D. Serega and Fun
time limit per test

4 seconds

memory limit per test

256 megabytes

input

standard input

output

standard output

Serega loves fun. However, everyone has fun in the unique manner. Serega has fun by solving query problems. One day Fedor came up with such a problem.

You are given an array a consisting of n positive integers and queries to it. The queries can be of two types:

  1. Make a unit cyclic shift to the right on the segment from l to r (both borders inclusive). That is rearrange elements of the array in the following manner:a[l], a[l + 1], ..., a[r - 1], a[r] → a[r], a[l], a[l + 1], ..., a[r - 1].
  2. Count how many numbers equal to k are on the segment from l to r (both borders inclusive).

Fedor hurried to see Serega enjoy the problem and Serega solved it really quickly. Let's see, can you solve it?

Input

The first line contains integer n (1 ≤ n ≤ 105) — the number of elements of the array. The second line contains n integers a[1], a[2], ..., a[n] (1 ≤ a[i] ≤ n).

The third line contains a single integer q (1 ≤ q ≤ 105) — the number of queries. The next q lines contain the queries.

As you need to respond to the queries online, the queries will be encoded. A query of the first type will be given in format: 1 l'i r'i. A query of the second type will be given in format: 2 l'i r'i k'i. All the number in input are integer. They satisfy the constraints: 1 ≤ l'i, r'i, k'i ≤ n.

To decode the queries from the data given in input, you need to perform the following transformations:

li = ((l'i + lastans - 1) mod n) + 1; ri = ((r'i + lastans - 1) mod n) + 1; ki = ((k'i + lastans - 1) mod n) + 1.

Where lastans is the last reply to the query of the 2-nd type (initially, lastans = 0). If after transformation li is greater than ri, you must swap these values.

Output

For each query of the 2-nd type print the answer on a single line.

Examples
input
7
6 6 2 7 4 2 5
7
1 3 6
2 2 4 2
2 2 4 7
2 2 2 5
1 2 6
1 1 4
2 1 7 3
output
2
1
0
0
input
8
8 4 2 2 7 7 8 8
8
1 8 8
2 8 1 7
1 8 1
1 7 3
2 8 8 3
1 1 4
1 2 7
1 4 5
output
2
0

题目大意:给你一个长度为n的a数组,然后有2个操作

①输入l, r 把a[l],a[l+1]……,a[r]变成 a[r],a[l],a[l+1]……,a[r-1]

②输入l,r,v,求[l,r]中等于v的数有多少

且强制在线

思路:

用deque维护一个序列即可(md一个字母写错了debug一个下午加晚上,TAT)

不过我是很单纯的每次把deque里面的东西每次都取出来,所以跑了2000+ms,不过这个人的代码跑了500ms左右,大家如果要看可以学习一下:http://blog.csdn.net/blankcqk/article/details/38468729

我的代码:

