[Educational Codeforces Round 81 (Rated for Div. 2)]E. Permutation Separation(线段树,思维,前缀和)

E. Permutation Separation

time limit per test

2 seconds

memory limit per test

256 megabytes

input

standard input

output

standard output

You are given a permutation p1,p2,…,pnp1,p2,…,pn (an array where each integer from 11 to nn appears exactly once). The weight of the ii-th element of this permutation is aiai.

At first, you separate your permutation into two non-empty sets — prefix and suffix. More formally, the first set contains elements p1,p2,…,pkp1,p2,…,pk, the second — pk+1,pk+2,…,pnpk+1,pk+2,…,pn, where 1≤k<n1≤k<n.

After that, you may move elements between sets. The operation you are allowed to do is to choose some element of the first set and move it to the second set, or vice versa (move from the second set to the first). You have to pay aiai dollars to move the element pipi.

Your goal is to make it so that each element of the first set is less than each element of the second set. Note that if one of the sets is empty, this condition is met.

For example, if p=[3,1,2]p=[3,1,2] and a=[7,1,4]a=[7,1,4], then the optimal strategy is: separate pp into two parts [3,1][3,1] and [2][2] and then move the 22-element into first set (it costs 44). And if p=[3,5,1,6,2,4]p=[3,5,1,6,2,4], a=[9,1,9,9,1,9]a=[9,1,9,9,1,9], then the optimal strategy is: separate pp into two parts [3,5,1][3,5,1] and [6,2,4][6,2,4], and then move the 22-element into first set (it costs 11), and 55-element into second set (it also costs 11).

Calculate the minimum number of dollars you have to spend.

Input

The first line contains one integer nn (2≤n≤2⋅1052≤n≤2⋅105) — the length of permutation.

The second line contains nn integers p1,p2,…,pnp1,p2,…,pn (1≤pi≤n1≤pi≤n). It's guaranteed that this sequence contains each element from 11 to nn exactly once.

The third line contains nn integers a1,a2,…,ana1,a2,…,an (1≤ai≤1091≤ai≤109).

Output

Print one integer — the minimum number of dollars you have to spend.

Examples

input

Copy

3
3 1 2
7 1 4

output

Copy

4

input

Copy

4
2 4 1 3
5 9 8 3

output

Copy

3

input

Copy

6
3 5 1 6 2 4
9 1 9 9 1 9

output

Copy

2

题意:

给定一个整数n以及一个1~n的全排列,和一个数组a[i]

让你找一个位置划开,将排列分成2个非空的部分,

然后将p[i] 移动到另一个部分的成本是a[i],

问使左边的所有数都比右边的所有的数小需要的最小成本是多少?

思路:

在区间\([1,n-1]\)枚举切割位置i (从p[i] 右边切开,不取n是因为要保证2部分初始非空) ,时间复杂度\(O(n)\)

那么我们知道所有p[i] 位置后面的小于等于 i 的数都要移动到左边,p[i] 位置即前面的大于 i 的数都要移动到右边。

我们尚若可以\(O(logn)\) 时间内得到上面移动操作的cost就可以维护出最小值。

我们想到了线段树。

首先对a[i] 做一个前缀和 数组 sum[i] ,代表将1~i的数都移动到右边需要的cost。

以sum[i] 数组建立区间修改,区间查询最小值的线段树。

然后从1到n-1枚举割切位置i,

对于第i个位置,我们将找到i在p中的位置 id ,将区间\([1,id-1]\) 减去 a[id],区间\([id,n-1]\) 加上a[id]

意义为:在切割位置大于等于i 的时候,不需要将数字i移动到右边部分,又结合线段树维护的是前缀和数组,

那么对于每一个位置i 用线段树的根节点维护的最小值去更新答案ans即可。

代码:

