from typing import List, Tuple # class Tree class LCA_HLD: __slots__ = ( "depth", "parent", "tree", "depthWeighted", "lid", "rid", "_idToNode", "_top", "_heavySon", "_dfn", ) def __init__(self, n: int) -> None: self.depth = [0] * n self.parent = [-1] * n self.tree = [[] for _ in range(n)] self.depthWeighted = [0] * n self.lid = [0] * n self.rid = [0] * n self._idToNode = [0] * n self._top = [0] * n self._heavySon = [0] * n self._dfn = 0 def build(self, root=-1) -> None: if root != -1: self._build(root, -1, 0, 0) self._markTop(root, root) return n = len(self.tree) for i in range(n): if self.parent[i] == -1: self._build(i, -1, 0, 0) self._markTop(i, i) def addEdge(self, from_: int, to: int, weight: int) -> None: self.tree[from_].append((to, weight)) self.tree[to].append((from_, weight)) def addDirectedEdge(self, from_: int, to: int, weight: int) -> None: self.tree[from_].append((to, weight)) def lca(self, u: int, v: int) -> int: while True: if self.lid[u] > self.lid[v]: u, v = v, u if self._top[u] == self._top[v]: return u v = self.parent[self._top[v]] def dist(self, u: int, v: int, weighted=True) -> int: if weighted: return ( self.depthWeighted[u] + self.depthWeighted[v] - 2 * self.depthWeighted[self.lca(u, v)] ) return self.depth[u] + self.depth[v] - 2 * self.depth[self.lca(u, v)] def kthAncestor(self, root: int, k: int) -> int: """k:0-indexed;如果不存在,返回-1. kthAncestor(root,0) = root """ if k > self.depth[root]: return -1 while True: u = self._top[root] if self.lid[root] - k >= self.lid[u]: return self._idToNode[self.lid[root] - k] k -= self.lid[root] - self.lid[u] + 1 root = self.parent[u] def jump(self, from_: int, to: int, step: int) -> int: """ 从 from 节点跳向 to 节点,跳过 step 个节点(0-indexed). 返回跳到的节点,如果不存在这样的节点,返回-1. """ if step == 1: if from_ == to: return -1 if self.isInSubtree(to, from_): return self.kthAncestor(to, self.depth[to] - self.depth[from_] - 1) return self.parent[from_] c = self.lca(from_, to) dac = self.depth[from_] - self.depth[c] dbc = self.depth[to] - self.depth[c] if step > dac + dbc: return -1 if step <= dac: return self.kthAncestor(from_, step) return self.kthAncestor(to, dac + dbc - step) def getPath(self, from_: int, to: int) -> List[int]: res = [] composition = self.getPathDecomposition(from_, to, True) for a, b in composition: if a <= b: res += self._idToNode[a : b + 1] else: res += self._idToNode[a : b - 1 : -1] return res def getPathDecomposition(self, from_: int, to: int, vertex: bool) -> List[Tuple[int, int]]: """返回沿着`路径顺序`的 [起点,终点] 的 欧拉序 `左闭右闭` 数组. 注意不一定沿着欧拉序, 但是沿着路径顺序. """ up, down = [], [] while True: if self._top[from_] == self._top[to]: break if self.lid[from_] < self.lid[to]: down.append((self.lid[self._top[to]], self.lid[to])) to = self.parent[self._top[to]] else: up.append((self.lid[from_], self.lid[self._top[from_]])) from_ = self.parent[self._top[from_]] offset = 1 ^ vertex if self.lid[from_] < self.lid[to]: down.append((self.lid[from_] + offset, self.lid[to])) elif self.lid[to] + offset <= self.lid[from_]: up.append((self.lid[from_], self.lid[to] + offset)) return up + down[::-1] def isInSubtree(self, child: int, root: int) -> bool: """child是否在root的子树中(child和root不能相等)""" return self.lid[root] <= self.lid[child] < self.rid[root] def subtreeSize(self, root: int) -> int: return self.rid[root] - self.lid[root] def id(self, root) -> Tuple[int, int]: """返回 root 的欧拉序区间, 左闭右开, 0-indexed.""" return self.lid[root], self.rid[root] def eid(self, u: int, v: int) -> int: """返回边 u-v 对应的 欧拉序起点编号, 0-indexed.""" id1, id2 = self.lid[u], self.lid[v] return id1 if id1 > id2 else id2 def _build(self, cur: int, pre: int, dep: int, dist: int) -> int: subSize, heavySize, heavySon = 1, 0, -1 for next, weight in self.tree[cur]: if next != pre: nextSize = self._build(next, cur, dep + 1, dist + weight) subSize += nextSize if nextSize > heavySize: heavySize, heavySon = nextSize, next self.depth[cur] = dep self.depthWeighted[cur] = dist self.parent[cur] = pre self._heavySon[cur] = heavySon return subSize def _markTop(self, cur: int, top: int) -> None: self._top[cur] = top self.lid[cur] = self._dfn self._idToNode[self._dfn] = cur