//看看会不会爆int!数组会不会少了一维!
//取物问题一定要小心先手胜利的条件
#include <bits/stdc++.h>
using namespace std;
#pragma comment(linker,"/STACK:102400000,102400000")
#define LL long long
#define ALL(a) a.begin(), a.end()
#define pb push_back
#define mk make_pair
#define fi first
#define se second
#define haha printf("haha\n")
const int maxn = 1e5 + ;
int a[maxn];
deque<int> que[maxn];
int cnt[][maxn];
int n, q;
int belong[maxn], L[maxn], R[maxn], num, block; void build(){
block = sqrt(n); num = n / block;
if (n % block) num++;
for (int i = ; i <= num; i++)
L[i] = (i - ) * block + , R[i] = i * block;
R[num] = n;
for (int i = ; i <= n; i++)
belong[i] = (i - ) / block + ;
for (int i = ; i <= num; i++)
for (int j = L[i]; j <= R[i]; j++){
cnt[i][a[j]]++;
que[i].push_back(a[j]);
}
} int tmp[], t1[], t2[];
void update(int x, int y){
if (belong[x] == belong[y]){
int px = x - L[belong[x]] + ;///在原来的里面的位置
int py = y - L[belong[y]] + ;
int t = ;
while (!que[belong[x]].empty()){
tmp[++t] = que[belong[x]].front(); que[belong[x]].pop_front();
}
for (int i = ; i <= t; i++){
if (i == px) que[belong[x]].push_back(tmp[py]);
if (i == py) continue;
que[belong[x]].push_back(tmp[i]);
}
return ;
}
int px = x - L[belong[x]] + ;///在原来的里面的位置
int py = y - L[belong[y]] + ;
int tt1 = , tt2 = ;
while (!que[belong[x]].empty()){
t1[++tt1] = que[belong[x]].front(); que[belong[x]].pop_front();
}
while (!que[belong[y]].empty()){
t2[++tt2] = que[belong[y]].front(); que[belong[y]].pop_front();
}
for (int i = ; i <= tt1; i++){
if (i == px) {
que[belong[x]].push_back(t2[py]);
cnt[belong[x]][t2[py]]++;
}
que[belong[x]].push_back(t1[i]);
}
for (int i = ; i <= tt2; i++){
if (i == py) {
cnt[belong[y]][t2[i]]--;
continue;
}
que[belong[y]].push_back(t2[i]);
}
for (int i = belong[x] + ; i <= belong[y]; i++){
int val = que[i - ].back(); que[i - ].pop_back();
cnt[i - ][val]--; cnt[i][val]++;
que[i].push_front(val);
}
} int query(int x, int y, int val){
int ans = ;
if (belong[x] == belong[y]){
int lb = x + - L[belong[x]];
int rb = y + - L[belong[y]];///修改了
int t = ;
while (!que[belong[x]].empty()){
tmp[++t] = que[belong[x]].front(); que[belong[x]].pop_front();
}
for (int i = lb; i <= rb; i++) if (tmp[i] == val) ans++;
for (int i = ; i <= t; i++) que[belong[x]].push_back(tmp[i]);
return ans;
}
///对x的操作
int lb = x + - L[belong[x]], rb = R[belong[x]] + - L[belong[x]];
int t = ;
while (!que[belong[x]].empty()){
tmp[++t] = que[belong[x]].front(); que[belong[x]].pop_front();
}
for (int i = lb; i <= rb; i++) if (tmp[i] == val) ans++;
for (int i = ; i <= t; i++) que[belong[x]].push_back(tmp[i]);
//printf("ans = %d\n", ans); ///对y的操作
lb = , rb = y + - L[belong[y]];
t = ;
while (!que[belong[y]].empty()){
tmp[++t] = que[belong[y]].front(); que[belong[y]].pop_front();
}
for (int i = lb; i <= rb; i++) if (tmp[i] == val) ans++;
for (int i = ; i <= t; i++) que[belong[y]].push_back(tmp[i]);
//printf("ans = %d\n", ans); ///对整体的操作
for (int i = belong[x] + ; i < belong[y]; i++)
ans += cnt[i][val];
return ans;
} int main(){
cin >> n;
for (int i = ; i <= n; i++)
scanf("%d", a + i);
build();
cin >> q;
int lastans = ;
for (int i = ; i <= q; i++){
int ty, l, r, k;
scanf("%d%d%d", &ty, &l, &r);
l = (l + lastans - ) % n + , r = (r + lastans - ) % n + ;
if (l > r) swap(l, r);
if(ty == ){
if (l == r) continue;
update(l, r);
}
else {
scanf("%d", &k);
k = (k + lastans - ) % n + ;
printf("%d\n", lastans = query(l, r, k));
}
}
return ;
}

分块+deque维护 Codeforces Round #260 (Div. 1) D. Serega and Fun的更多相关文章

  1. Codeforces Round #260 (Div. 1) D. Serega and Fun 分块

    D. Serega and Fun Time Limit: 20 Sec Memory Limit: 256 MB 题目连接 http://codeforces.com/contest/455/pro ...

  2. DP Codeforces Round #260 (Div. 1) A. Boredom

    题目传送门 /* 题意:选择a[k]然后a[k]-1和a[k]+1的全部删除,得到点数a[k],问最大点数 DP:状态转移方程:dp[i] = max (dp[i-1], dp[i-2] + (ll) ...

