Question

我正在尝试开发一种用于在图表中查找关节点的算法。我正在使用Spark的GraphX库，它有许多有用的算法，如PageRank和连接组件，但据我所知，没有任何东西可以找到切边或清晰点。

我发现an algorithm找到用Java编写的清晰点。我将算法翻译成Scala，如下所示：

// A recursive function that find articulation points using DFS
// u --> The vertex to be visited next
// visited[] --> keeps tract of visited vertices
// disc[] --> Stores discovery times of visited vertices
// parent[] --> Stores parent vertices in DFS tree
// ap[] --> Store articulation points
@tailrec
def APUtil(u: Int, visited: Array[Boolean], disc: Array[Int], low: Array[Int], parent: Array[Int], ap: Array[Boolean]): Unit = {
  var children = 0;
  visited(u) = true;
  time += 1;
  disc(u) = time;
  low(u) = time;
  val vertexId = graph.vertices.zipWithIndex.map {
    case(k,v) => (v,k)
  }.lookup(u).toMap.keys.iterator.next().toInt;
  val direction: EdgeDirection = EdgeDirection.Out;
  val neighbors = graph.collectNeighborIds(direction).lookup(vertexId)(0);
  for(neighborId <- neighbors) {
    val index =  graph.vertices.zipWithIndex.filter { _._1._1 == neighborId }.collect()(0)._2.toInt;
    if(!visited(index)) {
      children += 1;
      parent(index) = u;
      APUtil(index, visited, disc, low, parent, ap);

      low(u) = math.min(low(u), low(index));

      if(parent(u) == -1  && children > 1) {
        ap(u) = true;
      }

      if(parent(u) != -1 && low(index) >= disc(u)) {
        ap(u) = true;
      }

    } else if (index != parent(u)) {
      low(u) = math.min(low(u), disc(index));
    }
  }
}

当我在小数据上运行时，算法正常工作。但是，当我使用更大的文件时，Spark会向我抛出一个StackOverflowError。我读到了关于尾递归的内容，而且我无法弄清楚如何将这个递归函数转换为尾递归函数。

任何帮助表示赞赏！

Scala递归尾递归关节点算法

0 个答案: