Graph: Depth First Search (N citizens with pairs of friends)

Question

I have solved problem 10608 on UVA Online Judge using Python 3.5.1. My solution works, but it takes too long to run when the online judge evaluates it.

Problem

There is a town with N citizens. It is known that some pairs of people are friends. According to the famous saying that “The friends of my friends are my friends, too” it follows that if A and B are friends and B and C are friends then A and C are friends, too. Your task is to count how many people there are in the largest group of friends.

Input

Input consists of several datasets. The first line of the input consists of a line with the number of test cases to follow.

The first line of each dataset contains tho numbers N and M, where N is the number of town’s citizens (1 ≤ N ≤ 30000) and M is the number of pairs of people (0 ≤ M ≤ 500000), which are known to be friends. Each of the following M lines consists of two integers A and B (1 ≤ A ≤ N, 1 ≤ B ≤ N, A ̸= B) which describe that A and B are friends. There could be repetitions among the given pairs.

Output

The output for each test case should contain (on a line by itself) one number denoting how many people there are in the largest group of friends on a line by itself.

Sample Input

2
3 2
1 2
2 3
10 12
1 2
3 1
3 4
5 4
3 5
4 6
5 2
2 1
7 1
1 2
9 10
8 9

Sample Output

3
7

testCases = int(input())

for x in range(testCases):
    temp = input().split()
    N = int(temp[0])
    M = int(temp[1])

    nodes = []
    edges = []
    for _ in range(M):
        temp = input().split()
        A = int(temp[0])
        B = int(temp[1])
        edges.append([A, B])

    counter = 0
    for y in range(N):
        counter += 1
        nodes.append(counter)

    hashmap = {}
    for h in range(len(nodes)):
        neighbours = []
        for j in range(len(edges)):
            if edges[j].__contains__(nodes[h]):
                index_of_node = edges[j].index(nodes[h])
                if index_of_node == 0:
                    neighbours.append(edges[j][1])
                    hashmap[h + 1] = neighbours
                else:
                    neighbours.append(edges[j][0])
                    hashmap[h + 1] = neighbours
    current_group = 0
    highest_group = 0

    def reset_array():
        visited = []
        for _ in range(1, N + 2):
            visited.append(False)
        return visited

    visited = reset_array()

    def dfs(at):
        if visited[at]:
            return
        else:
            visited[at] = True
            global current_group
            current_group += 1
        if at in hashmap:
            neighbours = hashmap[at]
            for next in neighbours:
                dfs(next)
        else:
            return

    counter = 0
    for i in range(len(nodes)):
        dfs(i + 1)
        if current_group > highest_group:
            highest_group = current_group
        visited = reset_array()
        current_group = 0

    print(highest_group)

Answer 1

Review of your code

---

• You should mostly prefer the in operator instead __contains__. See @HeapOverflow's comment for more details.

• According to PEP 8, you should not prefer CamelCase for variable names. Use snake_case instead.

---

temp = input().split()
N = int(temp[0])
M = int(temp[1])

Can be replaced with

M, N = [int(x) for x in input().split()]

And the same applies for a similar case.

---

  counter = 0
  for y in range(N):
      counter += 1
      nodes.append(counter)

nodes is just equal to the values from 1 to N, which is just equal to list(range(1, N+1))
Therefore, you can remove counter completely.

---

  if index_of_node == 0:
      neighbours.append(edges[j][1])
      hashmap[h+1] = neighbours
  else:
      neighbours.append(edges[j][0])
      hashmap[h+1] = neighbours

Since hashmap[h+1] = neighbours is executed regardless of the if statement, you can move it outside the scope.

---

def reset_array():
    visited = []
    for _ in range(1, N+2):
        visited.append(False)
    return visited

visited is basically just equal to [False] * (N+1).
The whole function can be replaced to return [False] * (N+1)

Also, this is a personal preference, but you don't have to use a function for this.

---

Function dfs

• next is an inbuilt function's name, and therefore it should be avoided.
• neighbours = hashmap[at] Since this is used only once, the assignment is unnecessary
• else: return this is unnecessary, as the function does that anyway

Here's how dfs might look after applying the above changes:

def dfs(at):
    global current_group

    if visited[at]:
        return

    visited[at] = True
    current_group += 1

    if at in hashmap:
        for next_ in hashmap[at]:
            dfs(next_)

---

Faster Algorithm

The dfs will take O(N) time, and since it's executed N times, the time complexity will be O(N^2) which is clearly not feasible.

A disjoint set union on the other hand, will take a lot lesser time.

Here's my accepted code that uses DSU:

for _ in range(int(input())):
    n, m = map(int, input().split())
    dsu = [-1] * n

    for _ in range(m):
        u, v = map(int, input().split())
        u -= 1
        v -= 1

        while dsu[u] >= 0:
            u = dsu[u]

        while dsu[v] >= 0:
            v = dsu[v]

        if u == v:
            continue

        if u > v:
            u, v = v, u

        dsu[u] = dsu[u] + dsu[v]
        dsu[v] = u

    print(-min(dsu))

In case you have any queries about the above code, do ask me in the comments.

---

Also, I guess you're switching to python from java judging by the fact you named a variable hashmap. If that's the case, welcome to world of python!