bzoj3583 杰杰的女性朋友

如果暴力地把邻接矩阵建出来求解是 $O(n^3\log d)$的，显然无法接受。
我们把矩阵的元素写出来，$M_{i,j}=\sum_{x=1}^k O_{i,x}*I_{j,x}$
也就是$M=O\times I^T$
这样的话，我们要求的$M^d$就可以看成$(OI^T)^d=O(I^TO)^{d-1}I$。中间的矩阵是$k\times k$的，这样复杂度就成了$O(k^3\log d+k^2n)$。
还有一个问题，直接矩阵乘法求出来的是恰好$d$步走到，我们可以新建一个节点，从终点和他自己向他连边用来累计答案，这样求走$d+1$步的解就是答案。

#include<cstdio>
#include<algorithm>
using namespace std;
#define LL unsigned long long
const int maxn=1010,maxk=25,p=1000000007;
int o[maxn][maxk],it[maxk][maxn],o1[maxn][maxk],i1[maxk][maxn],a[maxn][maxn],
t1[maxn][maxn],t2[maxn][maxn];
int n,k;
int main()
{
    //freopen("in.txt","r",stdin);
    //freopen("out.txt","w",stdout);
    int q,u,v,d,ans;
    scanf("%d%d",&n,&k);
    for (int i=1;i<=n;i++)
    {
        for (int j=1;j<=k;j++)
            scanf("%d",&o[i][j]);
        for (int j=1;j<=k;j++)
            scanf("%d",&it[j][i]);
    }
    scanf("%d",&q);
    while (q--)
    {
        scanf("%d%d%d",&u,&v,&d);
        for (int i=1;i<=n;i++)
            for (int j=1;j<=k;j++)
            {
                o1[i][j]=o[i][j];
                i1[j][i]=it[j][i];
            }
        for (int i=1;i<=n;i++)
        {
            o1[i][k+1]=(i==v);
            i1[k+1][i]=0;
        }
        o1[n+1][k+1]=i1[k+1][n+1]=1;
        for (int i=1;i<=k+1;i++)
            for (int j=1;j<=k+1;j++)
            {
                a[i][j]=0;
                for (int x=1;x<=n+1;x++)
                    a[i][j]=(a[i][j]+(LL)i1[i][x]*o1[x][j])%p;
            }
        //mul(i1,o1,k+1,n+1,k+1);

        for (int i=1;i<=k+1;i++)
            for (int j=1;j<=k+1;j++)
            {
                t1[i][j]=a[i][j];
                a[i][j]=(i==j);
            }
        while (d)
        {
            if (d&1)
            {
                for (int i=1;i<=k+1;i++)
                    for (int j=1;j<=k+1;j++)
                    {
                        t2[i][j]=0;
                        for (int x=1;x<=k+1;x++)
                            t2[i][j]=(t2[i][j]+(LL)a[i][x]*t1[x][j])%p;
                    }
                for (int i=1;i<=k+1;i++)
                    for (int j=1;j<=k+1;j++)
                        a[i][j]=t2[i][j];
            }
            d>>=1;
            for (int i=1;i<=k+1;i++)
                for (int j=1;j<=k+1;j++)
                {
                    t2[i][j]=0;
                    for (int x=1;x<=k+1;x++)
                        t2[i][j]=(t2[i][j]+(LL)t1[i][x]*t1[x][j])%p;
                }
            for (int i=1;i<=k+1;i++)
                for (int j=1;j<=k+1;j++)
                    t1[i][j]=t2[i][j];
        }
        //pow(d,k+1);

        for (int i=1;i<=k+1;i++)
            for (int j=1;j<=k+1;j++)
                t1[i][j]=a[i][j];
        for (int i=1;i<=n+1;i++)
            for (int j=1;j<=k+1;j++)
            {
                a[i][j]=0;
                for (int x=1;x<=k+1;x++)
                    a[i][j]=(a[i][j]+(LL)o1[i][x]*t1[x][j])%p;
            }
        //mul(o1,a,n+1,k+1,k+1);

        ans=0;
        for (int i=1;i<=k+1;i++)
            ans=(ans+(LL)a[u][i]*i1[i][n+1])%p;
        printf("%d\n",ans);
        //mul(a,i1,n+1,k+1,n+1);
        //printf("%d\n",a[u][n+1]);
    }
}