pycopanbehave

Python implementation of COPAN:BEHAVE model

Provides code for reproducing the analysis reported in:

C.-F. Schleussner*, J.F. Donges*, D.A. Engemann*, and A. Levermann, Clustered marginalization of minorities during social transitions induced by co-evolution of behaviour and network structure, Nature Scientific Reports 6, 30790 (2016), DOI: 10.1038/srep30790, The first three authors share the lead authorship (*).

Please see the above paper for a detailed mathematical description of the model (Methods section) and references to the relevant scientific literature.

Software DOI of original release along with publication:

Dependencies

Main:

• numpy > 1.11.1 (http://www.numpy.org/)
• scipy > 0.17.1 (http://www.scipy.org)
• igraph > 0.7.0 (http://igraph.org/python/)
• pyunicorn > 0.5.1 (Unified Complex Network and RecurreNce analysis toolbox), http://www.pik-potsdam.de/~donges/pyunicorn/, get it here: https://github.com/pik-copan/pyunicorn, or via pip: pip install pyunicorn

For plotting:

• matplotlib > 1.5.1 (http://www.matplotlib.org/)
• seaborn > 0.7.1 (https://web.stanford.edu/~mwaskom/software/seaborn/)

Code structure

The modelling approach consists of two parts:

• The dynamic social network model (pycopanbehave/, class CopanBehaveModel)

  • Contains the main functionalities of the underlying adaptive network model
  • provides interfaces to specific applications, e.g. for the smoker evolution case

• The smoker transition simulations (bin/)

  • Specifically adapted to test case
  • Add simulation specific functionalities, e.g. the smoking disposition and the respective transition
  • Optimised to be run in parallel setup

Run the Model

Being rather computationally intensive, the modelling setup is optimised to run in a parallel MPI setup and individual ensemble members will be saved as separate outputs (requires post-processing, see below)

• A run_ens_smoker_*.py contains a dictionary including all relevant configuration parameters
• The setup in manuscript_runs/ resembles the full 1000 Ensemble member setups underlying the manuscript
• The setup under tests/ resembles a computationally less expensive test case of the model setup

Postprocessing and analysis

• The parallelisation procedure saves individual ensemble member representations
• bin/derive_percentiles_from_output.py derives percentiles from the respective output that can be used for further analysis

Plotting results

• manuscript_plot_scripts/plot_manuscript_figures.py allows to reproduce the manuscript figures from model runs