如果DFS发现一条边指向一个已经访问过的顶点，那么这里就有一个循环。

Tarjan的强连通分量算法的时间复杂度为O(|E| + |V|)。

有关其他算法，请参见维基百科上的强连接组件。

根据Cormen et al.， Introduction to Algorithms (CLRS)引理22.11:

有向图G是无环的当且仅当深度优先搜索G没有得到后边。

在几个回答中已经提到了这一点;在这里，我还将提供一个基于CLRS第22章的代码示例。示例图如下所示。

CLRS深度优先搜索的伪代码如下:

在CLRS图22.4中的示例中，图由两棵DFS树组成:一棵由节点u、v、x和y组成，另一棵由节点w和z组成。每棵树都包含一条后边:一条从x到v，另一条从z到z(一个自循环)。

关键的实现是，在DFS-VISIT函数中，当在u的邻居v上迭代时，遇到一个带有灰色的节点时，就会遇到后边缘。

下面的Python代码是CLRS伪代码的改编，添加了一个if子句，用于检测周期:

import collections


class Graph(object):
    def __init__(self, edges):
        self.edges = edges
        self.adj = Graph._build_adjacency_list(edges)

    @staticmethod
    def _build_adjacency_list(edges):
        adj = collections.defaultdict(list)
        for edge in edges:
            adj[edge[0]].append(edge[1])
            adj[edge[1]] # side effect only
        return adj


def dfs(G):
    discovered = set()
    finished = set()

    for u in G.adj:
        if u not in discovered and u not in finished:
            discovered, finished = dfs_visit(G, u, discovered, finished)


def dfs_visit(G, u, discovered, finished):
    discovered.add(u)

    for v in G.adj[u]:
        # Detect cycles
        if v in discovered:
            print(f"Cycle detected: found a back edge from {u} to {v}.")
            break

        # Recurse into DFS tree
        if v not in finished:
            dfs_visit(G, v, discovered, finished)

    discovered.remove(u)
    finished.add(u)

    return discovered, finished


if __name__ == "__main__":
    G = Graph([
        ('u', 'v'),
        ('u', 'x'),
        ('v', 'y'),
        ('w', 'y'),
        ('w', 'z'),
        ('x', 'v'),
        ('y', 'x'),
        ('z', 'z')])

    dfs(G)

注意，在本例中，CLRS伪代码中的时间没有被捕获，因为我们只对检测周期感兴趣。还有一些样板代码，用于从边列表构建图的邻接表表示。

当这个脚本执行时，它输出如下:

Cycle detected: found a back edge from x to v.
Cycle detected: found a back edge from z to z.

这些正是CLRS图22.4示例中的后边缘。

我已经在sml(命令式编程)中实现了这个问题。这是大纲。找到所有入度或出度为0的节点。这样的节点不能成为循环的一部分(因此将它们删除)。接下来，从这些节点中删除所有传入或传出边。 递归地将此过程应用于结果图。如果最后你没有剩下任何节点或边，图就没有任何循环，否则就有。

There is no algorithm which can find all the cycles in a directed graph in polynomial time. Suppose, the directed graph has n nodes and every pair of the nodes has connections to each other which means you have a complete graph. So any non-empty subset of these n nodes indicates a cycle and there are 2^n-1 number of such subsets. So no polynomial time algorithm exists. So suppose you have an efficient (non-stupid) algorithm which can tell you the number of directed cycles in a graph, you can first find the strong connected components, then applying your algorithm on these connected components. Since cycles only exist within the components and not between them.

如果DFS发现一条边指向一个已经访问过的顶点，那么这里就有一个循环。