Question about tree swap implementation

[winni2k]

Hi, Thanks for this excellent tool and paper.

I was reading through the SCITE source code to understand the Swap subtrees section of the Supplemental Material section of the SCITE paper. In particular, I am trying to figure out whether d(i) and d(k) are calculated before or after the tree swap move.

From the source code at https://github.com/cbg-ethz/SCITE/blob/master/mcmcTreeMove.cpp#L65 it appears that d(k) is calculated before the swap, and that d(i) is calculated after the swap. Please correct me if I'm wrong.

However, I also noticed a potential bug in the calculation of the neighborhood correction at https://github.com/cbg-ethz/SCITE/blob/master/mcmcTreeMove.cpp#L72

nbhcorrection = 1.0*descendants.size()/nextdescendants.size();

What happens if nextdescendants is empty? Or does nextdescendants always contain the parent node (i.e. i) as well?