Population Model

A population of people is comprised of “meta-groups.” Within a meta-group, all parameters such as hospitalization rate remain constant with the exception of contact level. For example, in modeling a university, possible meta-groups could be undergraduates, graduates, professional students, and faculty/staff. A meta-group contact matrix encodes how often these meta-groups come into contact with one another. Within each meta-group, there are multiple groups that represent varying contact levels.

The simpar.groups.MetaGroup class allows one to specify a vector of number of contacts across meta-groups as well as a vector indicating what fraction of the meta-group population is in each of these groups. Alternatively, this can be specified with a truncated Pareto distribution. Given a shape parameter $\alpha$ and truncation point $ub$ , the fraction of the meta-group population with $i$ contacts is $f(i;\alpha,ub)$ where $f$ is the probability density function of the Pareto distribution. Note these values are normalized to sum to 1.

The interactions among individuals within the same meta-group is assumed to be well-mixed in that the amount of interaction group $i$ has with group $j$ is proportional to both groups contact levels and the fraction of the population of group $j$ . Hence, it is not assumed that more social people tend to interact with more social people and vice versa. Similarly, while the meta-group contact matrix encodes how much contact takes place between two meta-groups, the interaction between them is assumed to be well-mixed with respect to the groups that comprise them. See simpar.groups.MetaGroup and simpar.groups.Population for more details.