population

population

Defines functions for making contact edge lists, e.g. for custom layers.

These are not used by cv.Sim, which creates its networks with Starsim (see network.py). They return a dict of p1 and p2 arrays that can be used to make a layer, e.g. cv.Layer(**cv.make_random_contacts(pop_size, 10), label='random'). They use Covasim’s global random number generator, so call cv.set_seed() first for reproducible edges.

Functions

Name	Description
make_hybrid_contacts	Create “hybrid” contacts – microstructured contacts for households and
make_microstructured_contacts	Create microstructured contacts – i.e. for households.
make_random_contacts	Make random static contacts for a single layer as an edgelist.

make_hybrid_contacts

population.make_hybrid_contacts(
    pop_size,
    ages,
    contacts,
    school_ages=None,
    work_ages=None,
)

Create “hybrid” contacts – microstructured contacts for households and random contacts for schools and workplaces, both of which have extremely basic age structure. A combination of both make_random_contacts() and make_microstructured_contacts().

make_microstructured_contacts

population.make_microstructured_contacts(pop_size, cluster_size, mapping=None)

Create microstructured contacts – i.e. for households.

Parameters

Name	Type	Description	Default
pop_size	int	total number of people	required
cluster_size	int	the average size of each cluster (Poisson-sampled)	required
mapping	array	optionally map the generated indices onto new indices	None

New in version 3.1.1: optimized updated arguments.

make_random_contacts

population.make_random_contacts(
    pop_size,
    n,
    overshoot=1.2,
    dispersion=None,
    mapping=None,
)

Make random static contacts for a single layer as an edgelist.

Parameters

Name	Type	Description	Default
pop_size (int)		number of agents to create contacts between (N)	required
n (int)		the average number of contacts per person for this layer	required
overshoot (float)		to avoid needing to take multiple Poisson draws	required
dispersion (float)		if not None, use a negative binomial distribution with this dispersion parameter instead of Poisson to make the contacts	required
mapping (array)		optionally map the generated indices onto new indices	required

Returns

Name	Type	Description
		Dictionary of two arrays defining UIDs of the edgelist (sources and targets)

New in 3.1.1: optimized and updated arguments.