|
| 1 | +## 题目地址(690. Maximize the Number of Equivalent Pairs After Swaps) |
| 2 | + |
| 3 | +https://binarysearch.com/problems/Maximize-the-Number-of-Equivalent-Pairs-After-Swaps |
| 4 | + |
| 5 | +## 题目描述 |
| 6 | + |
| 7 | +``` |
| 8 | +You are given a list of integers of the same length A and B. You are also given a two-dimensional list of integers C where each element is of the form [i, j] which means that you can swap A[i] and A[j] as many times as you want. |
| 9 | +
|
| 10 | +Return the maximum number of pairs where A[i] = B[i] after the swapping. |
| 11 | +
|
| 12 | +Constraints |
| 13 | +
|
| 14 | +n ≤ 100,000 where n is the length of A and B |
| 15 | +m ≤ 100,000 where m is the length of C |
| 16 | +Example 1 |
| 17 | +Input |
| 18 | +A = [1, 2, 3, 4] |
| 19 | +B = [2, 1, 4, 3] |
| 20 | +C = [ |
| 21 | + [0, 1], |
| 22 | + [2, 3] |
| 23 | +] |
| 24 | +Output |
| 25 | +4 |
| 26 | +Explanation |
| 27 | +We can swap A[0] with A[1] then A[2] with A[3]. |
| 28 | +``` |
| 29 | + |
| 30 | +## 前置知识 |
| 31 | + |
| 32 | +- 并查集 |
| 33 | +- BFS |
| 34 | +- DFS |
| 35 | + |
| 36 | +## 并查集 |
| 37 | + |
| 38 | +### 思路 |
| 39 | + |
| 40 | +这道题的核心在于如果 A 中的 [0,1] 可以互换,并且 [1,2] 可以互换,那么 [0,1,2] 可以任意互换。 |
| 41 | + |
| 42 | +也就是说互换**具有联通性**。这种联通性对做题有帮助么?有的! |
| 43 | + |
| 44 | +根据 C 的互换关系,我们可以将 A 分为若干联通域。对于每一个联通域我们可以任意互换。因此我们可以枚举每一个联通域,对联通域中的每一个索引 i ,我们看一下 B 中是否有对应 B[j] == A[i] 其中 i 和 j 为同一个联通域的两个点。 |
| 45 | + |
| 46 | +具体算法: |
| 47 | + |
| 48 | +- 首先根据 C 构建并查集。 |
| 49 | +- 然后根据将每一个联通域存到一个字典 group 中,其中 group[i] = list,i 为联通域的元,list 为联通域的点集合列表。 |
| 50 | +- 枚举每一个联通域,对联通域中的每一个索引 i ,我们看一下 B 中是否有对应 B[j] == A[i] 其中 i 和 j 为同一个联通域的两个点。累加答案即可 |
| 51 | + |
| 52 | +### 代码 |
| 53 | + |
| 54 | +代码支持:Python3 |
| 55 | + |
| 56 | +Python3 Code: |
| 57 | + |
| 58 | +```py |
| 59 | + |
| 60 | +class UF: |
| 61 | + def __init__(self, M): |
| 62 | + self.parent = {} |
| 63 | + self.cnt = 0 |
| 64 | + # 初始化 parent,size 和 cnt |
| 65 | + for i in range(M): |
| 66 | + self.parent[i] = i |
| 67 | + self.cnt += 1 |
| 68 | + |
| 69 | + def find(self, x): |
| 70 | + if x != self.parent[x]: |
| 71 | + self.parent[x] = self.find(self.parent[x]) |
| 72 | + return self.parent[x] |
| 73 | + return x |
| 74 | + def union(self, p, q): |
| 75 | + if self.connected(p, q): return |
| 76 | + leader_p = self.find(p) |
| 77 | + leader_q = self.find(q) |
| 78 | + self.parent[leader_p] = leader_q |
| 79 | + self.cnt -= 1 |
| 80 | + def connected(self, p, q): |
| 81 | + return self.find(p) == self.find(q) |
| 82 | + |
| 83 | +class Solution: |
| 84 | + def solve(self, A, B, C): |
| 85 | + n = len(A) |
| 86 | + uf = UF(n) |
| 87 | + for fr, to in C: |
| 88 | + print(fr, to) |
| 89 | + uf.union(fr, to) |
| 90 | + group = collections.defaultdict(list) |
| 91 | + |
| 92 | + for i in uf.parent: |
| 93 | + group[uf.find(i)].append(i) |
| 94 | + ans = 0 |
| 95 | + for i in group: |
| 96 | + indices = group[i] |
| 97 | + values = collections.Counter([A[i] for i in indices]) |
| 98 | + for i in indices: |
| 99 | + if values[B[i]] > 0: |
| 100 | + values[B[i]] -= 1 |
| 101 | + ans += 1 |
| 102 | + return ans |
| 103 | + |
| 104 | +``` |
| 105 | + |
| 106 | +**复杂度分析** |
| 107 | + |
| 108 | +令 n 为数组 A 的长度,v 为图的点数,e 为图的边数。 |
| 109 | + |
| 110 | +- 时间复杂度:$O(n+v+e)$ |
| 111 | +- 空间复杂度:$O(n)$ |
| 112 | + |
| 113 | +## 总结 |
| 114 | + |
| 115 | +我们也可以使用 BFS 或者 DFS 来生成 group,生成 group 后的逻辑大家都是一样的,这两种解法留给大家来实现吧。 |
| 116 | + |
| 117 | +力扣的小伙伴可以[关注我](https://leetcode-cn.com/u/fe-lucifer/),这样就会第一时间收到我的动态啦~ |
| 118 | + |
| 119 | +以上就是本文的全部内容了。大家对此有何看法,欢迎给我留言,我有时间都会一一查看回答。更多算法套路可以访问我的 LeetCode 题解仓库:https://github.com/azl397985856/leetcode 。 目前已经 46K star 啦。大家也可以关注我的公众号《力扣加加》带你啃下算法这块硬骨头。 |
0 commit comments