  3. 递推DP Codeforces Round #260 (Div. 1) A. Boredom

    题目传送门 /* DP:从1到最大值,dp[i][1/0] 选或不选,递推更新最大值 */ #include <cstdio> #include <algorithm> #in ...

  4. Codeforces Round #260 (Div. 2)AB

    http://codeforces.com/contest/456/problem/A A. Laptops time limit per test 1 second memory limit per ...

  5. Codeforces Round #260 (Div. 1) C. Civilization 并查集,直径

    C. Civilization Time Limit: 20 Sec Memory Limit: 256 MB 题目连接 http://codeforces.com/contest/455/probl ...

  6. Codeforces Round #260 (Div. 1) A - Boredom DP

    A. Boredom Time Limit: 20 Sec Memory Limit: 256 MB 题目连接 http://codeforces.com/contest/455/problem/A ...

  7. Codeforces Round #260 (Div. 1) A. Boredom (简单dp)

    题目链接:http://codeforces.com/problemset/problem/455/A 给你n个数,要是其中取一个大小为x的数,那x+1和x-1都不能取了,问你最后取完最大的和是多少. ...

  8. 暴力+树状数组维护 Codeforces Round #378 (Div. 2) C

    题目大意:给你一个长度为n的数组a,然后数值大的可以合并数值小的,且合并了以后该数组的长度-1.给你一个长度为k目标数组b,问,是否可以从a数组变到b数组,是就yes并且输出步骤.否就输出no 思路: ...

  9. Codeforces Round #260 (Div. 1) 455 A. Boredom (DP)

    题目链接:http://codeforces.com/problemset/problem/455/A A. Boredom time limit per test 1 second memory l ...

随机推荐

  1. 20181120-6 Beta阶段第2周/共2周 Scrum立会报告+燃尽图 03

    此作业要求参见:[https://edu.cnblogs.com/campus/nenu/2018fall/homework/2411] 版本控制地址    [https://git.coding.n ...

  2. struts通配符*的使用

    <action name="user_*" class="com.wangcf.UserAction" method="{1}"> ...

  3. POJ 2376 (区间问题,贪心)

    题目链接:http://poj.org/problem?id=2376 题目大意:选择一些区间使得能够覆盖1-T中的每一个点,并且区间数最少 题目分析:这道题目很明显可以用贪心法来解决.但题目没有看起 ...

  4. Alpha 冲刺(4/10)

    队名 火箭少男100 组长博客 林燊大哥 作业博客 Alpha 冲鸭鸭鸭鸭! 成员冲刺阶段情况 林燊(组长) 过去两天完成了哪些任务 协调各成员之间的工作 协助前后端接口的开发 测试项目运行的服务器环 ...

  5. C++ Primer Plus学习:第九章

    C++第九章:内存模型与名称空间 C++在内存中存储数据方面提供了多种选择.可直接选择保留在内存中的时间长度(存储持续性)以及程序哪一部分可以访问数据(作用域和链接)等. 单独编译 程序分为三个部分: ...

  6. ssh: Could not resolve hostname问题终于解决了?

    1.如果系统为64位,无法启动启动hdfs: ./sbin/start-dfs.sh.并有以下错误: sed: -e expression #1, char 6: unknown option to  ...

  7. QMap使用

    版权声明:若无来源注明,Techie亮博客文章均为原创. 转载请以链接形式标明本文标题和地址: 本文标题:QMap使用     本文地址:https://www.techieliang.com/201 ...

  8. 【Python】tuple and list 练习

    Tuple: 1.       __add__ a = (,,,,,,,,,) b = (,,,,,,,), c = (,,,,), result = a.__add__(b) print(resul ...

  9. BZOJ 1835 基站选址(DP+线段树)

    # include <cstdio> # include <cstring> # include <cstdlib> # include <iostream& ...

  10. 秒杀多线程第七篇 经典线程同步 互斥量Mutex(续)

    java使用Synchronized关键字实现互斥,而同时有Lock支持. 这两个的效果是等同的,Synchronized性能的起伏较大,而lock比较收敛. 为了代码的可读性,Synchronize ...