Closed chad-klumb closed 1 year ago
Looks like a regression from the refactor of ergm()
. Will fix.
Actually no, looks like the problem precedes the refactor.
The problem is that if init="zeros"
, MPLE call is bypassed, and then if MLE ≡ MPLE, MCMLE is never run. Thus, it's not just a likelihood calculation error, and it has nothing to do with target stats: there is no estimate in the first place.
When running
with
ergm@master
I get the errorapparently coming from this line:
https://github.com/statnet/ergm/blob/8eeb514f81b9a72113a5492341bf303169df3f50/R/logLik.ergm.R#L110