THỰC HÀNH

Cities in Alberta tend to be laid out as rectangular grids of blocks. Blocks are labeled with coordinates 0 to R-1 from north to south and 0 to C-1 from west to east.

The quality of living in each particular block has been ranked by a distinct number, called quality rank, between 1 and R*C, where 1 is the best and R*C is the worst.

The city planning department wishes to identify a rectangular set of blocks with dimensions H from north to south and W from west to east, such that the median quality rank among all blocks in the rectangle is the best.

H and W are odd numbers not exceeding R and C respectively. The median quality rank among an odd number of quality ranks is defined to be the quality rank m in the set such that the number of quality ranks better than m equals the number of quality ranks worse than m.

Input Format

The first line of input will contain the four integers R C H W, where R and C represent the total size of the city, and H and W represent the dimensions of the set of blocks.

The next R lines each contain C integers, denoting an array Q where Q[a][b] is the quality rank for the block labeled a from north to south and b from west to east.

Output Format

A single integer, the best (numerically smallest) possible median quality rank of an H by W rectangle of blocks.

Example

Input:
5 5 3 3
5 11 12 16 25
17 18 2 7 10
4 23 20 3 1
24 21 19 14 9
6 22 8 13 15
Output:
9

Limit:

Subtask 1: Assume R and C do not exceed 30.

Subtask 2: Assume R and C do not exceed 100.

Subtask 3: Assume R and C do not exceed 300.

Subtask 4: Assume R and C do not exceed 1 000.