【Codefoces487E/UOJ#30】Tourists Tarjan 点双连通分量 + 树链剖分

时间:2023-03-09 07:00:07
【Codefoces487E/UOJ#30】Tourists       Tarjan 点双连通分量 + 树链剖分

E. Tourists

time limit per test:

2 seconds

memory limit per test:

256 megabytes

input:

standard input

output

:standard output

There are n cities in Cyberland, numbered from 1 to n, connected by m bidirectional roads. The j-th road connects city aj and bj.

For tourists, souvenirs are sold in every city of Cyberland. In particular, city i sell it at a price of wi.

Now there are q queries for you to handle. There are two types of queries:

  • "C a w": The price in city a is changed to w.
  • "A a b": Now a tourist will travel from city a to b. He will choose a route, he also doesn't want to visit a city twice. He will buy souvenirs at the city where the souvenirs are the cheapest (possibly exactly at city a or b). You should output the minimum possible price that he can buy the souvenirs during his travel.

More formally, we can define routes as follow:

  • A route is a sequence of cities [x1, x2, ..., xk], where k is a certain positive integer.
  • For any 1 ≤ i < j ≤ k, xi ≠ xj.
  • For any 1 ≤ i < k, there is a road connecting xi and xi + 1.
  • The minimum price of the route is min(wx1, wx2, ..., wxk).
  • The required answer is the minimum value of the minimum prices of all valid routes from a to b.

Input

The first line of input contains three integers n, m, q (1 ≤ n, m, q ≤ 105), separated by a single space.

Next n lines contain integers wi (1 ≤ wi ≤ 109).

Next m lines contain pairs of space-separated integers aj and bj (1 ≤ aj, bj ≤ n, aj ≠ bj).

It is guaranteed that there is at most one road connecting the same pair of cities. There is always at least one valid route between any two cities.

Next q lines each describe a query. The format is "C a w" or "A a b" (1 ≤ a, b ≤ n, 1 ≤ w ≤ 109).

Output

For each query of type "A", output the corresponding answer.

Examples

input
3 3 3
1
2
3
1 2
2 3
1 3
A 2 3
C 1 5
A 2 3
output
1
2
input
7 9 4
1
2
3
4
5
6
7
1 2
2 5
1 5
2 3
3 4
2 4
5 6
6 7
5 7
A 2 3
A 6 4
A 6 7
A 3 3

output

2
1
5
3

Note

For the second sample, an optimal routes are:

From 2 to 3 it is [2, 3].

From 6 to 4 it is [6, 5, 1, 2, 4].

From 6 to 7 it is [6, 5, 7].

From 3 to 3 it is [3].

【Codefoces487E/UOJ#30】Tourists       Tarjan 点双连通分量 + 树链剖分

Solution

中文题面见UOJ#30  : UOJ#30

思路比较显然,把图缩点重建,再利用数据结构去维护。

要求简单路径,显然可以考虑点双连通分量,我们先对图Tarjan求点双连通分量,然后将这些BCC维护一个最小值,然后缩成一个点,这样就形成一棵树,就可以树剖。

但是这里有一个问题,割点可能会存在于多个点双连通分量中,而这会比较蛋疼,因为割点的答案显然应该是距离它最小的那个。

所以我们对割点单独处理,每个割点新建一个点单独向它所在的所有点双连通分量连一条边; 这样我们就可以在询问的时候,搞定割点的问题了。

那么修改一个点,这个点所在点双的值就有可能发生改变,如果修改割点,那么就会对很多点双造成影响,所以要特殊处理这种情况。

但是我们发现,我们这样建出来的新图(树)一定是满足 块->割点->块->割点 的。

所以我们不妨定义一个点双维护的是,这个点双中除了它fa的割点的所有点的权值+连向它的割点的权值最小,这样如果修改一个割点,就只会影响到它的fa点双。

查询的时候,我们查询$<u,v>$,如果$LCA(u,v)$是一个割点,我们可以直接统计答案。 如果$LCA(u,v)$是一个点双,那么我们还得查它的fa割点的值。

点双内的最小值,可以用multiset来维护。

这样的查询复杂度是$O(log^{2}N)$,修改的复杂度是$O(logN)$,总的复杂度就是$O(Nlog^{2}N)$

对于Tarjan点双连通分量,网上有些人说要边入栈,但很多人却写的点入栈,其实都是可以的。 但是这道题,如果用边入栈的方法,在连边的时候会连出问题的,需要额外判,所以就用来点入栈的方法。

