[分块分治]CodeForce #567C Points on Plane

题面

传送门
C. Points on Plane
time limit per test2 seconds
memory limit per test256 megabytes
inputstandard input
outputstandard output
On a plane are n points (xi, yi) with integer coordinates between 0 and 106. The distance between the two points with numbers a and b is said to be the following value: (the distance calculated by such formula is called Manhattan distance).

We call a hamiltonian path to be some permutation pi of numbers from 1 to n. We say that the length of this path is value .

Find some hamiltonian path with a length of no more than 25 × 108. Note that you do not have to minimize the path length.

Input

The first line contains integer n (1 ≤ n ≤ 106).

The i + 1-th line contains the coordinates of the i-th point: xi and yi (0 ≤ xi, yi ≤ 106).

It is guaranteed that no two points coincide.

Output

Print the permutation of numbers pi from 1 to n — the sought Hamiltonian path. The permutation must meet the inequality .

If there are multiple possible answers, print any of them.

It is guaranteed that the answer exists.

Examples

input

5
0 7
8 10
3 4
5 0
9 12

output

4 3 1 2 5 

Note

In the sample test the total distance is:

(|5 - 3| + |0 - 4|) + (|3 - 0| + |4 - 7|) + (|0 - 8| + |7 - 10|) + (|8 - 9| + |10 - 12|) = 2 + 4 + 3 + 3 + 8 + 3 + 1 + 2 = 26

解题思路

分块思维，带有一定的贪心思想。
开始我们考虑先将X的距离优化至最短，那么就把点按坐标排序，这样最坏情况（排好序后每相邻点Y距离都是1e6）哈密顿距离和是1e12显然不行，那么再考虑分块，如果在X最短的基础上X轴每1000分成1块，在每块中再对y排序，那么每块最坏情况2e6（排好序后块中每相邻点X距离都是1e3）这样,极限哈密顿距离和是2e9<2.5e9，就可以接受了
但是实现的时候我犯了个蠢，这个根本没有X轴每1000分成1块，而是每1000点分一块，感觉数据强可能会卡掉，虽然A了，但也是侥幸，特此记录引以为戒。

错误示范

#include<cstdio>
#include<cstdlib>
#include<algorithm>
#include<cmath>
#include<cstring>
using namespace std;

struct node
{
    int x;
    int y;
    int sn;
};

bool comp1(node a, node b){
    return a.x < b.x;
}

bool comp2(node a, node b){
    return a.y < b.y;
}

node co[1000001];
int n;

int main(){
    scanf("%d", &n);
    for(int i = 0; i < n; i++){
        scanf("%d%d", &co[i].x, &co[i].y);
        co[i].sn = i;
    }
    sort(co, co+n, comp1);
    int i;
    for(i = 1005; i < n; i+=1005){
        sort(co+i-1005, co+i, comp2);
    }
    sort(co+i-1005, co+n, comp2);
    for(i = 0; i < n; i++){
        printf("%d ", co[i].sn + 1);
    }
    // system("pause");
    return 0;
}

正解

#include<cstdio>
#include<cstdlib>
#include<algorithm>
#include<cmath>
#include<cstring>
using namespace std;

struct node
{
    int x;
    int y;
    int sn;
};

bool comp1(node a, node b){
    if(a.x==b.x)return a.y<b.y;
    return a.x < b.x;
}


node co[1000001];
int n;

int main(){
    scanf("%d", &n);
    for(int i = 0; i < n; i++){
        scanf("%d%d", &co[i].x, &co[i].y);
        co[i].x/=1005;//反正块内x顺序不重要
        co[i].sn = i;
    }
    sort(co, co+n, comp1);
    for(int i = 0; i < n; i++){
        printf("%d ", co[i].sn + 1);
    }
    // system("pause");
    return 0;
}