내가 정리하는 자료구조 06 - 힙(heap)

대표적인 데이터 구조8: 힙

1. 힙 (Heap) 이란?

  • 힙: 데이터에서 최대값과 최소값을 빠르게 찾기 위해 고안된 완전 이진 트리(Complete Binary Tree)
    • 완전 이진 트리: 노드를 삽입할 때 최하단 왼쪽 노드부터 차례대로 삽입하는 트리

  • 힙을 사용하는 이유
    • 배열에 데이터를 넣고, 최대값과 최소값을 찾으려면 O(n) 이 걸림
    • 이에 반해, 힙에 데이터를 넣고, 최대값과 최소값을 찾으면, $ O(log n) $ 이 걸림
    • 우선순위 큐와 같이 최대값 또는 최소값을 빠르게 찾아야 하는 자료구조 및 알고리즘 구현 등에 활용됨

2. 힙 (Heap) 구조

  • 힙은 최대값을 구하기 위한 구조 (최대 힙, Max Heap) 와, 최소값을 구하기 위한 구조 (최소 힙, Min Heap) 로 분류할 수 있음
  • 힙은 다음과 같이 두 가지 조건을 가지고 있는 자료구조임
  1. 각 노드의 값은 해당 노드의 자식 노드가 가진 값보다 크거나 같다. (최대 힙의 경우)
    • 최소 힙의 경우는 각 노드의 값은 해당 노드의 자식 노드가 가진 값보다 크거나 작음
  2. 완전 이진 트리 형태를 가짐

힙과 이진 탐색 트리의 공통점과 차이점

  • 공통점: 힙과 이진 탐색 트리는 모두 이진 트리임
  • 차이점:
  • 힙은 각 노드의 값이 자식 노드보다 크거나 같음(Max Heap의 경우)
  • 이진 탐색 트리는 왼쪽 자식 노드의 값이 가장 작고, 그 다음 부모 노드, 그 다음 오른쪽 자식 노드 값이 가장 큼
  • 힙은 이진 탐색 트리의 조건인 자식 노드에서 작은 값은 왼쪽, 큰 값은 오른쪽이라는 조건은 없음
    • 힙의 왼쪽 및 오른쪽 자식 노드의 값은 오른쪽이 클 수도 있고, 왼쪽이 클 수도 있음
  • 이진 탐색 트리는 탐색을 위한 구조, 힙은 최대/최소값 검색을 위한 구조 중 하나로 이해하면 됨

3. 힙 (Heap) 동작

  • 데이터를 힙 구조에 삽입, 삭제하는 과정을 그림을 통해 선명하게 이해하기

힙에 데이터 삽입하기 - 기본 동작

  • 힙은 완전 이진 트리이므로, 삽입할 노드는 기본적으로 왼쪽 최하단부 노드부터 채워지는 형태로 삽입

힙의 데이터 삭제하기 (Max Heap 의 예)

  • 보통 삭제는 최상단 노드 (root 노드)를 삭제하는 것이 일반적임
    • 힙의 용도는 최대값 또는 최소값을 root 노드에 놓아서, 최대값과 최소값을 바로 꺼내 쓸 수 있도록 하는 것임
  • 상단의 데이터 삭제시, 가장 최하단부 왼쪽에 위치한 노드 (일반적으로 가장 마지막에 추가한 노드) 를 root 노드로 이동
  • root 노드의 값이 child node 보다 작을 경우, root 노드의 child node 중 가장 큰 값을 가진 노드와 root 노드 위치를 바꿔주는 작업을 반복함 (swap)

4. 힙 구현

힙과 배열

  • 일반적으로 힙 구현시 배열 자료구조를 활용
  • 배열은 인덱스가 0번부터 시작하지만, 힙 구현의 편의를 위해, root 노드 인덱스 번호를 1로 지정하면, 구현이 좀더 수월함
    • 부모 노드 인덱스 번호 (parent node’s index) = 자식 노드 인덱스 번호 (child node’s index) // 2
    • 왼쪽 자식 노드 인덱스 번호 (left child node’s index) = 부모 노드 인덱스 번호 (parent node’s index) * 2
    • 오른쪽 자식 노드 인덱스 번호 (right child node’s index) = 부모 노드 인덱스 번호 (parent node’s index) * 2 + 1
예1 - 10 노드의 부모 노드 인덱스

