Vasya has decided to build a zip-line on trees of a nearby forest. He wants the line to be as long as possible but he doesn't remember exactly the heights of all trees in the forest. He is sure that he remembers correct heights of all trees except, possibly,
one of them.

It is known that the forest consists of n trees staying in a row numbered from left to right with integers from 1 to n.
According to Vasya, the height of the i-th tree is equal to hi.
The zip-line of length k should hang over k (1 ≤ k ≤ n)
trees i1, i2, ..., ik (i1 < i2 < ... < ik)
such that their heights form an increasing sequence, that is hi1 < hi2 < ... < hik.

Petya had been in this forest together with Vasya, and he now has q assumptions about the mistake in Vasya's sequence h.
His i-th assumption consists of two integers ai and bi indicating
that, according to Petya, the height of the tree numbered ai is
actually equal to bi.
Note that Petya's assumptions are independent from each other.

Your task is to find the maximum length of a zip-line that can be built over the trees under each of the q assumptions.

In this problem the length of a zip line is considered equal to the number of trees that form this zip-line.

Input

The first line of the input contains two integers n and m (1 ≤ n, m ≤ 400 000) —
the number of the trees in the forest and the number of Petya's assumptions, respectively.

The following line contains n integers hi (1 ≤ hi ≤ 109) —
the heights of trees according to Vasya.

Each of the following m lines contains two integers ai and bi (1 ≤ ai ≤ n, 1 ≤ bi ≤ 109).

Output

For each of the Petya's assumptions output one integer, indicating the maximum length of a zip-line that can be built under this assumption.

Examples
input
4 4
1 2 3 4
1 1
1 4
4 3
4 5
output
4
3
3
4
input
4 2
1 3 2 6
3 5
2 4
output
4
3
Note

Consider the first sample. The first assumption actually coincides with the height remembered by Vasya. In the second assumption the heights of the trees are (4, 2, 3, 4),
in the third one they are (1, 2, 3, 3) and in the fourth one they are (1, 2, 3, 5).

题意:给你n个数,有q个询问,每一次替换c上的位置为d,问替换后的最长严格上升子序列的长度是多少。

思路:如果普通的替换再查找肯定超时了,所以我们要用线段树来处理。我们设f[i],g[i]分别为以i位置为尾点和起始点的最长严格上升子序列的长度,这个可以用线段树O(nlogn)的复杂度求出来,记录最长上升子序列的长度为maxlen,然后我们再设f1[i],f2[i]表示询问i替换后,询问替换的位置为c,以c位置为尾点和起始点的最长上升子序列长度。接下来我们要判断替换的节点是不是"关键点","关键点"的意思是,如果原来序列没有这个位置的点,那么原来序列的最长上升子序列的长度达不到maxlen。那么这个要怎么判断呢,我们可以开一个数组cnt[i],表示对于一个节点j,以j为尾节点的最长上升子序列的长度为i,且f[j]+g[j]-1==maxlen的这样符合条件的j点的个数总和。对于每一个位置,先判断f[i]+g[i]-1是不是等于maxlen,如果等于maxlen,那么我们就把cnt[f[i]]++。然后对于每一个询问,ans=max(是不是为关键点?maxlen:maxlen-1
,f1[i]+g1[i]-1 )。

