CodeForces_1353F Decreasing Heights(动态规划)

Decreasing Heights

time limit per test:2.5 seconds
memory limit per test:256 megabytes
Problem Description

You are playing one famous sandbox game with the three-dimensional world. The map of the world can be represented as a matrix of size n×m, where the height of the cell ( i , j ) (i,j) (i,j) is a i , j a_{i,j} ai,j.

You are in the cell ( 1 , 1 ) (1,1) (1,1) right now and want to get in the cell ( n , m ) (n,m) (n,m). You can move only down (from the cell ( i , j ) (i,j) (i,j) to the cell ( i + 1 , j ) (i+1,j) (i+1,j)) or right (from the cell ( i , j ) (i,j) (i,j) to the cell ( i , j + 1 ) (i,j+1) (i,j+1)). There is an additional restriction: if the height of the current cell is x x x then you can move only to the cell with height x + 1 x+1 x+1.

Before the first move you can perform several operations. During one operation, you can decrease the height of any cell by one. I.e. you choose some cell ( i , j ) (i,j) (i,j) and assign (set) a i , j : = a i , j − 1 a_{i,j}:=a_{i,j}−1 ai,j:=ai,j1. Note that you can make heights less than or equal to zero. Also note that you can decrease the height of the cell ( 1 , 1 ) (1,1) (1,1).

Your task is to find the minimum number of operations you have to perform to obtain at least one suitable path from the cell ( 1 , 1 ) (1,1) (1,1) to the cell ( n , m ) (n,m) (n,m). It is guaranteed that the answer exists.

You have to answer t independent test cases.

Input

The first line of the input contains one integer t t t ( 1 ≤ t ≤ 100 1≤t≤100 1t100) — the number of test cases. Then t test cases follow.

The first line of the test case contains two integers n n n and m m m ( 1 ≤ n , m ≤ 100 1≤n,m≤100 1n,m100) — the number of rows and the number of columns in the map of the world. The next n lines contain m integers each, where the j j j-th integer in the i i i-th line is a i , j a_{i,j} ai,j ( 1 ≤ a i , j ≤ 1 0 15 1≤a_{i,j}≤10^{15} 1ai,j1015) — the height of the cell ( i , j ) (i,j) (i,j).

It is guaranteed that the sum of n (as well as the sum of m) over all test cases does not exceed 100 ( ∑ n ≤ 100 ; ∑ m ≤ 100 ∑n≤100;∑m≤100 n100;m100).

Output

For each test case, print the answer — the minimum number of operations you have to perform to obtain at least one suitable path from the cell ( 1 , 1 ) (1,1) (1,1) to the cell ( n , m ) (n,m) (n,m). It is guaranteed that the answer exists.

Sample Input

5
3 4
1 2 3 4
5 6 7 8
9 10 11 12
5 5
2 5 4 8 3
9 10 11 5 1
12 8 4 2 5
2 2 5 4 1
6 8 2 4 2
2 2
100 10
10 1
1 2
123456789876543 987654321234567
1 1
42

Sample Output

9
49
111
864197531358023
0

题意

一个三维地图,大小为n*m,第i行第j列的高度为 a i , j a_{i,j} ai,j。初始在 ( 1 , 1 ) (1,1) (1,1),目标为 ( n , m ) (n,m) (n,m)。每次移动只能向下方或右方移动,同时处于高度为x的位置时,下一步只能到高度为x+1的位置。
可以进行多次操作,每次操作可以将某个位置的高度减一。求最少进行多少次操作使存在路径到达 ( n , m ) (n,m) (n,m)

