baoyinyuan / covid19dispersion

Official implementation of "Why ODE models for COVID-19 fail: Heterogeneity shapes epidemic dynamics"

Stochastic COVID-19 Simulation

Stochastic simulation of a simple COVID-19 model on Networks with individual infectiousness variations.

Copyright: 2021, Gerrit Großmann, Group of Modeling and Simulation at Saarland University

Version: 0.1 (Please note that this is proof-of-concept code in a very early development stage.)

Caveat lector: This is an academic model, do not use academic models as a basis for political decision-making.

Overview

---

Installation

---

The tool is based on Python3. Install the required dependencies using: With:

pip install -r requirements.txt

Example Usage

---

With

python simulation.py

Output

---

Two output folders are created: output_graphs/ and output_dynamics/.

output_graph

---

output_graphs/ contains example contact networks which are generated. Note that for each simulation run a new contact network is generated using a random graph model. Moreover, the folder contains summary statistics (degree distribution) and network visalizations (only for networks < 200 nodes).

output_dynamics

---

output_dynamics/ contains different files describing the stochastic dynamics of the system. Files with evolution in their name report the fraction of nodes in each compartment over time. The rvalues file reports for each infected node in each simulatino run: (1) when the node became infected, (2) number of secondary infections of that node. Visualization code is not provided.