#include<stdio.h>
#include<stdlib.h>
#include<string.h>
#include<math.h>
#include<vector>
#include<map>
#include<set>
#include<queue>
#include<stack>
#include<string>
#include<bitset>
#include<algorithm>
using namespace std;
typedef long long ll;
typedef long double ldb;
#define inf 99999999
#define pi acos(-1.0)
#define maxn 400050
vector<pair<int,int> >ques[maxn]; //<idx,num>
int ans[maxn],a[maxn],pos[2*maxn];
int f1[maxn],g1[maxn],f[maxn],g[maxn];
int cnt[maxn];
int c[maxn],d[maxn]; struct node1{
struct node{
int l,r,maxlen;
}b[8*maxn];
void build(int l,int r,int th)
{
int mid;
b[th].l=l;b[th].r=r;
b[th].maxlen=0;
if(l==r)return;
mid=(l+r)/2;
build(l,mid,th*2);
build(mid+1,r,th*2+1);
}
int question(int l,int r,int th)
{
int mid;
if(b[th].l==l && b[th].r==r){
return b[th].maxlen;
}
mid=(b[th].l+b[th].r)/2;
if(r<=mid)return question(l,r,th*2);
else if(l>mid)return question(l,r,th*2+1);
return max(question(l,mid,th*2),question(mid+1,r,th*2+1) );
}
void update(int idx,int num,int th)
{
int mid;
if(b[th].l==idx && b[th].r==idx){
b[th].maxlen=num;return;
}
mid=(b[th].l+b[th].r)/2;
if(idx<=mid)update(idx,num,th*2);
else update(idx,num,th*2+1);
b[th].maxlen=max(b[th*2].maxlen,b[th*2+1].maxlen);
}
}L,R; int main()
{
int n,m,i,j,tot;
while(scanf("%d%d",&n,&m)!=EOF)
{
int tot=0;
for(i=1;i<=n;i++){
scanf("%d",&a[i]);
tot++;pos[tot]=a[i];
}
for(i=1;i<=m;i++){
scanf("%d%d",&c[i],&d[i]);
ques[c[i] ].push_back(make_pair(i,d[i] ) );
tot++;pos[tot]=d[i];
}
sort(pos+1,pos+1+tot);
tot=unique(pos+1,pos+1+tot)-pos-1;
L.build(1,tot,1);
R.build(1,tot,1); int maxlen,num,t;
maxlen=0; for(i=1;i<=n;i++){
for(j=0;j<ques[i].size();j++){
t=lower_bound(pos+1,pos+1+tot,ques[i][j].second)-pos;
if(t==1)num=1;
else num=L.question(1,t-1,1)+1;
f1[ques[i][j].first ]=num;
}
t=lower_bound(pos+1,pos+1+tot,a[i])-pos;
if(t==1)f[i]=1;
else f[i]=L.question(1,t-1,1)+1;
L.update(t,f[i],1);
maxlen=max(maxlen,f[i]);
}
for(i=n;i>=1;i--){
for(j=0;j<ques[i].size();j++){
t=lower_bound(pos+1,pos+1+tot,ques[i][j].second)-pos;
if(t==tot)num=1;
else num=R.question(t+1,tot,1)+1;
g1[ques[i][j].first ]=num;
}
t=lower_bound(pos+1,pos+1+tot,a[i])-pos;
if(t==tot)g[i]=1;
else g[i]=R.question(t+1,tot,1)+1;
R.update(t,g[i],1);
}
for(i=1;i<=n;i++){
cnt[i]=0;
} for(i=1;i<=n;i++){
if(f[i]+g[i]-1==maxlen){
cnt[f[i] ]++;
}
} int ans;
for(i=1;i<=m;i++){
if(f[c[i] ]+g[c[i] ]-1==maxlen && cnt[f[c[i] ] ]==1 ){
ans=maxlen-1;
}
else ans=maxlen;
ans=max(f1[i]+g1[i]-1,ans );
printf("%d\n",ans);
}
}
return 0;
} /*
15 14
76 9 32 82 40 91 46 5 12 69 44 97 30 13 29
4 73
13 84
14 51
5 99
7 47
14 32
4 12
11 20
9 65
15 95
10 26
5 25
2 62
11 81
*/

codeforces #345 (Div. 1) D. Zip-line (线段树+最长上升子序列)的更多相关文章

  1. Codeforces #345 Div.1

    Codeforces #345 Div.1 打CF有助于提高做题的正确率. Watchmen 题目描述:求欧拉距离等于曼哈顿距离的点对个数. solution 签到题,其实就是求有多少对点在同一行或同 ...

