Running a Simulation#

While cosasi’s focus is on the inverse problem of inferring the origin of a diffusion process, we provide a simulator of the forward problem for convenience. This functionality is found in cosasi.contagion, which is built on NetworkX and NDlib.

StaticNetworkContagion objects need:

A contact network to spread on (NetworkX graph)
An epidemic model to run (one of SI, SIS, or SIR)
An infection_rate specifying the infectivity of the contagion

We first define a contact network for the diffusion process to spread on; these are implemented as NetworkX graphs.

import networkx as nx
import cosasi
import random
import numpy as np

seed = 42
random.seed(seed)
np.random.seed(seed)

G = nx.fast_gnp_random_graph(200, 0.01, seed=seed)

Then define a contagion object:

contagion = cosasi.StaticNetworkContagion(
    G=G,
    model="si",
    infection_rate=0.2,
    seed=seed
)

By default, one vertex, selected uniformly at random, is “infected” at initialization. Alternately, the user can specify a number of “patients zero”:

contagion = cosasi.StaticNetworkContagion(
    G=G,
    model="si",
    infection_rate=0.2,
    number_infected=4,
    seed=seed
)

or a fraction to be “infected” at initialization:

contagion = cosasi.StaticNetworkContagion(
    G=G,
    model="si",
    infection_rate=0.2,
    fraction_infected=0.05,
    seed=seed
)

If the epidemic model is SIR or SIS, the user must provide a recovery rate at which vertices in the infected compartment switch to the recovered or susceptible compartments:

contagion = cosasi.StaticNetworkContagion(
    G=G,
    model="si",
    infection_rate=0.2,
    recovery_rate=0.005,
    seed=seed
)

After initialization, run the contagion process for as many steps as desired:

contagion.forward(steps=100)

To reset compartmental histories, simply run:

contagion.reset_sim()

Tutorial

Extracting Simulation Information