Bergm (version 3.0.1)

bergm: Bayesian exponential random graph models

Description

Function to fit Bayesian exponential random graphs models using the exchange algorithm. Two are the sampling approaches available: block update and population MCMC with parallel Adaptive Direction Sampling (ADS).

Usage

bergm(formula,  burn.in=10, main.iters=1000, aux.iters=1000,  m.prior = NULL,  sigma.prior = NULL,  nchains = NULL,  gamma = 0.5,  sigma.epsilon = NULL, ...)

Arguments

formula

formula; an R formula object, of the form ~ where is a network object and are ergm-terms.

burn.in

count; number of burn-in iterations at the beginning of an MCMC run. If population MCMC is performed, it refers to the number of burn-in iterations for every chain of the population.

main.iters

count; number of iterations for the MCMC chain(s) excluding burn-in. If population MCMC is performed, it refers to the number of iterations for every chain of the population.

aux.iters

count; number of auxiliary iterations used for network simulation.

m.prior

vector; mean of the multivariate Normal prior. By default set to a vector of 0's.

sigma.prior

variance/covariance matrix for the multivariate Normal prior. By default set to a diagonal matrix with every diagonal entry equal to 100.

nchains

count; number of chains of the population MCMC. By default set to twice the model dimension (number of model terms). If the model is one-dimensional, nchains is set to 1.

gamma

scalar; ``parallel ADS move factor.'' In case of one-dimensional models, the population MCMC procedure is disabled and gamma is used as variance of the Normal proposal distribution.

sigma.epsilon

variance/covariance matrix for the multivariate Normal proposal or ``parallel ADS move parameter''. By default set to a diagonal matrix with every diagonal entry equal to 0.0025. If the model is one-dimensional, sigma.espilon is set equal to gamma.

...

additional arguments, to be passed to lower-level functions.

See Also

bergm.output, bgof.

Examples

data(florentine)

# Parameter estimation via approximate exchange algorithm with ADS:

flo <- bergm(flomarriage ~ edges + kstar(2),
             burn.in=10,
             aux.iters=500,
             main.iters=500,
             gamma=1)

# MCMC diagnostics:

bergm.output(flo)

# Bayesian goodness-of-fit test:

bgof(flo,
     aux.iters=500,
     sample.size=50,
     n.deg=10,
     n.dist=9,
     n.esp=6)