In MCMC SE calculation, handle the case of the Hessian not SNND.

statnet · Jan 25, 2024 · b2f9a06 · b2f9a06
1 parent ff704e4
commit b2f9a06
Showing 1 changed file with 1 addition and 1 deletion.
diff --git a/R/ergm.MCMCse.R b/R/ergm.MCMCse.R
@@ -119,7 +119,7 @@ ergm.MCMCse <- function(model, theta, init, statsmatrices, statsmatrices.obs,
 
   mc.cov <- matrix(NA,ncol=length(novar),nrow=length(novar))
 
-  if(sum(!novar)==0 || srcond(-H) < .Machine$double.eps){
+  if(sum(!novar)==0 || ERRVL(try(srcond(-H) < .Machine$double.eps, TRUE), TRUE)){
     warning("Approximate Hessian matrix is singular. Standard errors due to MCMC approximation of the likelihood cannot be evaluated. This is likely due to insufficient MCMC sample size or highly correlated model terms.", call.=FALSE)
   }else{
     mc.cov0 <- sandwich_ssolve(-H, cov.zbar)