self._dfn += 1 if self._heavySon[cur] != -1: self._markTop(self._heavySon[cur], top) for next, _ in self.tree[cur]: if next != self._heavySon[cur] and next != self.parent[cur]: self._markTop(next, next) self.rid[cur] = self._dfn from collections import defaultdict from typing import Callable, Generic, List, TypeVar, Dict T = TypeVar("T") class RerootingForest(Generic[T]): __slots__ = ("adjList", "_n", "_decrement") def __init__(self, n: int, decrement: int = 0): self.adjList = [[] for _ in range(n)] self._n = n self._decrement = decrement def addEdge(self, u: int, v: int) -> None: u -= self._decrement v -= self._decrement self.adjList[u].append(v) self.adjList[v].append(u) def rerooting( self, e: Callable[[int], T], op: Callable[[T, T], T], composition: Callable[[T, int, int, int], T], groupRoot=0, ) -> Dict[int, "T"]: """groupRoot 是联通分量的根节点.""" groupRoot -= self._decrement assert 0 <= groupRoot < self._n parents = defaultdict(lambda: -1) order = [groupRoot] stack = [groupRoot] while stack: cur = stack.pop() for next in self.adjList[cur]: if next == parents[cur]: continue parents[next] = cur order.append(next) stack.append(next) dp1 = dict({v: e(v) for v in order}) dp2 = dict({v: e(v) for v in order}) for cur in order[::-1]: res = e(cur) for next in self.adjList[cur]: if parents[cur] == next: continue dp2[next] = res res = op(res, composition(dp1[next], cur, next, 0)) res = e(cur) for next in self.adjList[cur][::-1]: if parents[cur] == next: continue dp2[next] = op(res, dp2[next]) res = op(res, composition(dp1[next], cur, next, 0)) dp1[cur] = res for newRoot in order[1:]: parent = parents[newRoot] dp2[newRoot] = composition(op(dp2[newRoot], dp2[parent]), parent, newRoot, 1) dp1[newRoot] = op(dp1[newRoot], dp2[newRoot]) return dp1 import sys from typing import Tuple, List, DefaultDict from collections import defaultdict sys.setrecursionlimit(int(1e7)) input = lambda: sys.stdin.readline().rstrip("\r\n") INF = int(4e18) class _UF: __slots__ = ("n", "part", "_parent", "_rank") def __init__(self, n: int): self.n = n self.part = n self._parent = list(range(n)) self._rank = [1] * n def find(self, x: int) -> int: while x != self._parent[x]: self._parent[x] = self._parent[self._parent[x]] x = self._parent[x] return x def union(self, x: int, y: int) -> bool: rootX = self.find(x) rootY = self.find(y) if rootX == rootY: return False if self._rank[rootX] > self._rank[rootY]: rootX, rootY = rootY, rootX self._parent[rootX] = rootY self._rank[rootY] += self._rank[rootX] self.part -= 1 return True def isConnected(self, x: int, y: int) -> bool: return self.find(x) == self.find(y) def getGroups(self) -> DefaultDict[int, List[int]]: groups = defaultdict(list) for key in range(self.n): root = self.find(key) groups[root].append(key) return groups if __name__ == "__main__": n, m, q = map(int, input().split()) # 树中保留的边 edges = list((a - 1, b - 1) for a, b in tuple(map(int, input().split()) for _ in range(m))) queries = list((a - 1, b - 1) for a, b in tuple(map(int, input().split()) for _ in range(q))) tree = [[] for _ in range(n)] uf = _UF(n) LCA = LCA_HLD(n) for u, v in edges: tree[u].append((v, 1)) tree[v].append((u, 1)) uf.union(u, v) LCA.addEdge(u, v, 1) LCA.build(root=-1) res = 0 weights = [0] * n # 特殊点的权重 for u, v in queries: if uf.isConnected(u, v): res += LCA.dist(u, v) else: weights[u] += 1 weights[v] += 1 """求`原树`中每个点到可达的(连通)所有特殊点的距离之和.""" E = Tuple[int, int] # (distSum,size) 树中距离之和, 子树中特殊点的大小 def e(root: int) -> E: return (0, 0) def op(childRes1: E, childRes2: E) -> E: dist1, size1 = childRes1 dist2, size2 = childRes2 return (dist1 + dist2, size1 + size2) def composition(fromRes: E, parent: int, cur: int, direction: int) -> E: """direction: 0: cur -> parent, 1: parent -> cur""" dist, size = fromRes from_ = cur if direction == 0 else parent count = weights[from_] return (dist + size + count, size + count) R = RerootingForest(n) for u, v in edges: R.addEdge(u, v) groups = uf.getGroups() for root, nodes in groups.items(): dp = R.rerooting(e, op, composition, groupRoot=root) values = dp.values() res += min(d for d, _ in values) print(res)