Skip to contents

Build the full spatial-correlation matrix across all locations

Source: R/calc_spatial_correlation_matrix.R
calc_spatial_correlation_matrix.Rd

Given simulation outputs, this function constructs the \(J \times J\) matrix \(\mathbf C = [\mathcal{C}_{ij}]\) of Pearson spatial-correlation coefficients (Keeling & Rohani 2002) for every pair of locations.

Usage

calc_spatial_correlation_matrix(I_sym, I_asym, N)

Arguments

I_sym

Numeric matrix (J x T). Symptomatic infections.

I_asym

Numeric matrix (J x T). Asymptomatic infections. Must have the same dimensions as I_sym.

N

Population sizes. Accepts

  • a single scalar (same for every j & t),

  • a length-J vector (different per location, constant in time), or

  • a full J x T matrix (time-varying populations).

Value

A J x J symmetric numeric matrix whose \((i,j)\) entry is \(\mathcal{C}_{ij}\); diagonal elements are set to 1. Pairs that cannot be computed (all NA, zero variance, etc.) are NA_real_.

Details

All three input objects must be arranged as (locations J x time T) (rows are the J locations; columns are the T time steps).

Total infections at time t in location j are taken as \(I_{jt} = I^{(\mathrm{sym})}_{jt} + I^{(\mathrm{asym})}_{jt}\); these are converted to prevalence by dividing by the matching population size \(N_{jt}\). Pair-wise correlations are then computed with calc_spatial_correlation().

See also

calc_spatial_correlation