The Pregel implementation of PageRank

We have already seen that GraphX has a PageRank API. In the following, let us see how this famous web search algorithmic can be easily implemented using Pregel. Since we already explained in the previous chapter how PageRank works, we will now simply explain its Pregel implementation:

First of all, we need to initialize the ranking graph with each edge attribute set to 1, divided by the out-degree, and each vertex attribute to set 1.0:

val rankGraph: Graph[(Double, Double), Double] = 
    // Associate the degree with each vertex
    graph.outerJoinVertices(graph.outDegrees) {
        (vid, vdata, deg) => deg.getOrElse(0)
    }.mapTriplets( e => 1.0 / e.srcAttr )
     .mapVertices( (id, attr) => (0.0, 0.0) )

Following the Pregel abstraction, we define the three functions that are needed to implement PageRank in GraphX. First, we define the vertex program as follows:

val resetProb = 0.15
def vProg(id: VertexId, attr: (Double, Double), msgSum: Double...