file does not have to be named “main.py”.

python main.py -n (name) -m (mode)

When continuing a previous simulation (mode=1), the ABM will prompt for the updated end-step number, though this can be passed like above.

python main.py -n (name) -m (mode) -es (end-step)

NVIDIA CUDA support

For GPU parallelization, PythonABM requires a CUDA compatible GPU and NVIDIA’s CUDA toolkit. If you don’t have the toolkit, download the toolkit either directly from NVIDIA here or with Anaconda's conda command show below.

conda install cudatoolkit