On 13.10.2014 21:44, tanzairatier wrote:
> I'm trying to read data into a graph using Graph-Tool. My code is further
> below. I'm having the issue of invalid vertex descriptors.
>
> Each row in my.data has a start, end, and transition weight. My goal is to
> iterate through each row, adding the edges (start,end) that have that edge
> weight of trans.
>
> Suppose on the first iteration, my.data has (1,1,2) = (start, trans, end)
> which is to describe an edge between vertices labeled 1 and 2, with the
> weight of '1'.
>
> Then on the second iteration, suppose my.data has (1,5,3) = (start, trans,
> end), which is to describe an edge between vertices labeled 1 and 3, with
> the weight of '5'.
>
> Then I'd want a graph with three vertices, 1, 2, 3 and has two edges (1,2)
> and (1,3).
>
> My strategy is to check if a vertex of each label (start,end) exists, and
> then to add those vertices if they do not exist. The try-catch statements
> seem to work properly. But when we try to access the added vertices, I get
> the invalid vertex descriptor (value error) exception.
>
>
> def gtBuildGraph(my):
> """build a graph using 'Graph-Tool'
> """
>
> my.graph = gt.Graph()
> my.weights = my.graph.new_edge_property("int")
>
> for row in my.data:
> start = int(row['start_state'])
> end = int(row['end_state'])
> trans = int(row['transition'])
>
> #sometimes, start and end nodes are non-unique
> try:
> my.graph.vertex(start)
> except:
> my.graph.add_vertex(start)
> try:
> my.graph.vertex(end)
> except:
> my.graph.add_vertex(end)
>
> start = my.graph.vertex(start)
> end = my.graph.vertex(end)
> newEdge = my.graph.add_edge(start, end)
> my.weights[newEdge] = transVertices are indexed from 0 to N - 1. When you call g.add_vertex(n) it adds n vertices to the graph. If you had zero vertices before, the vertex with index n does not exist yet in the graph (the last one has index n-1), and hence g.vertex(n) is invalid. Best, Tiago -- Tiago de Paula Peixoto <[email protected]>
signature.asc
Description: OpenPGP digital signature
_______________________________________________ graph-tool mailing list [email protected] http://lists.skewed.de/mailman/listinfo/graph-tool
