题目
设
p
=
1
m
p = \frac 1m
p=m1
a
n
s
=
∑
i
=
0
k
{
k
i
}
p
i
(
n
i
)
i
!
=
∑
i
=
0
k
∑
j
=
0
i
(
i
j
)
(
n
i
)
p
i
(
−
1
)
j
(
i
−
j
)
k
=
∑
j
=
0
k
(
−
1
)
j
(
n
j
)
∑
i
=
j
k
(
n
−
j
i
−
j
)
p
i
(
i
−
j
)
k
=
∑
j
=
0
k
(
−
1
)
j
p
j
(
n
j
)
∑
i
=
0
k
−
j
(
n
−
j
i
)
p
i
i
k
=
∑
i
=
0
k
i
k
p
i
∑
j
=
0
k
−
i
(
−
p
)
j
(
n
j
)
(
n
−
j
i
)
=
∑
i
=
0
k
i
k
p
i
(
n
i
)
∑
j
=
0
k
−
i
(
−
p
)
j
(
n
−
i
j
)
\begin{aligned} ans &= \sum_{i=0}^k \begin{Bmatrix}k\\i\end{Bmatrix}p^i\binom nii! \\ &=\sum_{i=0}^k \sum_{j=0}^i \binom ij\binom nip^i(-1)^j(i-j)^k\\ &=\sum_{j=0}^k(-1)^j\binom nj \sum_{i=j}^k \binom{n-j}{i-j}p^i(i-j)^k\\ &=\sum_{j=0}^k(-1)^jp^j\binom nj\sum_{i=0}^{k-j} \binom {n-j}ip^ii^k\\ &=\sum_{i=0}^ki^kp^i\sum_{j=0}^{k-i} (-p)^j\binom nj\binom{n-j}i\\ &=\sum_{i=0}^ki^kp^i\binom ni\sum_{j=0}^{k-i}(-p)^j\binom{n-i}{j}\end{aligned}
ans=i=0∑k{ki}pi(in)i!=i=0∑kj=0∑i(ji)(in)pi(−1)j(i−j)k=j=0∑k(−1)j(jn)i=j∑k(i−jn−j)pi(i−j)k=j=0∑k(−1)jpj(jn)i=0∑k−j(in−j)piik=i=0∑kikpij=0∑k−i(−p)j(jn)(in−j)=i=0∑kikpi(in)j=0∑k−i(−p)j(jn−i)
考虑 S i + 1 = ∑ j = 0 k − i − 1 ( − p ) j ( n − i − 1 j ) = ∑ j = 0 k − i − 1 ( − p ) j ( ( n − i j ) − ( n − i − 1 j − 1 ) ) = S i − ( − p ) k ( n − i k − i ) + p ( S i + 1 − ( − p ) k − i − 1 ( n − i − 1 k − i − 1 ) ) S_{i+1} = \sum_{j=0}^{k-i-1}(-p)^j\binom{n-i-1}j = \sum_{j=0}^{k-i-1}(-p)^j\left(\binom{n-i}j - \binom{n-i-1}{j-1}\right)\\=S_i-(-p)^k\binom {n-i}{k-i} + p\left(S_{i+1}-(-p)^{k-i-1}\binom {n-i-1}{k-i-1}\right) Si+1=j=0∑k−i−1(−p)j(jn−i−1)=j=0∑k−i−1(−p)j((jn−i)−(j−1n−i−1))=Si−(−p)k(k−in−i)+p(Si+1−(−p)k−i−1(k−i−1n−i−1))
可以得到
S
i
=
S
i
+
1
(
1
−
p
)
+
(
−
p
)
k
−
i
(
n
−
i
−
1
k
−
i
)
S_i = S_{i+1}(1-p) + (-p)^{k-i}\binom {n-i-1}{k-i}
Si=Si+1(1−p)+(−p)k−i(k−in−i−1)
边界条件
S
k
=
1
S_k = 1
Sk=1。
所以
a
n
s
=
∑
i
=
0
k
i
k
p
i
(
n
i
)
S
i
ans = \sum_{i=0}^k i^kp^i\binom ni S_i
ans=i=0∑kikpi(in)Si
O
(
k
)
O(k)
O(k)线性筛预处理
i
k
i^k
ik。
对于
(
n
i
)
\binom ni
(in)需要求
n
i
‾
n^{\underline i}
ni
对于
(
n
−
i
−
1
k
−
i
)
\binom {n-i-1}{k-i}
(k−in−i−1)需要求
(
n
−
k
)
k
−
i
‾
(n-k)^{\overline {k-i}}
(n−k)k−i
还有就是阶乘的逆元。
感觉可以搞搞模数搞拓展
l
u
c
a
s
lucas
lucas。
总复杂度
O
(
k
)
O(k)
O(k)
A
C
C
o
d
e
\mathcal AC \ Code
AC Code
#include<bits/stdc++.h>
#define maxn 10000007
#define rep(i,j,k) for(int i=(j),LIM=(k);i<=LIM;i++)
#define per(i,j,k) for(int i=(j),LIM=(k);i>=LIM;i--)
#define mod 998244353
using namespace std;
int Pow(int b,int k){ int r=1;for(;k;k>>=1,b=1ll*b*b%mod) if(k&1)r=1ll*r*b%mod;return r; }
int pr[maxn/8],cnt_pr,pwk[maxn],inv[maxn],s[maxn];
bool vis[maxn];
int n,m,K,p;
int main(){
scanf("%d%d%d",&n,&m,&K);p = Pow(m , mod-2);
pwk[1] = inv[0] = inv[1] = 1;
rep(i,2,K){
inv[i] = 1ll * (mod - mod / i) * inv[mod % i] % mod;
if(!vis[i]) pr[cnt_pr++] = i , pwk[i] = Pow(i,K);
for(int j=0;pr[j] * i <= K;j++){
vis[i * pr[j]] = 1 , pwk[i * pr[j]] = 1ll * pwk[i] * pwk[pr[j]] % mod;
if(i % pr[j] == 0) break;
}
}
int C = 1 , pw = 1;
rep(i,0,K){
s[K-i] = (s[K-i+1] * (1ll - p) + 1ll * pw * C) % mod;
pw = 1ll * pw * (-p) % mod;
C = 1ll * C * (n-K+i) % mod * inv[i+1] % mod;
}
C = 1 , pw = 1;
int ans = 0;
rep(i,0,K){
ans = (ans + 1ll * pwk[i] * pw % mod * C % mod * s[i]) % mod;
pw = 1ll * pw * p % mod;
C = 1ll * C * (n-i) % mod * inv[i+1] % mod;
}
printf("%d\n",(ans+mod)%mod);
}
94
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
