python/TangDou/AcWing_TiGao/T5/JuZhengChengFa/P3390_1.cpp

#include <bits/stdc++.h>
using namespace std;
#define int long long
#define endl "\n"
const int MOD = 1e9 + 7;
const int N = 110;
int n, k;             // 本题数据范围很大，用int直接wa哭了
int a[N][N], b[N][N]; // 原始矩阵，结果矩阵

// 矩阵乘法模板
void mul(int a[][N], int b[][N], int c[][N]) {
    int t[N][N] = {0};
    for (int i = 0; i < N; i++)
        for (int j = 0; j < N; j++)
            for (int k = 0; k < N; k++)
                t[i][j] = (t[i][j] + a[i][k] * b[k][j] % MOD) % MOD; // 矩阵乘法
    memcpy(c, t, sizeof t);
}

signed main() {
    cin >> n >> k;
    // 输入原始矩阵
    for (int i = 0; i < n; i++)
        for (int j = 0; j < n; j++)
            cin >> a[i][j];

    // 1、单位矩阵
    for (int i = 0; i < n; i++) b[i][i] = 1;

    // 2、矩阵快速幂
    for (int i = k; i; i >>= 1) {
        if (i & 1) mul(a, b, b);
        mul(a, a, a);
    }

    // 3、输出
    for (int i = 0; i < n; i++) {
        for (int j = 0; j < n; j++)
            printf("%lld ", b[i][j]);
        printf("\n");
    }
}