Hi, all Recently, I faced a big problem when I'm using the boost function of shorthest path computing including dijkstra_shortest_paths and johnson_all_pairs_shortest_paths. My projectet was that I extracted all the surface voxels of a volume as the vertexs of the graph, and 26-connection between them as a edge. Hence, I got a large sparse graph which composed of 12768 vertexs and around 86000 edges. In the next stage, I implemented the dijkstra_shortest_paths computation for each vertex. However, it succeded in some points and failed in other points. If I changed the weight of each edge into 1, all points worked. If I used other weights for different edges, then problem poped up. Also, some points worked for this special weight, and failed for others. I think the shortest path algorithm should be unrelated with the weight assignment. Initially, I think something was wrong in my graph buliding. But when I wrote the shorthest path algorithm by myself and all points worked, I felt there may be a bug in dijkstra_shortest_paths class. If someone else faces the same problem, feel free to tell me. If not, I'm willing to paste my graph code in spite of its complexity. Thanks Regards!