List of known bugs and issues in R package hergm

R package hergm is available at https://CRAN.R-project.org/package=hergm and has been published in the Journal of Statistical Software.

hergm version 4.1-3

  1. By mistake, the estimate and standard error of the between-block parameter under method "ml" were reported to be equal, because hergm printed the estimate of the between-block parameter twice - thanks to Angelo Mele for alerting us.

  2. There is a bug in the simulation of directed networks. It does not affect the simulation of undirected networks. Thanks to Edward McFowland III for alerting us.

hergm version 4.1-2

We changed the default parameterization of method = "ml" from the size-independent parameterization "standard" to the size-dependent parameterization "offset", which makes more sense when the sizes of blocks are not equal. In addition, we updated the help function along with the summary function to improve clarity. Thanks to Francesca Vantaggiato for alerting us to these issues.

hergm version 4.1-1

Some additional issues and bugs have been discovered due to the switch from default method = "bayes" to default method = "ml" - thanks to Dave Evenden for alerting me.
  1. Bug: If method = "ml" (default) and the number of blocks is 1, the gof method of hergm 4.1-1 generates an error.
  2. Issues: All of the examples in the help function were designed and tested with method = "bayes" in mind. The switch from default method = "bayes" to default method = "ml" necessitates changes in the examples and help function.

hergm version 4.1-0

Due to the split of hergm and mlergm in late 2018, hergm 4.1-0 had some bugs and issues - thanks to Dave Evenden and Jonathan Stewart for alerting me and helping them fix:
  1. Bug: If method = "ml" (default), no model term with extension "_ij" or "_ijk" was recognized.
  2. Bug: If method = "ml" (default) and all nodes belong to the same block, the between-block parameters cannot be estimated and, as a consequence, the goodness-of-fit function gof() returned an error.
  3. Issue: If method = "ml" (default), the sample size was much larger than needed.
  4. Issue: hergm exported all functions, which is undesirable.

hergm version 3.1-1

Bug: hergm 3.1-0 generated identical seeds in parallel computing when no seeds were supplied.