With igraph package through Rserver: Removing the trivial loops and combining part1 and part2 results in a DAG, so no nontrivial cycles. Or am I missing something? load'stats/r/rserver' Rcmd'library(igraph)' 'A' Rset part1 +.&(*. -.@e.@i.@#) part2 Rcmd'g <- graph_from_adjacency_matrix(A, mode="directed")' Rget'is.dag(g)' 1
On Thu, Apr 7, 2022 at 1:22 AM Raul Miller <rauldmil...@gmail.com> wrote: > > https://gist.githubusercontent.com/rdm/3ce3503ee63b51cd49c97ad185e1dd6f/raw/100e83ba7a87f10b844271180318c37fee1304dc/gph.txt > > The text at that gist represents a "defective graph" > > require'web/gethttp' > text=: gethttp url > 'A B C'=: |:640 #.inv<.0".text rplc LF,' ' > > part1=: A=/B > part2=: B=/C > > To get an idea of the structure, try inspecting part1 and part2 in viewmat. > > part1 and part2 are almost the same, but not quite. In some cases, we > have nodes which are reflexive (point to themselves) in the one, but > not in the other. > > Also, some nodes point "outside" the graph. In part1 and part2, these > connections are ignored, and that's fine for now. > > I am trying to figure out a good way of detecting nontrivial cycles > (cycles which include at least two graph nodes) in either part 1, or > part 2, or the combination of the two (part1 and part2 together > represent connections involving these same nodes). > > Does anyone here have a good approach for that? > > Thanks, > ---------------------------------------------------------------------- > For information about J forums see http://www.jsoftware.com/forums.htm > ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
