*Cycle detection* determines properties of cycles in a directed or undirected graph, such as whether each node of the graph is part of a cycle or just checking whether a cycle exists.
A related topic is **strongly connected components**, a platinum level concept.
### Functional Graphs
Links:
- CPH 16.3: successor paths
- CPH 16.4: cycle detection in successor graph
In silver-level directed cycle problems, it is generally the case that each node has exactly one edge going out of it. This is known as a **successor graph** or a **functional graph.**
The following sample code counts the number of cycles in such a graph. The "stack" contains nodes that can reach the current node. If the current node points to a node `v` on the stack (`on_stack[v]` is true), then we know that a cycle has been created. However, if the current node points to a node `v` that has been previously visited but is not on the stack, then we know that the current chain of nodes points into a cycle that has already been considered.
```cpp
//UNTESTED
//Each node points to next_node[node]
bool visited[MAXN], on_stack[MAXN];
int number_of_cycles = 0, next_node[MAXN];
void dfs(int n)
{
visited[n] = on_stack[n] = true;
int u = next_node[n];
if(on_stack[u])
number_of_cycles++;
else if(!visited[u])
dfs(u);
on_stack[n] = false;
}
int main()
{
//read input, etc
for(int i = 1;i <= N;i++)
if(!visited[i])
dfs(i);
}
```
The same general idea is implemented below to find any cycle in a directed graph (if one exists).