题解

对于任意一条路径,其经过的路径高度依次为 b 0 , b 1 , . . . b n + m − 2 b_0,b_1,...b_{n+m-2} b0,b1,...bn+m2,使高度减一的操作最少,则起始高度应为 m i n ( b 0 − 0 , b 1 − 1 , . . . , b n + m − 2 − n − m − 2 ) min(b_0-0,b_1-1,...,b_{n+m-2}-n-m-2) min(b00,b11,...,bn+m2nm2)
为了方便表述,设地图坐标从 ( 0 , 0 ) (0,0) (0,0)开始。位置 ( i , j ) (i,j) (i,j)只能对应上面的 b i + j b_{i+j} bi+j的位置,则起始高度的可能值为 a [ i ] [ j ] − i − j a[i][j]-i-j a[i][j]ij。起始高度共有n*m个可能值,枚举这些值,动态规划求解到 ( n − 1 , m − 1 ) (n-1,m-1) (n1,m1)需要进行的最少的操作次数。
设起始高度为ps,则递推式为:
d p [ i ] [ j ] = { L L I N F a [ i ] [ j ] < p s + i + j m i n ( L L I N F , m i n ( d p [ i − 1 ] [ j ] , d p [ i ] [ j − 1 ] ) + a [ i ] [ j ] − p s − i − j ) a [ i ] [ j ] ≥ p s + i + j dp[i][j] = \begin{cases}LLINF& a[i][j]<ps+i+j\\min(LLINF, min(dp[i-1][j],dp[i][j-1])+a[i][j]-ps-i-j) & a[i][j] \ge ps+i+j\end{cases} dp[i][j]={LLINFmin(LLINF,min(dp[i1][j],dp[i][j1])+a[i][j]psij)a[i][j]<ps+i+ja[i][j]ps+i+j

#include<stdio.h>
#include<iostream>
#include<cstdlib>
#include<cmath>
#include<algorithm>
#include<cstring>
#include<map>
#include<vector>
#include<queue>
#include<iterator>
#define dbg(x) cout<<#x<<" = "<<x<<endl;
#define INF 0x3f3f3f3f
#define eps 1e-6
 
using namespace std;
typedef long long LL;   
typedef pair<int, int> P;
const int maxn = 120;
const int mod = 1000000007;
const LL llinf = 1e18;
LL a[maxn][maxn], dp[maxn][maxn];

int main()
{
    int t, n, m, i, j, k;
    scanf("%d", &t);
    while(t--)
    {
        scanf("%d %d", &n, &m);
        for(i=0;i<n;i++)
            for(j=0;j<m;j++)
                scanf("%I64d", &a[i][j]);
        LL ans = llinf;
        for(int a1=0;a1<n;a1++)
            for(int a2=0;a2<m;a2++)
        if(a[0][0] >= a[a1][a2]-a1-a2)
        {
            LL ps = a[a1][a2]-a1-a2;
            dp[0][0] = a[0][0]-ps;
            for(j=1;j<n;j++)
                if(a[j][0]<ps+j || dp[j-1][0] == llinf)dp[j][0] = llinf;
                else dp[j][0] = min(llinf, dp[j-1][0] + a[j][0]-ps-j);
            for(j=1;j<m;j++)
                if(a[0][j]<ps+j || dp[0][j-1] == llinf)dp[0][j] = llinf;
                else dp[0][j] = min(llinf, dp[0][j-1] + a[0][j]-ps-j);
            for(j=1;j<n;j++)
                for(k=1;k<m;k++)
                {
                    dp[j][k] = llinf;
                    if(a[j][k]<ps+j+k)continue;
                    dp[j][k] = min(dp[j][k], dp[j-1][k]+a[j][k]-ps-j-k);
                    dp[j][k] = min(dp[j][k], dp[j][k-1]+a[j][k]-ps-j-k);
                }
            ans = min(ans, dp[n-1][m-1]);
        }
        printf("%I64d\n", ans);
    }
    return 0;
}
  • 1
    点赞
  • 1
    收藏
    觉得还不错? 一键收藏
  • 0
    评论

“相关推荐”对你有帮助么?

  • 非常没帮助
  • 没帮助
  • 一般
  • 有帮助
  • 非常有帮助
提交
评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值