1
2 // 2

결과

1
1

예1 - 15 노드의 왼쪽 자식 노드 인덱스 번호

1
1 * 2

결과

1
2

예1 - 15 노드의 오른쪽 자식 노드 인덱스 번호

1
2 * 2 + 1

결과

1
5

힙에 데이터 삽입 구현 (Max Heap 예)

  • 힙 클래스 구현1

1
2
3
4
5
6
7
8
class Heap:
def __init__(self, data):
# 배열 구조로 주로 하기 때문에
self.heap_array = list()
# 완전 이진 트리의 성질을 이용하여 부모, 자식노드를 쉽게 찾기 위해
# 인덱싱을 활용하기 위해 인덱스를 1부터 가져간다.
self.heap_array.append(None)
self.heap_array.append(data)

Test

1
2
heap = Heap(1)
heap.heap_array

결과

1
[None, 1]

  • 힙 클래스 구현2 - insert1
    • 인덱스 번호는 1번부터 시작하도록 변경


1
2
3
4
5
6
7
8
9
10
11
12
13
14
class Heap:
def __init__(self, data):
self.heap_array = list()
self.heap_array.append(None)
self.heap_array.append(data)

def insert(self, data):
if len(self.heap_array) == 0:
self.heap_array.append(None)
self.heap_array.append(data)
return True

self.heap_array.append(data)
return True

  • 힙 클래스 구현3 - insert2
    • 삽입한 노드가 부모 노드의 값보다 클 경우, 부모 노드와 삽입한 노드 위치를 바꿈
    • 삽입한 노드가 루트 노드가 되거나, 부모 노드보다 값이 작거나 같을 경우까지 반복

  • 특정 노드의 관련 노드 위치 알아내기
    • 부모 노드 인덱스 번호 (parent node’s index) = 자식 노드 인덱스 번호 (child node’s index) // 2
    • 왼쪽 자식 노드 인덱스 번호 (left child node’s index) = 부모 노드 인덱스 번호 (parent node’s index) * 2
    • 오른쪽 자식 노드 인덱스 번호 (right child node’s index) = 부모 노드 인덱스 번호 (parent node’s index) * 2 + 1


1
2
3
4
5
6
7
heap = Heap(15)
heap.insert(10)
heap.insert(8)
heap.insert(5)
heap.insert(4)
heap.insert(20)
heap.heap_array

결과

1
[None, 20, 10, 15, 5, 4, 8]


1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
class Heap:
def __init__(self, data):
self.heap_array = list()
self.heap_array.append(None)
self.heap_array.append(data)

def move_up(self, inserted_idx):
if inserted_idx <= 1:
return False

parent_idx = inserted_idx // 2
if self.heap_array[inserted_idx] > self.heap_array[parent_idx]:
return True
else:
return False

def insert(self, data):
# 배열에 데이터가 존재하지 않는다면 reset시켜줌.
if len(self.heap_array) == 0:
self.heap_array.append(None)
self.heap_array.append(data)
return True

self.heap_array.append(data)

# None이 0번째 index에 존재하므로
# index를 통해 부모,자식노드를 분별하기 쉽게 -1을 해줌
inserted_idx = len(self.heap_array) - 1

# 이제 index를 통해 heap 구조를 띄도록 추가한 노드의 값이
# Max heap인 경우 부모노드보다 작은지를 확인한 후에 계속해서
# 바꿔주게끔 함수 하나를 만들어 바꿔줘야하는 노드이면 True를 반환하여
# while문을 반복할 수 있게끔 한다.

while self.move_up(inserted_idx):
parent_idx = inserted_idx // 2
self.heap_array[inserted_idx], self.heap_array[parent_idx] = self.heap_array[parent_idx], self.heap_array[inserted_idx]
inserted_idx = parent_idx

return True

힙에 데이터 삭제 구현 (Max Heap 예)

