Partial Sums（CF-223B）

Problem Description

You've got an array a, consisting of n integers. The array elements are indexed from 1 to n. Let's determine a two step operation like that:

First we build by the array a an array s of partial sums, consisting of n elements. Element number i (1 ≤ i ≤ n) of array s equals . The operation x mod y means that we take the remainder of the division of number x by number y.
Then we write the contents of the array s to the array a. Element number i (1 ≤ i ≤ n) of the array s becomes the i-th element of the array a (ai = si).
You task is to find array a after exactly k described operations are applied.

Input

The first line contains two space-separated integers n and k (1 ≤ n ≤ 2000, 0 ≤ k ≤ 109). The next line contains n space-separated integers a1, a2, ..., an — elements of the array a (0 ≤ ai ≤ 109).

Output

Print n integers  — elements of the array a after the operations are applied to it. Print the elements in the order of increasing of their indexes in the array a. Separate the printed numbers by spaces.

Examples

Input

3 1
1 2 3

Output

1 3 6

Input

5 0
3 14 15 92 6

Output

3 14 15 92 6

题意：给出 n 个数，进行 k 次操作，每次操作将数组用其前缀数组代替，然后模 1E9+7，问 k 次操作后每个数的值

思路：矩阵快速幂

假设数组长度为 3，每一项为 a1、a2、a3

则有操作矩阵：

，其中，每一列的和为第 i 次操作的第 j 项

可以看出，题目实质是要求对矩阵  做 k 次乘法，使用矩阵快速幂即可解决

但由于裸的矩阵快速幂时间复杂度为 O(n*n*n*logk)，无疑会 TLE，需要利用矩阵本身的性质来进行压缩，以降低时间复杂度

观看可以发现，每条与主对角线平行的斜线上的值是相等的，因此对于整个矩阵来说，真正有意义的值只是最上面的一行，故而可以将矩阵压缩成一个 1*n 的矩阵，然后对其进行矩阵快速幂

Source Program

#include<iostream>
#include<cstdio>
#include<cstdlib>
#include<string>
#include<cstring>
#include<cmath>
#include<ctime>
#include<algorithm>
#include<utility>
#include<stack>
#include<queue>
#include<vector>
#include<set>
#include<map>
#include<bitset>
#define EPS 1e-9
#define PI acos(-1.0)
#define INF 0x3f3f3f3f
#define LL long long
const int MOD = 1E9+7;
const int N = 5000+5;
const int dx[] = {-1,1,0,0,-1,-1,1,1};
const int dy[] = {0,0,-1,1,-1,1,-1,1};
using namespace std;
struct Matrix{
    LL s[N];//1*n的矩阵
};
Matrix e;//单位矩阵E
Matrix x;//构造矩阵x
LL a[N];
Matrix mul(Matrix A,Matrix B,LL n){//矩阵乘法
    Matrix temp;//临时矩阵，存放矩阵A*矩阵B的结果
    for(int i=1;i<=n;i++)
        temp.s[i]=0;

    for(int i=1;i<=n;i++)
        for(int j=1;j<=i;j++)
            temp.s[i]=(temp.s[i]+(A.s[j]*B.s[i-j+1])%MOD)%MOD;

    return temp;
}

Matrix quickPower(Matrix A,LL k,LL n){//矩阵快速幂
    Matrix res=e;
    while(k){
        if(k&1)
            res=mul(A,res,n);//res=A*e
        A=mul(A,A,n);//A=A*A
        k>>=1;
    }
    return res;
}

int main(){
    LL n,k;
    scanf("%lld%lld",&n,&k);

    for(int i=1;i<=n;i++)
        scanf("%d",&a[i]);

    //单位矩阵的第一行
    for(int i=1;i<=n;i++)
        e.s[i]=0;
    e.s[1]=1;
    //构造矩阵初始值
    for(int i=1;i<=n;i++)
        x.s[i]=1;

    Matrix res=quickPower(x,k,n);//k次幂

    for(int i=1;i<=n;i++){
        LL val=0;
        for(int j=1;j<=i;j++)
            val=(val+a[i-j+1]*res.s[j])%MOD;
        printf("%lld ",val);
    }
    printf("\n");

    return 0;
}