QOJ.ac

QOJ

Time Limit: 1.5 s Memory Limit: 256 MB Total points: 100

# 11565. Manhattan Pairing

الإحصائيات

For a pair of points on the Cartesian plane $(x_1, y_1)$ and $(x_2, y_2)$, we define the Manhattan distance between them as $|x_1-x_2|+|y_1-y_2|$. For example, for the pair of points $(4, 1)$ and $(2, 7)$, the Manhattan distance between them is $|4-2|+|1-7| = 2+6 = 8$.

You are given $2 \cdot n$ points on the Cartesian plane, whose coordinates are integers. All $y$-coordinates of the given points are either $0$ or $1$.

Split the given points into $n$ pairs such that each of these points belongs to exactly one pair, and the maximum Manhattan distance between the points of one pair is minimized.

Input

The first line contains a single integer $n$ $(1 \le n \le 10^5)$.

In the following $2 \cdot n$ lines, each line contains two integers $x_i$ and $y_i$ $(0 \le x_i \le 10^9, 0 \le y_i \le 1)$~--- the coordinates of the corresponding point.

Output

Output a single integer~--- the maximum Manhattan distance between the points of one pair in the optimal partition.

Example

Input

1
3 1
1 0

Output

3

Input

3
18 0
3 0
1 0
10 0
8 0
14 0

Output

4

Input

4
3 0
0 1
5 0
2 1
6 0
3 0
5 1
2 1

Output

2

Notes

In the second example, the pairing $[(18, 0), (14, 0)]$, $[(3, 0), (1, 0)]$, and $[(8, 0), (10, 0)]$ is the only optimal partition. The Manhattan distances between the points of one pair in this partition are $4$, $2$, and $2$, respectively.

In the third example, the pairing $[(0, 1), (2, 1)]$, $[(2, 1), (3, 0)]$, $[(3, 0), (5, 0)]$, and $[(5, 1), (6, 0)]$ is an optimal partition. All Manhattan distances between the points of one pair in this partition are equal to $2$.

problem_11565_1.png
Illustration for the third example

Scoring

  1. ($2$ points): $n = 1$;
  2. ($3$ points): $x_i = 0$ for $1 \le i \le 2\cdot n$;
  3. ($4$ points): $n \le 4$;
  4. ($11$ points): $n \le 10$;
  5. ($14$ points): $y_i = 0$ for $1 \le i \le 2\cdot n$;
  6. ($10$ points): $x_i \neq x_j$ for $1 \le i < j \le 2\cdot n$;
  7. ($29$ points): $n \le 1000$;
  8. ($27$ points): no additional restrictions.

对于笛卡尔平面上的一对点 $(x_1, y_1)$ 和 $(x_2, y_2)$,我们定义它们之间的 曼哈顿距离 为 $|x_1-x_2|+|y_1-y_2|$。例如,对于点对 $(4, 1)$ 和 $(2, 7)$,它们之间的 曼哈顿距离 为 $|4-2|+|1-7| = 2+6 = 8$。

你将得到平面上 $2 \cdot n$ 个点,它们的坐标是整数。所有给出的点的 $y$ 坐标要么是 $0$,要么是 $1$。

请将这些点划分成 $n$ 对,使得每个点恰好属于一对,并且使得每对点之间的最大 曼哈顿距离 尽可能小。

输入

第一行包含一个整数 $n$ $(1 \le n \le 10^5)$。

接下来的 $2 \cdot n$ 行中,每行包含两个整数 $x_i$ 和 $y_i$ $(0 \le x_i \le 10^9, 0 \le y_i \le 1)$~--- 表示该点的坐标。

输出

输出一个整数~--- 最优划分中每对点之间的最大 曼哈顿距离

示例

输入

1
3 1
1 0

输出

3

输入

3
18 0
3 0
1 0
10 0
8 0
14 0

输出

4

输入

4
3 0
0 1
5 0
2 1
6 0
3 0
5 1
2 1

输出

2

说明

在第二个样例中,划分为 $[(18, 0), (14, 0)]$, $[(3, 0), (1, 0)]$, 和 $[(8, 0), (10, 0)]$ 是唯一的最优划分。在这个划分中,每对点之间的 曼哈顿距离 分别为 $4$, $2$, 和 $2$。

在第三个样例中,划分为 $[(0, 1), (2, 1)]$, $[(2, 1), (3, 0)]$, $[(3, 0), (5, 0)]$, 和 $[(5, 1), (6, 0)]$ 是一个最优划分。这个划分中每对点之间的 曼哈顿距离 都为 $2$。

problem_11565_1.png
第三个样例的图示

评分

  1. ($2$ 分):$n = 1$;
  2. ($3$ 分):$x_i = 0$ 对于 $1 \le i \le 2\cdot n$;
  3. ($4$ 分):$n \le 4$;
  4. ($11$ 分):$n \le 10$;
  5. ($14$ 分):$y_i = 0$ 对于 $1 \le i \le 2\cdot n$;
  6. ($10$ 分):$x_i \neq x_j$ 对于 $1 \le i < j \le 2\cdot n$;
  7. ($29$ 分):$n \le 1000$;
  8. ($27$ 分):没有额外限制。