#include <iostream>
#include <cstdio>
#include <cstring>
#include <algorithm>
#include <cmath>
#include <queue>
#include <stack>
#include <map>
#include <set>
#include <vector>
#include <iomanip>
#define ALL(x) (x).begin(), (x).end()
#define sz(a) int(a.size())
#define rep(i,x,n) for(int i=x;i<n;i++)
#define repd(i,x,n) for(int i=x;i<=n;i++)
#define pii pair<int,int>
#define pll pair<long long ,long long>
#define gbtb ios::sync_with_stdio(false),cin.tie(0),cout.tie(0)
#define MS0(X) memset((X), 0, sizeof((X)))
#define MSC0(X) memset((X), '\0', sizeof((X)))
#define pb push_back
#define mp make_pair
#define fi first
#define se second
#define eps 1e-6
#define chu(x) cout<<"["<<#x<<" "<<(x)<<"]"<<endl
#define du3(a,b,c) scanf("%d %d %d",&(a),&(b),&(c))
#define du2(a,b) scanf("%d %d",&(a),&(b))
#define du1(a) scanf("%d",&(a));
using namespace std;
typedef long long ll;
ll gcd(ll a, ll b) {return b ? gcd(b, a % b) : a;}
ll lcm(ll a, ll b) {return a / gcd(a, b) * b;}
ll powmod(ll a, ll b, ll MOD) { if (a == 0ll) {return 0ll;} a %= MOD; ll ans = 1; while (b) {if (b & 1) {ans = ans * a % MOD;} a = a * a % MOD; b >>= 1;} return ans;}
void Pv(const vector<int> &V) {int Len = sz(V); for (int i = 0; i < Len; ++i) {printf("%d", V[i] ); if (i != Len - 1) {printf(" ");} else {printf("\n");}}}
void Pvl(const vector<ll> &V) {int Len = sz(V); for (int i = 0; i < Len; ++i) {printf("%lld", V[i] ); if (i != Len - 1) {printf(" ");} else {printf("\n");}}}
inline long long readll() {long long tmp = 0, fh = 1; char c = getchar(); while (c < '0' || c > '9') {if (c == '-') fh = -1; c = getchar();} while (c >= '0' && c <= '9') tmp = tmp * 10 + c - 48, c = getchar(); return tmp * fh;}
inline int readint() {int tmp = 0, fh = 1; char c = getchar(); while (c < '0' || c > '9') {if (c == '-') fh = -1; c = getchar();} while (c >= '0' && c <= '9') tmp = tmp * 10 + c - 48, c = getchar(); return tmp * fh;}
const int maxn = 300010;
const int inf = 0x3f3f3f3f;
/*** TEMPLATE CODE * * STARTS HERE ***/
int n;
int p[maxn];
int id[maxn];
ll a[maxn];
ll sum[maxn];
struct node
{
int l, r;
ll laze;
ll val;
} segment_tree[maxn << 2];
void pushup(int rt)
{
segment_tree[rt].val = min(segment_tree[rt << 1].val, segment_tree[rt << 1 | 1].val);
}
void build(int rt, int l, int r)
{
segment_tree[rt].l = l;
segment_tree[rt].r = r;
if (l == r)
{
segment_tree[rt].val = sum[l];
segment_tree[rt].laze = 0ll;
return ;
}
int mid = (l + r) >> 1;
build(rt << 1, l, mid);
build(rt << 1 | 1, mid + 1, r);
pushup(rt);
}
void pushdown(int rt)
{
segment_tree[rt << 1 | 1].val += segment_tree[rt].laze;
segment_tree[rt << 1 | 1].laze += segment_tree[rt].laze;
segment_tree[rt << 1 ].val += segment_tree[rt].laze;
segment_tree[rt << 1 ].laze += segment_tree[rt].laze;
segment_tree[rt].laze = 0;
}
void update(int rt, int l, int r, int num)
{
if (segment_tree[rt].laze && l != r)
{
pushdown(rt);
}
if (segment_tree[rt].l >= l && segment_tree[rt].r <= r)
{
segment_tree[rt].val += num;
segment_tree[rt].laze += num;
return ;
}
int mid = segment_tree[rt].l + segment_tree[rt].r >> 1;
if (r > mid)
{
update(rt << 1 | 1, l, r, num);
}
if (l <= mid)
{
update(rt << 1, l, r, num);
}
pushup(rt);
}
int main()
{
//freopen("D:\\code\\text\\input.txt","r",stdin);
//freopen("D:\\code\\text\\output.txt","w",stdout);
n = readint();
repd(i, 1, n)
{
p[i] = readint();
id[p[i]] = i;
}
repd(i, 1, n)
{
a[i] = readll();
}
repd(i, 1, n)
{
sum[i] = sum[i - 1] + a[i];
}
ll ans = a[n];
build(1, 1, n - 1);
ans = min(ans, segment_tree[1].val);
int x;
repd(i, 1, n)
{
x = id[i];
if (x != n)
{
update(1, x, n - 1, -a[x]);
}
if (x != 1)
{
update(1, 1, x - 1, a[x]);
}
ans = min(ans, segment_tree[1].val);
}
printf("%lld\n", ans ); return 0;
}

[Educational Codeforces Round 81 (Rated for Div. 2)]E. Permutation Separation(线段树,思维,前缀和)的更多相关文章

  1. Educational Codeforces Round 51 (Rated for Div. 2) G. Distinctification(线段树合并 + 并查集)

    题意 给出一个长度为 \(n\) 序列 , 每个位置有 \(a_i , b_i\) 两个参数 , \(b_i\) 互不相同 ,你可以进行任意次如下的两种操作 : 若存在 \(j \not = i\) ...

