Microsoft Corporation Question:
How could you determine if a linked list contains a cycle in it, and, at what node the cycle starts?
Answer:
There are a number of approaches. The approach I shared is in time N (where N is the number of nodes in your linked list). Assume that the node definition contains a boolean flag, bVisited.
struct Node
{
...
bool bVisited;
};
Then, to determine whether a node has a loop, you could first set this flag to false for all of the nodes:
// Detect cycle
// Note: pHead points to the head of the list (assume already exists)
Node *pCurrent = pHead;
while (pCurrent)
{
pCurrent->bVisited = false;
pCurrent = pCurrent->pNext;
}
A much better approach was submitted by 4Guys visitor George R., a Microsoft interviewer/employee. He recommended using the following technique, which is in time O(N) and space O(1).
Use two pointers.
// error checking and checking for NULL at end of list omitted
p1 = p2 = head;
do {
p1 = p1-gt;next;
p2 = p2-gt;next->next;
} while (p1 != p2);
p2 is moving through the list twice as fast as p1. If the list is circular, (i.e. a cycle exists) it will eventually get around to that sluggard, p1.
struct Node
{
...
bool bVisited;
};
Then, to determine whether a node has a loop, you could first set this flag to false for all of the nodes:
// Detect cycle
// Note: pHead points to the head of the list (assume already exists)
Node *pCurrent = pHead;
while (pCurrent)
{
pCurrent->bVisited = false;
pCurrent = pCurrent->pNext;
}
A much better approach was submitted by 4Guys visitor George R., a Microsoft interviewer/employee. He recommended using the following technique, which is in time O(N) and space O(1).
Use two pointers.
// error checking and checking for NULL at end of list omitted
p1 = p2 = head;
do {
p1 = p1-gt;next;
p2 = p2-gt;next->next;
} while (p1 != p2);
p2 is moving through the list twice as fast as p1. If the list is circular, (i.e. a cycle exists) it will eventually get around to that sluggard, p1.