This is a lower level MCMC-MLE fitting function for ERNM. Users should generally use the ernm() function instead.
ernmFit(
sampler,
theta0,
mcmcBurnIn = 10000,
mcmcInterval = 100,
mcmcSampleSize = 10000,
minIter = 3,
maxIter = 40,
objectiveTolerance = 0.5,
gradTolerance = 0.25,
meanStats,
verbose = 1,
method = c("bounded", "newton")
)an ernm object
the ErnmModel
initial starting values
MCMC burn in
MCMC interval
MCMC sample size
minimum number of MCMC-MLE iterations
maximum number of MCMC-MLE iterations
convergence criteria on change in log likelihood ratio
convergence criteria on scaled gradient
optional target statistics for the mean value parameters
level of verbosity 0, 1, or 2
the optimization method to use. "bounded" uses trust regions around the MCMC sample and is generally preferable. See Fellows (2012) for details.
Fellows, Ian Edward. Exponential family random network models. University of California, Los Angeles, 2012.