  2. Codeforces Round #244 (Div. 2) B. Prison Transfer 线段树rmq

    B. Prison Transfer Time Limit: 20 Sec Memory Limit: 256 MB 题目连接 http://codeforces.com/problemset/pro ...

  3. Codeforces Round #397 by Kaspersky Lab and Barcelona Bootcamp (Div. 1 + Div. 2 combined) F. Souvenirs 线段树套set

    F. Souvenirs 题目连接: http://codeforces.com/contest/765/problem/F Description Artsem is on vacation and ...

  4. Codeforces Round #603 (Div. 2) E. Editor(线段树)

    链接: https://codeforces.com/contest/1263/problem/E 题意: The development of a text editor is a hard pro ...

  5. Codeforces Round #222 (Div. 1) D. Developing Game 线段树有效区间合并

    D. Developing Game   Pavel is going to make a game of his dream. However, he knows that he can't mak ...

  6. Codeforces Round #530 (Div. 2) F (树形dp+线段树)

    F. Cookies 链接:http://codeforces.com/contest/1099/problem/F 题意: 给你一棵树,树上有n个节点,每个节点上有ai块饼干,在这个节点上的每块饼干 ...

  7. Codeforces Round #406 (Div. 2) D. Legacy (线段树建图dij)

    D. Legacy time limit per test 2 seconds memory limit per test 256 megabytes input standard input out ...

  8. Codeforces Round #426 (Div. 2) D. The Bakery 线段树优化DP

    D. The Bakery   Some time ago Slastyona the Sweetmaid decided to open her own bakery! She bought req ...

  9. Codeforces Round #305 (Div. 2) D题 (线段树+RMQ)

    D. Mike and Feet time limit per test 1 second memory limit per test 256 megabytes input standard inp ...

随机推荐

  1. 【MyBatis】MyBatis 缓存

    MyBatis 缓存 文章源码 什么是缓存 像大多数的持久化框架一样,MyBatis 也提供了缓存策略,通过缓存策略来减少数据库的查询次数,从而提高性能. Mybatis 中缓存分为一级缓存,二级缓存 ...

  2. LeetCode167 两数之和 II - 输入有序数组

    给定一个已按照升序排列 的有序数组,找到两个数使得它们相加之和等于目标数. 函数应该返回这两个下标值 index1 和 index2,其中 index1 必须小于 index2. 说明: 返回的下标值 ...

  3. unixbench性能测试跑分工具

    UnixBench是一个类unix系(Unix,BSD,Linux)统下的性能测试工具,一个开源工具,被广泛用与测试linux系统主机的性能 所谓跑分工具,不仅各项的测试有得分,最后跑完也会有一个综合 ...

  4. Python输出有颜色的文字

    原创链接: https://www.cnblogs.com/easypython/p/9084426.html   我们在使用python运维与开发的过程中,经常需要打印显示各种信息.海量的信息堆砌在 ...

  5. Flask的“中间件”

    特殊装饰器 from flask import Flask,render_template,request app = Flask(__name__) @app.before_request def ...

  6. ROS教程(二):创建工作空间(图文)

    ros教程:创建工作空间 目录 前言 一.工作空间? 二.创建一个工作空间 1.创建目录 2.编译 前言 使用catkin创建一个工作空间 一.工作空间? 在ROS系统下,我们所有的项目都放在一个工作 ...

  7. python3多进程 进程池 协程并发

    一.进程           我们电脑的应用程序,都是进程,进程是资源分配的单位.进程切换需要的资源最大,效率低.         进程之间相互独立         cpu密集的时候适合用多进程 #多 ...

  8. 同步与异步 Python 有何不同?

    你是否听到人们说过,异步 Python 代码比"普通(或同步)Python 代码更快?果真是那样吗? 1 "同步"和"异步"是什么意思? Web 应用 ...

  9. Soul API 网关源码解析 03

    目标 使用 soul 代理 dubbo 服务 dubbo 服务如何注册到网关的? dubbo 插件是如何工作的? 理清 http --> 网关--> dubbo provider 整条链路 ...

  10. hive 时间相关的函数

    yyyy-MM-dd与yyyyMMdd000000转换的三种方法 方法一:date_format(只支持yyyy-MM-dd -> yyyyMMdd000000) select date_for ...