[graph-tool] slow shortest path

I want to compute shortest path repeatedly, each time some edge weight is changed. The following implementation run very slowly, is this my problem? for e in elist: w=wt[e] wt[g.edge_index[e]]=sys.maxint d2=shortest_distance(g, g.vertex(s), g.vertex(t),weights=wt) wt[e]=w Profiling says most of the time (132 sec in 153 sec)is spend on __init__.py:539(__setitem__). It seems each time all weights are updated using wt, if there is a large number of edges, this will cost a lot time. But in my case, only one edge weight is changed, is there some other way to do this? Thanks.