yannickoswald / covpol

COVPOL: An agent-based model of the international COVID-19 policy response

COVPOL: An agent-based model of the international COVID-19 policy response

This repository is for the publication

Oswald, Malleson, Suchak (2023). An agent-based model of the 2020 international policy diffusion in response to the COVID-19 pandemic with particle filter

available as preprint at https://arxiv.org/abs/2302.11277

This work has been implemented using the Python Anaconda distribution and the agent-based model package MESA in particular. It is provided with a .yml file specifying a conda environment which contains the required packages. In order to set up the environment, run the following command from the terminal/conda prompt:

conda env create -f env.yml

This will create a new conda environment titled cov-pol. The environment can then be activated using the following command:

conda activate cov-pol

To reproduce the full body of work, take the following steps:

1. Run the script file run_base_model_and_filter_with_plotting.py in the covpol directory as follows:

   cd covpol
   python run_base_model_and_filter_with_plotting.py
   

   This reproduces a substantial amount of the above paper including Figure 2, 4 and 5. To reproduce Figure 4 exactly, the notebook has to take the parameter no_of_iterations = 100. To reproduce Figure 5 exactly, the notebeook has to take the parameter no_of_iterations = 1000.

2. To reproduce Figure 6 in full several intermediate steps are necessary (time-expensive):

   1. Run the script file number_of_particles_experiment_MSE.py to reproduce the data points where iterations = 20.

   2. To reproduce the datapoints where iterations = 1, run the following:

      1. particle_filter_only.py and collect the data.
      2. run_base_model_only_parallelized.py and collect the data.

   3. Alternatively Figure 6 can be reproduced exactly in a time-cheap manner as it is from the script file graph_mse_number_of_particles_experiment.py.

   Best run from anaconda command prompt.

3. To reproduce Figure 7 run the script experiment_da_window_size.py (time-expensive). Best run from anaconda command prompt.