  • 힙 클래스 구현4 - delete1
    • 보통 삭제는 최상단 노드 (root 노드)를 삭제하는 것이 일반적임
    • 힙의 용도는 최대값 또는 최소값을 root 노드에 놓아서, 최대값과 최소값을 바로 꺼내 쓸 수 있도록 하는 것임
  • 힙 클래스 구현4 - delete2
    • 상단의 데이터 삭제시, 가장 최하단부 왼쪽에 위치한 노드 (일반적으로 가장 마지막에 추가한 노드) 를 root 노드로 이동
    • root 노드의 값이 child node 보다 작을 경우, root 노드의 child node 중 가장 큰 값을 가진 노드와 root 노드 위치를 바꿔주는 작업을 반복함 (swap)
  • 특정 노드의 관련 노드 위치 알아내기
    • 부모 노드 인덱스 번호 (parent node’s index) = 자식 노드 인덱스 번호 (child node’s index) // 2
    • 왼쪽 자식 노드 인덱스 번호 (left child node’s index) = 부모 노드 인덱스 번호 (parent node’s index) * 2
    • 오른쪽 자식 노드 인덱스 번호 (right child node’s index) = 부모 노드 인덱스 번호 (parent node’s index) * 2 + 1


1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
class Heap:
def __init__(self, data):
self.heap_array=list()
self.heap_array.append(None)
self.heap_array.append(data)

def move_down(self, popped_idx):
left_child_popped_idx = popped_idx * 2
right_child_popped_idx = popped_idx * 2 + 1

# case1: 왼쪽 자식 노드도 없을 때
if left_child_popped_idx >= len(self.heap_array):
return False
# case2: 오른쪽 자식 노드만 없을 때
elif right_child_popped_idx >= len(self.heap_array):
if self.heap_array[popped_idx] < self.heap_array[left_child_popped_idx]:
return True
else:
return False
# case3: 왼쪽, 오른쪽 자식 노드 모두 있을 때
# 자식노드가 둘다있다면, 먼저 자식노드끼리 비교한다.
else:
if self.heap_array[left_child_popped_idx] > self.heap_array[right_child_popped_idx]:
if self.heap_array[popped_idx] < self.heap_array[left_child_popped_idx]:
return True
else:
return False
else:
if self.heap_array[popped_idx] < self.heap_array[right_child_popped_idx]:
return True
else:
return False

def pop(self):
if len(self.heap_array) <= 1:
return None

returned_data = self.heap_array[1]
self.heap_array[1] = self.heap_array[-1]
del self.heap_array[-1]
popped_idx = 1

while self.move_down(popped_idx):
left_child_popped_idx = popped_idx * 2
right_child_popped_idx = popped_idx * 2 + 1

# case2: 오른쪽 자식 노드만 없을 때
if right_child_popped_idx >= len(self.heap_array):
if self.heap_array[popped_idx] < self.heap_array[left_child_popped_idx]:
self.heap_array[popped_idx], self.heap_array[left_child_popped_idx] = self.heap_array[left_child_popped_idx], self.heap_array[popped_idx]
popped_idx = left_child_popped_idx
# case3: 왼쪽, 오른쪽 자식 노드 모두 있을 때
else:
if self.heap_array[left_child_popped_idx] > self.heap_array[right_child_popped_idx]:
if self.heap_array[popped_idx] < self.heap_array[left_child_popped_idx]:
self.heap_array[popped_idx], self.heap_array[left_child_popped_idx] = self.heap_array[left_child_popped_idx], self.heap_array[popped_idx]
popped_idx = left_child_popped_idx
else:
if self.heap_array[popped_idx] < self.heap_array[right_child_popped_idx]:
self.heap_array[popped_idx], self.heap_array[right_child_popped_idx] = self.heap_array[right_child_popped_idx], self.heap_array[popped_idx]
popped_idx = right_child_popped_idx

return returned_data

Heap 구현

  • 위의 모든 function들을 모아 하나의 class로 만들어 놓은 파일이다.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
class Heap:
def __init__(self, data):
self.heap_array = list()
self.heap_array.append(None)
self.heap_array.append(data)

def move_down(self, popped_idx):
left_child_popped_idx = popped_idx * 2
right_child_popped_idx = popped_idx * 2 + 1

