k shortest paths implementation in Igraph/networkx (Yen's algorithm)

Question

After thorough research and based on this , this and a lot more I was suggested to implement k shortest paths algorithm in order to find first, second, third ... k-th shortest path in a large undirected, cyclic, weighted graph. About 2000 nodes.

The pseudocode on Wikipedia is this:

function YenKSP(Graph, source, sink, K):
  //Determine the shortest path from the source to the sink.
 A[0] = Dijkstra(Graph, source, sink);
 // Initialize the heap to store the potential kth shortest path.
 B = [];

for k from 1 to K:
   // The spur node ranges from the first node to the next to last node in the shortest path.
   for i from 0 to size(A[i]) − 1:

       // Spur node is retrieved from the previous k-shortest path, k − 1.
       spurNode = A[k-1].node(i);
       // The sequence of nodes from the source to the spur node of the previous k-shortest path.
       rootPath = A[k-1].nodes(0, i);

       for each path p in A:
           if rootPath == p.nodes(0, i):
               // Remove the links that are part of the previous shortest paths which share the same root path.
               remove p.edge(i, i) from Graph;

       // Calculate the spur path from the spur node to the sink.
       spurPath = Dijkstra(Graph, spurNode, sink);

       // Entire path is made up of the root path and spur path.
       totalPath = rootPath + spurPath;
       // Add the potential k-shortest path to the heap.
       B.append(totalPath);

       // Add back the edges that were removed from the graph.
       restore edges to Graph;

   // Sort the potential k-shortest paths by cost.
   B.sort();
   // Add the lowest cost path becomes the k-shortest path.
   A[k] = B[0];
return A;

The main problem is that I couldn't write the correct python script yet for this (delete edges and places them back in place correctly) so I've only got this far with reliyng on Igraph as usual:

def yenksp(graph,source,sink, k):
    global distance
    """Determine the shortest path from the source to the sink."""
    a = graph.get_shortest_paths(source, sink, weights=distance, mode=ALL, output="vpath")[0]
    b = [] #Initialize the heap to store the potential kth shortest path
    #for xk in range(1,k):
    for xk in range(1,k+1):
        #for i in range(0,len(a)-1):
        for i in range(0,len(a)):
            if i != len(a[:-1])-1:
                spurnode = a[i]
                rootpath = a[0:i]
                #I should remove edges part of the previous shortest paths, but...:
                for p in a:
                    if rootpath == p:
                        graph.delete_edges(i) 

            spurpath = graph.get_shortest_paths(spurnode, sink, weights=distance, mode=ALL, output="vpath")[0]
            totalpath = rootpath + spurpath
            b.append(totalpath)
            # should restore the edges
            # graph.add_edges([(0,i)]) <- this is definitely not correct.
            graph.add_edges(i)
        b.sort()
        a[k] = b[0]
    return a

It's a really poor try and it returns only a list in a list

I'm not very sure anymore what am I doing and I'm very desperate with this issue already and in the last days my point of view on this was changed with 180 degrees and even once. I'm just a noob doing its best. Please help. Networkx implementation can also be suggested.

P.S. It's likely that there are no other working ways about this because we researched it here already . I've already received lots of suggestions and I owe the community alot. DFS or BFS wont work. Graph is huge.

Edit: I keep correcting the python script. In a nutshell the aim of this question is the correct script.

Answer 1

I had the same problem as you so I ported Wikipedia's pseudocode for Yen's algorithm for use in Python with the igraph library.

You can find it there : https://gist.github.com/ALenfant/5491853

Answer 2

There is a python implementation of Yen's KSP on Github, YenKSP. Giving full credit to the author, the heart of the algorithm is given here:

def ksp_yen(graph, node_start, node_end, max_k=2):
    distances, previous = dijkstra(graph, node_start)

    A = [{'cost': distances[node_end], 
          'path': path(previous, node_start, node_end)}]
    B = []

    if not A[0]['path']: return A

    for k in range(1, max_k):
        for i in range(0, len(A[-1]['path']) - 1):
            node_spur = A[-1]['path'][i]
            path_root = A[-1]['path'][:i+1]

            edges_removed = []
            for path_k in A:
                curr_path = path_k['path']
                if len(curr_path) > i and path_root == curr_path[:i+1]:
                    cost = graph.remove_edge(curr_path[i], curr_path[i+1])
                    if cost == -1:
                        continue
                    edges_removed.append([curr_path[i], curr_path[i+1], cost])

            path_spur = dijkstra(graph, node_spur, node_end)

            if path_spur['path']:
                path_total = path_root[:-1] + path_spur['path']
                dist_total = distances[node_spur] + path_spur['cost']
                potential_k = {'cost': dist_total, 'path': path_total}

                if not (potential_k in B):
                    B.append(potential_k)

            for edge in edges_removed:
                graph.add_edge(edge[0], edge[1], edge[2])

        if len(B):
            B = sorted(B, key=itemgetter('cost'))
            A.append(B[0])
            B.pop(0)
        else:
            break

    return A