Strange Results when Filtering Vertices Iteratively


I have been trying to filter vertices iteratively but am seeing very strange

Summary of what my code is doing:
I loop through groups of vertices. I set the vertex property map to False
for all vertices. I then set it to True for all vertices in the current
group. I then loop through all vertices in the graph setting the vertex
property for the current vertex also to True, creating a graph_view object
using that property map, resetting the value for that vertex to zero and
then moving on to the next vertex repeating the procedure. For some reason
the number of vertices in the resulting graph_view object bears no
resemblance to the number of vertices in my current group. Can anybody help
me with what is going on here?

I am using '2.27dev (commit 31421e3c, Sun Jun 17 19:18:54 2018 +0200)'. If
the code looks like it should work I can upload a copy of the state and
network if that would be useful.

Here is the code:

g = gt.load_graph('')
with open('state.pkl','rb') as state_pkl:

groups = collections.defaultdict(list)
l0 = state.levels[0]
for v in g.vertices():
v_filter = g.new_vertex_property('bool')

P_part = collections.defaultdict(int)
for target_vertices in groups:
    target_vertices = groups[target_vertices]
    for v_index in target_vertices:
        vertex = g.vertex(v_index)
        v_filter[vertex] = True
    for v in g.vertices():
        v_filter[v] = True
        f_g = gt.GraphView(g,v_filter)
        if int(v) in target_vertices:
            if f_g.num_vertices() != len(target_vertices):
                print 'f_g.num_vertices()=', f_g.num_vertices(), 'expected:
', len(target_vertices)
        elif f_g.num_vertices() != len(target_vertices)+1:
            'f_g.num_vertices()=', f_g.num_vertices(), 'expected: ',
        v_filter[v] = False

Best wishes,


I couldn't run your code since it depends on some data you haven't provided
--- it is _always_ a bad idea to give code that can't be run. Please provide
*self-contained* examples. Otherwise we have to guess what the intended
behavior is, instead of looking at it ourselves.

But looking briefly, it seems you are resetting the filter even for vertices
that do belong to the current group, thus destroying it.


You are perfectly right, I am indeed! I will rerun it and get back in touch
if the problem persists. Thank you for spotting this though!