# case1: 왼쪽 자식 노드도 없을 때
if left_child_popped_idx >= len(self.heap_array):
return False
# case2: 오른쪽 자식 노드만 없을 때
elif right_child_popped_idx >= len(self.heap_array):
if self.heap_array[popped_idx] < self.heap_array[left_child_popped_idx]:
return True
else:
return False
# case3: 왼쪽, 오른쪽 자식 노드 모두 있을 때
# 자식노드가 둘다있다면, 먼저 자식노드끼리 비교한다.
else:
if self.heap_array[left_child_popped_idx] > self.heap_array[right_child_popped_idx]:
if self.heap_array[popped_idx] < self.heap_array[left_child_popped_idx]:
return True
else:
return False
else:
if self.heap_array[popped_idx] < self.heap_array[right_child_popped_idx]:
return True
else:
return False

def pop(self):
if len(self.heap_array) <= 1:
return None

returned_data = self.heap_array[1]
self.heap_array[1] = self.heap_array[-1]
del self.heap_array[-1]
popped_idx = 1

while self.move_down(popped_idx):
left_child_popped_idx = popped_idx * 2
right_child_popped_idx = popped_idx * 2 + 1

# case2: 오른쪽 자식 노드만 없을 때
if right_child_popped_idx >= len(self.heap_array):
if self.heap_array[popped_idx] < self.heap_array[left_child_popped_idx]:
self.heap_array[popped_idx], self.heap_array[left_child_popped_idx] = self.heap_array[left_child_popped_idx], self.heap_array[popped_idx]
popped_idx = left_child_popped_idx
# case3: 왼쪽, 오른쪽 자식 노드 모두 있을 때
else:
if self.heap_array[left_child_popped_idx] > self.heap_array[right_child_popped_idx]:
if self.heap_array[popped_idx] < self.heap_array[left_child_popped_idx]:
self.heap_array[popped_idx], self.heap_array[left_child_popped_idx] = self.heap_array[left_child_popped_idx], self.heap_array[popped_idx]
popped_idx = left_child_popped_idx
else:
if self.heap_array[popped_idx] < self.heap_array[right_child_popped_idx]:
self.heap_array[popped_idx], self.heap_array[right_child_popped_idx] = self.heap_array[right_child_popped_idx], self.heap_array[popped_idx]
popped_idx = right_child_popped_idx

return returned_data

def move_up(self, inserted_idx):
if inserted_idx <= 1:
return False
parent_idx = inserted_idx // 2
if self.heap_array[inserted_idx] > self.heap_array[parent_idx]:
return True
else:
return False

def insert(self, data):
if len(self.heap_array) == 1:
self.heap_array.append(data)
return True

self.heap_array.append(data)
inserted_idx = len(self.heap_array) - 1

while self.move_up(inserted_idx):
parent_idx = inserted_idx // 2
self.heap_array[inserted_idx], self.heap_array[parent_idx] = self.heap_array[parent_idx], self.heap_array[inserted_idx]
inserted_idx = parent_idx
return True

Test


1
2
3
4
5
6
7
heap = Heap(15)
heap.insert(10)
heap.insert(8)
heap.insert(5)
heap.insert(4)
heap.insert(20)
heap.heap_array

결과

1
[None, 20, 10, 15, 5, 4, 8]


1
heap.pop()

결과

1
20


1
heap.heap_array

결과

1
[None, 15, 10, 8, 5, 4]

5. 힙 (Heap) 시간 복잡도

  • depth (트리의 높이) 를 h라고 표기한다면,
  • n개의 노드를 가지는 heap 에 데이터 삽입 또는 삭제시, 최악의 경우 root 노드에서 leaf 노드까지 비교해야 하므로 $h = log_2{n} $ 에 가까우므로, 시간 복잡도는 $ O(log{n}) $
    • 참고: 빅오 표기법에서 $log{n}$ 에서의 log의 밑은 10이 아니라, 2입니다.
    • 한번 실행시마다, 50%의 실행할 수도 있는 명령을 제거한다는 의미. 즉 50%의 실행시간을 단축시킬 수 있다는 것을 의미함