Koudou

Modular small community simulator

This repository contains the simulator used for the paper: Shiyu Jiang, Hee Joong Kim, Fabio Tanaka, Claus Aranha, Anna Bogdanova, Kimia Ghobadi and Anton Dahbura. 2023. [Simulating Disease Spread During Disaster Scenarios]() In Proceedings of the International Conference on Artificial Life (ALife 2023).

How to reproduce the results:

Set up environment

git clone -b ALIFE_2023 https://github.com/caranha/Koudou.git
cd Koudou
pip install -r requirements.txt

Deploy Map .osm file

Download the .osm file for the tsukuba area and place it on: osm_files/Tx-To-TU.osm

• To download the file go to: https://www.openstreetmap.org/export and set the following coordinates:
  • minlat="36.0777000"
  • minlon="140.0921000"
  • maxlat="36.1217000"
  • maxlon="140.1201000"
• Or download using the API req: https://www.openstreetmap.org/api/0.6/map?bbox=140.0921%2C36.0777%2C140.1201%2C36.1217

  Modify the config files

  Referring to the directory of config and parameters, modify the files to the desired configuration as introduced in the paper. Here is some default parameters as tables for reference:

  Reduced infection rate for mask config/infection/covid.json --> "precautionary_measures"

  The infection will multiply by the reduced infection rate based on agent's mask condition.

Chances of changing mask config/infection/covid.json --> "precautionary_measures"

When self-infected, the chance of changing mask to a better one.

Agent profession with mask /config/behavioral/profession.csv

Proportion of agents with mask for each profession at initialization. The sum of the proportion of each profession should be 1.

Run the simulation

-s: seed for random number generator. It can help with the reproducibility of the results. If not specified, the default seed is 1111.

python main.py -p parameters/default.py -s 1111

Analyze the results

Use dashboard to analyze the results with visualization and statistics. It locates at /src/dashapp/App.py Please refer to the README.md in the dashapp folder for more details.