CF - 580B. Kefa and Company - 排序+暴力+二分查找上界

1.题目描述：

B. Kefa and Company

time limit per test

2 seconds

memory limit per test

256 megabytes

input

standard input

output

standard output

Kefa wants to celebrate his first big salary by going to restaurant. However, he needs company.

Kefa has n friends, each friend will agree to go to the restaurant if Kefa asks. Each friend is characterized by the amount of money he has and the friendship factor in respect to Kefa. The parrot doesn't want any friend to feel poor compared to somebody else in the company (Kefa doesn't count). A friend feels poor if in the company there is someone who has at least d units of money more than he does. Also, Kefa wants the total friendship factor of the members of the company to be maximum. Help him invite an optimal company!

Input

The first line of the input contains two space-separated integers, n and d (1 ≤ n ≤ 105, ) — the number of Kefa's friends and the minimum difference between the amount of money in order to feel poor, respectively.

Next n lines contain the descriptions of Kefa's friends, the (i + 1)-th line contains the description of the i-th friend of type mi, si(0 ≤ mi, si ≤ 109) — the amount of money and the friendship factor, respectively.

Output

Print the maximum total friendship factir that can be reached.

Examples

input

4 5
75 5
0 100
150 20
75 1

output

100

input

5 100
0 7
11 32
99 10
46 8
87 54

output

111

Note

In the first sample test the most profitable strategy is to form a company from only the second friend. At all other variants the total degree of friendship will be worse.

In the second sample test we can take all the friends.

2.题意概述：

给你n个元素，每个元素由m[i]和s[i]组成，要你构造一个集合，使得集合内的最大m和最小m相差不超过d，且这个集合的s[i]的和最大，输出最大值。

3.解题思路：

既然最大和最小不超过d，考虑每次以集合下界（即最小m）来找它的上界，并把所有的位于上界内的s[i]求和。

考虑按m[i]排序，这样对于对于每个m[i]只需要向上去搜索不大于m[i] + d的元素，这里可以手写二分搜索也可以用STL函数lower_bound，然后是集合内s[i]和，有单调性以后就不需要线性求和，直接维护一个前缀和即可。

ps：这里结构体貌似不能直接用STL的lower_bound，具体怎么写我也不会QAQ

4.AC代码：

#include <bits/stdc++.h>
#define INF 0x3f3f3f3f
#define maxn 100100
#define lson root << 1
#define rson root << 1 | 1
#define lent (t[root].r - t[root].l + 1)
#define lenl (t[lson].r - t[lson].l + 1)
#define lenr (t[rson].r - t[rson].l + 1)
#define N 1111
#define eps 1e-6
#define pi acos(-1.0)
#define e exp(1.0)
using namespace std;
const int mod = 1e9 + 7;
typedef long long ll;
typedef unsigned long long ull;
struct node
{
	int m, s;
} p[maxn];
ll sum[maxn];
bool cmp(node a, node b)
{
	return a.m < b.m;
}
int pp[maxn];
int main()
{
#ifndef ONLINE_JUDGE
	freopen("in.txt", "r", stdin);
	freopen("out.txt", "w", stdout);
	long _begin_time = clock();
#endif
	int n, d;
	while (~scanf("%d%d", &n, &d))
	{
		for (int i = 1; i <= n; i++)
			scanf("%d%d", &p[i].m, &p[i].s);
		sort(p + 1, p + n + 1, cmp);
		sum[0] = 0;
		for (int i = 1; i <= n; i++)
		{
			pp[i] = p[i].m;
			sum[i] = sum[i - 1] + p[i].s;
		}
		ll ans = 0;
		for (int i = 1; i <= n; i++)
		{
			int pos = lower_bound(pp + 1, pp + n + 1, p[i].m + d) - pp - 1;
			ans = max(ans, sum[pos] - sum[i - 1]);
		}
		printf("%lld\n", ans);
	}
#ifndef ONLINE_JUDGE
	long _end_time = clock();
	printf("time = %ld ms.", _end_time - _begin_time);
#endif
	return 0;
}