b1degree and b2degree count the other partition nodes as isolates?

[mbojan]

Is this correct and intended? I'd expect b1degree(0) to be 1 and b2degree(0) to be 2...

suppressPackageStartupMessages({
  library(ergm)
})
network.initialize(5, directed=FALSE, bipartite=2) -> net
add.edge(net, 1, 3)
net
#>  Network attributes:
#>   vertices = 5 
#>   directed = FALSE 
#>   hyper = FALSE 
#>   loops = FALSE 
#>   multiple = FALSE 
#>   bipartite = 2 
#>   total edges= 1 
#>     missing edges= 0 
#>     non-missing edges= 1 
#> 
#>  Vertex attribute names: 
#>     vertex.names 
#> 
#> No edge attributes
as.matrix(net)
#>   3 4 5
#> 1 1 0 0
#> 2 0 0 0
summary(net ~ b1degree(0:1) + b2degree(0:1))
#> b1degree0 b1degree1 b2degree0 b2degree1 
#>         4         1         4         1

I spotted that because of the degree=0 counts in:

> degreedist(davis)
Bipartite mode 2 degree distribution:
 0  3  4  5  6  8 10 12 14 
18  4  2  1  2  2  1  1  1 
Bipartite mode 1 degree distribution:
 0  2  3  4  5  6  7  8 
14  3  1  6  1  1  3  3 

[krivit]

I think that's a bug. The good news is that the statistics are off by a constant, so model coefficients are unaffected.

[mbojan]

oops