Code

#include<iostream>
#include<cstdio>
#include<algorithm>
#include<cmath>
#include<vector>
#include<set>
#include<cstring>
using namespace std;
inline int read()
{
int x=; char ch=getchar();
while (ch<'' || ch>'') {ch=getchar();}
while (ch>='' && ch<='') {x=x*+ch-''; ch=getchar();}
return x;
}
#define INF 0x7fffffff
#define MAXN 100100
int N,M,Q,val[MAXN],Val[MAXN<<];
struct RoadNode{int next,to;}road[MAXN<<];
int tot=,first[MAXN];
struct EdgeNode{int next,to;}edge[MAXN<<];
int cnt=,head[MAXN<<];
inline void AddEdge(int u,int v) {cnt++; edge[cnt].next=head[u]; head[u]=cnt; edge[cnt].to=v;}
inline void InsertEdge(int u,int v) {AddEdge(u,v); AddEdge(v,u);}
multiset<int>::iterator ist;
struct BlockNode
{
multiset<int>st; int val;
inline void insert(int x) {st.insert(x); val=*st.begin();}
inline void change(int x,int y) {st.erase(*st.find(x)); st.insert(y); val=*st.begin();}
}B[MAXN<<];
namespace Graph
{
inline void AddRoad(int u,int v) {tot++; road[tot].next=first[u]; first[u]=tot; road[tot].to=v;}
inline void InsertRoad(int u,int v) {AddRoad(u,v); AddRoad(v,u);}
int dfn[MAXN],low[MAXN],dfsn,bcc,now,belong[MAXN],size[MAXN],cut[MAXN],st[MAXN],top;
inline void Tarjan(int now)
{
dfn[now]=low[now]=++dfsn; st[++top]=now;
for (int i=first[now]; i; i=road[i].next)
if (!dfn[road[i].to])
{
Tarjan(road[i].to); low[now]=min(low[now],low[road[i].to]);
if (dfn[now]<=low[road[i].to])
{
bcc++; cut[now]=; int tp=;
while ()
{
tp=st[top--]; belong[tp]=bcc;
if (cut[tp]) InsertEdge(tp+N,bcc); if (tp==road[i].to) break;
}
InsertEdge(now+N,bcc);
}
}
else low[now]=min(low[now],dfn[road[i].to]);
}
inline void reBuild()
{
for (int i=; i<=N; i++) if (!dfn[i]) Tarjan(i);
for (int i=; i<=N; i++)
{
if (i!=) B[belong[i]].insert(val[i]);
if (cut[i]) B[i+N].insert(INF);
}
}
}
namespace SegmentTree
{
struct SegmentTreeNode{int l,r,minx;}tree[MAXN<<];
#define ls now<<1
#define rs now<<1|1
inline void Update(int now) {tree[now].minx=min(tree[ls].minx,tree[rs].minx);}
inline void BuildTree(int now,int l,int r)
{
tree[now].l=l; tree[now].r=r;
if (l==r) {tree[now].minx=Val[l]; return;}
int mid=(l+r)>>;
BuildTree(ls,l,mid); BuildTree(rs,mid+,r);
Update(now);
}
inline void Change(int now,int pos,int D)
{
int l=tree[now].l,r=tree[now].r;
if (l==r) {tree[now].minx=D; return;}
int mid=(l+r)>>;
if (pos<=mid) Change(ls,pos,D);
if (pos>mid) Change(rs,pos,D);
Update(now);
}
inline int Query(int now,int L,int R)
{
int l=tree[now].l,r=tree[now].r;
if (L<=l && R>=r) return tree[now].minx;
int mid=(l+r)>>,re=INF;
if (L<=mid) re=min(re,Query(ls,L,R));
if (R>mid) re=min(re,Query(rs,L,R));
return re;
}
}
namespace TreePartition
{
int fa[MAXN<<],size[MAXN<<],son[MAXN<<],deep[MAXN<<],pl[MAXN<<],dfn,pre[MAXN<<],top[MAXN<<];
inline void DFS_1(int now)
{
size[now]=;
for (int i=head[now]; i; i=edge[i].next)
if (edge[i].to!=fa[now])
{
fa[edge[i].to]=now;
deep[edge[i].to]=deep[now]+;
DFS_1(edge[i].to);
size[now]+=size[edge[i].to];
if (size[son[now]]<size[edge[i].to]) son[now]=edge[i].to;
}
}
inline void DFS_2(int now,int chain)
{
pl[now]=++dfn; pre[dfn]=now; top[now]=chain; Val[dfn]=B[now].val;
if (son[now]) DFS_2(son[now],chain);
for (int i=head[now]; i; i=edge[i].next)
if (edge[i].to!=fa[now] && edge[i].to!=son[now])
DFS_2(edge[i].to,edge[i].to);
}
inline int LCA(int u,int v)
{
while (top[u]!=top[v])
{
if (deep[top[u]]<deep[top[v]]) swap(u,v);
u=fa[top[u]];
}
if (deep[u]>deep[v]) swap(u,v);
return u;
}
inline int Query(int x,int y)
{
x=Graph::cut[x]? x+N:Graph::belong[x]; y=Graph::cut[y]? y+N:Graph::belong[y];
int re=INF; int lca=LCA(x,y);
if (lca<=Graph::bcc) lca=fa[lca]; re=val[lca-N];
while (top[x]!=top[y])
{
if (deep[top[x]]<deep[top[y]]) swap(x,y);
re=min(re,SegmentTree::Query(,pl[top[x]],pl[x]));
x=fa[top[x]];
}
if (deep[x]>deep[y]) swap(x,y);
re=min(re,SegmentTree::Query(,pl[x],pl[y]));
return re;
}
inline void Change(int pos,int D) {SegmentTree::Change(,pl[pos],D);}
}
int main()
{
N=read(),M=read(),Q=read();
for (int i=; i<=N; i++) val[i]=read();
for (int x,y,i=; i<=M; i++) x=read(),y=read(),Graph::InsertRoad(x,y);
Graph::reBuild();
// for (int i=1; i<=N; i++) printf("%d %d %d %d\n",dfn[i],low[i],cut[i],belong[i]);
TreePartition::DFS_1(N+); TreePartition::DFS_2(N+,N+);
SegmentTree::BuildTree(,,TreePartition::dfn);
while (Q--)
{
char opt[]; scanf("%s",opt); int x=read(),y=read();
switch (opt[])
{
case 'C': if (x!=) {B[Graph::belong[x]].change(val[x],y); TreePartition::Change(Graph::belong[x],B[Graph::belong[x]].val);} val[x]=y; break;
case 'A': if (x==y) printf("%d\n",val[x]); else printf("%d\n",TreePartition::Query(x,y)); break;
}
}
return ;
}

UOJ上跑的还挺快QAQ...就是死活压不到6000ms+....