  2. Educational Codeforces Round 47 (Rated for Div. 2)F. Dominant Indices 线段树合并

    题意:有一棵树,对于每个点求子树中离他深度最多的深度是多少, 题解:线段树合并快如闪电,每个节点开一个权值线段树,递归时合并即可,然后维护区间最多的是哪个权值,到x的深度就是到根的深度减去x到根的深度 ...

  3. Educational Codeforces Round 37 (Rated for Div. 2)C. Swap Adjacent Elements (思维,前缀和)

    Educational Codeforces Round 37 (Rated for Div. 2)C. Swap Adjacent Elements time limit per test 1 se ...

  4. Educational Codeforces Round 81 (Rated for Div. 2) A-E简要题解

    链接:https://codeforces.com/contest/1295 A. Display The Number 贪心思路,尽可能放置更多位,如果n为奇数,消耗3去放置一个7,剩下的放1 AC ...

  5. Educational Codeforces Round 81 (Rated for Div. 2) B. Infinite Prefixes

    题目链接:http://codeforces.com/contest/1295/problem/B 题目:给定由0,1组成的字符串s,长度为n,定义t = sssssss.....一个无限长的字符串. ...

  6. Educational Codeforces Round 81 (Rated for Div. 2) C. Obtain The String

    题目链接:http://codeforces.com/contest/1295/problem/C 题目:给定字符串s,t.  给定一个空串z,需要按照规则把z构造成 string z == stri ...

  7. Educational Codeforces Round 81 (Rated for Div. 2)

    A 0~9需要多少笔画,自取7和1,判奇偶 #include<bits/stdc++.h> using namespace std; #define ll long long #defin ...

  8. Educational Codeforces Round 81 (Rated for Div. 2) 题解

    过了n天补的题解:D AB就不用说了 C. Obtain The String 思路挺简单的,就是贪心,但是直接贪心的复杂度是O(|s|*|t|),会超时,所以需要用到序列自动机 虽然名字很高端但是就 ...

  9. Educational Codeforces Round 81 (Rated for Div. 2)E(线段树)

    预处理把左集划分为大小为1~i-1时,把全部元素都移动到右集的代价,记作sum[i]. 然后枚举终态时左集的大小,更新把元素i 留在/移动到 左集的代价. 树状数组/线段树处理区间修改/区间查询 #d ...

随机推荐

  1. 最全的Java操作Redis的工具类,使用StringRedisTemplate实现,封装了对Redis五种基本类型的各种操作!

    转载自:https://github.com/whvcse/RedisUtil 代码 ProtoStuffSerializerUtil.java import java.io.ByteArrayInp ...

  2. 创业学习--《预判行业机会》--B-2.预判模块---HHR计划--以太一堂

    一,<开始学习> 1,行业机会的判断,是可以通过不断地训练提高自己的判准的概率的,要科学思考创业. 2,创业者在行业机会上的三个问题: a. 对市场变化,敏感性太弱,没有洞察行业的意识. ...

  3. 吴裕雄 Bootstrap 前端框架开发——Bootstrap 网格系统实例:堆叠的水平

    <!DOCTYPE html> <html> <head> <meta charset="utf-8"> <title> ...

  4. onclick="this.src=this.src+'?'"是什么意思?

    onclick="this.src=this.src+'?'" 这是表示当前图片链接 在当前链接值的基础上添加了一个问号 譬如当前src="check.aspx" ...

  5. 各大厂商发力5G新机,未来全球手机市场或将呈现新格局

    随着5G商用将正式于今年开启落地,运营商和手机厂商都在为新一代网络制式积极做好准备.对于运营商来说,它们在不断增加5G基站的建设,让5G信号覆盖更广泛的范围.而对于手机厂商来说,它们在努力推出旗下的5 ...

  6. iOS 使用 Xcode8 制作动态库及静态库

    在使用第三方 SDK 时,经常遇到他们提供的仅仅只有一个动态或静态库,并不能获取源码.使用动态库 FrameWork 或 静态库 Lib,可以满足不想把核心代码的具体实现向使用者展示,又能避免其他人错 ...

  7. Python基础语法笔记2

    ------------------------------------------------------------------------------- 常量和Pylint的规范 1.常量:常量 ...

  8. 「ZJOI2011」最小割

    「ZJOI2011」最小割 传送门 建出最小割树,然后暴力计算任意两点之间最小割即可. 多组数据记得初始化. 参考代码: #include <algorithm> #include < ...

  9. HashMap中推荐使用entrySet方式遍历Map类集合KV而不是keySet方式遍历

    我们先来做一个测试 public class HashMapTest { private HashMap<String, String> map = new HashMap<> ...

  10. java动态代理中的invoke方法是如何被自动调用的

    转载声明:本文转载至 zcc_0015的专栏 一.动态代理与静态代理的区别. (1)Proxy类的代码被固定下来,不会因为业务的逐渐庞大而庞大:(2)可以实现AOP编程,这是静态代理无法实现的:(3) ...