【CF908D New Year and Arbitrary Arrangement】

【Analysis】

设$f_{i,j}$表示有$i$个A，$j$个AB期望长度。目标状态是$f_{1,0}$，因为$f_{0,0}$会转移自己。容易得到方程：
$$f_{i,j}=\frac{pa}{pa+pb}\times f_{i+1,j}+\frac{pb}{pa+pb}\times f_{i,j+i}$$

初始转态是$f_{i,j}(i+j⩾k)$，因为再加一个B就完成了。设期望$x$次中一次B，$P_a\leftarrow \frac{pa}{pa+pb}$，$P_b\leftarrow \frac{pb}{pa+pb}$，$c\leftarrow i+j$
$$x\leftarrow P_b\times \sum _{i=0}^{\infty }(c+i)\times P_a^i$$
设数列 { a n } \{a_n\} {an​}满足$a_n=(c+n)\times P_a^n$

设$S_n=\sum _{i=0}^n{a}_i$

$$P_a\times S_n=\frac{pa}{pa+pb}\times \sum _{i=0}^n(c+i)\times \frac{pa}{pa+pb}$$

$$=\sum _{i=0}^n(c+i)\times (\frac{pa}{pa+pb}{)}^{i+1}$$

$$\because P_a+P_b=1$$

$$\therefore x\leftarrow \underset{n\to \infty }{\lim }{P}_b\times S_n=\underset{n\to \infty }{\lim }{S}_n-P_a\times S_n$$
$$=\underset{n\to \infty }{\lim }\sum _{i=0}^n(c+i)\times \frac{pa}{pa+pb}-\sum _{i=0}^n(c+i)\times (\frac{pa}{pa+pb}{)}^{i+1}$$
$$=\underset{n\to \infty }{\lim }(c+0)\times P_a^0+\sum _{i=1}^n(c+i)\times P_a^i-\sum _{i=1}^n(c+i-1)\times P_a^i+(c+n)\times P_a^{n+1}$$
$$=\underset{n\to \infty }{\lim }c+\sum _{i=1}^n{P}_a^i+(c+n)\times P_a^{n+1}$$
$$=\underset{n\to \infty }{\lim }c+\frac{P_a-P_a^n\times P_a}{1-P_a}+(c+n)\times P_a^{n+1}$$
$$=\underset{n\to \infty }{\lim }c+\frac{pa}{pb}\times (1-P_a^n)+(c+n)\times P_a^{n+1}$$
$$=\underset{n\to \infty }{\lim }c+\frac{pa}{pb}+(c+n)\times P_a^{n+1}$$
$$=\underset{n\to \infty }{\lim }c+\frac{pa}{pb}+\frac{c+n}{(\frac{pa+pb}{pa}{)}^{n+1}}$$
$$\because (c+n)随着n的增大单调递增，(\frac{pa+pb}{pa}{)}^{n+1}随着n的增大单调递增$$
$$\therefore 满足洛必达法则$$
$$根据洛必达法则，\frac{f(x)}{g(x)}如果在x=a处变成(0/0)或是(\infty /\infty )，那么\frac{f(x)}{g(x)}=\frac{f^{\prime }(x)}{g^{\prime }(x)}$$
$$\therefore x\leftarrow \underset{n\to \infty }{\lim }c+\frac{pa}{pb}+\frac{c+n}{(\frac{pa+pb}{pa}{)}^{n+1}}$$
$$=\underset{n\to \infty }{\lim }c+\frac{pa}{pb}+\frac{1}{\ln (\frac{pa+pb}{pa})\times (\frac{pa+pb}{pa}{)}^{n+1}}$$
$$=c+\frac{pa}{pb}$$

时间复杂度$\Theta (k^2)$

【Code】

#include <bits/stdc++.h>
using namespace std;
 
typedef long long int64;
const int P = 1e9 + 7;
inline int qPow(int x, int y)
{
	int res = 1, base = x;
	while (y > 0)
	{
		if (y & 1)
			res = 1ll * res * base % P;
		base = 1ll * base * base % P;
		y >>= 1;
	}
	return res;
}
inline int inv(int x) { return qPow(x, P - 2); }
inline int add(int x, int y) { x += y; x -= x >= P ? P : 0; return x; }
inline int mul(int x, int y) { return 1ll * x * y % P; }

int main(void)
{
	int n = 1, a = 1, b = 1;
	cin >> n >> a >> b;
	const int A = 1ll * a * inv(a + b) % P;
	const int B = (1 - A + P) % P;
	const int V = 1ll * a * inv(b) % P;
	vector <vector <int>> dp(n + 1);
	for (int i = 1;i <= n; ++i)
		dp[i].resize(n + 1);
	
	for (int i = n;i >= 1; --i)
		for (int j = n;j >= 0; --j)
			dp[i][j] = (i + j >= n) ? (add(i, add(j, V))) : (add(mul(dp[i + 1][j], A), mul(dp[i][i + j], B)));
	cout << dp[1][0] << '\n';
	return 0;
}