2991: [2012北大校赛]Do Not Gamble

2991: [2012北大校赛]Do Not Gamble

Time Limit: 10 Sec   Memory Limit: 128 MB
Submit: 8   Solved: 3
[ Submit][ Status][ Discuss]

Description

Have you ever gambled? I wish you not. Being a professional gambler for three years, I learned every

 trick they play. Gambling is has nothing to do with luck. It is all about cheating.One of the most 

elementary tricks is "mercury dice". Instead of being solid, there is a hollow chamber, filled with 

mercury, inside of the dice. This little modification makes the dice asymmetric, so that, the probab

ility of each outcome is not equal. Imprecisely, which face is up, depends on the previous state.To 

simplify the model, we use "mercury coin" here. As we mentioned, the state (head or tail) after toss

ing depends on the state before. The probability of that the two states are same, is p. For example,

 if the head side is up now, the probability that, the coin is still head up after tossing is p. And

 the probability that the tail side come up is 1-p. If p=0.5, it is a common coin. When p≠ 0.5, thi

s coin is mercury coin. Assume the coin is head up now. After we toss Ntimes, what is the probabilit

y that, the number of heads is less than a certain number K?

Input

The test file contains multiple lines, and each line is a test case. 

Each test cases consists of three numbers

N K p

N and K are integers, and p is a float number. 

N ≤ 30000

K ≤ N

0 ≤ p ≤ 1

Output

For each test case, output the probability that, after tossing N times, the number of heads is less 

than K. Rounded to the three digit after the decimal point.

Sample Input

171 88 0.107

Sample Output

0.83785

HINT

Source

[ Submit][ Status][ Discuss]
题意： 有一枚奇怪的硬币，每次投它有p的概率和上次同一面，1-p的概率和上次反面，一开始正面朝上，问投掷n次后正面朝上次数小于m次的概率是多少？？ solution： Fi,j：投了i次，正面朝上j次，最后一次正面朝上的概率， Gi,j：投了i次，正面朝上j次，最后一次反面朝上的概率。 这样搞个dp，转移方程显然，但是n是3W的，这样就TLE了-- 尝试重新定义———— Fi,j：投了i次，和第一次朝向相同j次，最后一次和第一次朝向相同 Gi,j：类似 这样就能F2i,j = ∑Fi,t*Fi,j-t + ∑Gi,t*Gi,i-j+t 类似列出G2i,j的式子，把F，G分别翻转，记作Fr,Gr，这样就是卷积的形式了 最后和快速幂的处理类似，，瞎搞一下FFT就行

卡精度了。。GG。。bzoj没有spj，输出要特判。。

#include<iostream>  
#include<cstdio>  
#include<cstring>  
#include<algorithm>  
#include<cmath>  
using namespace std;  
   
const int maxn = 2E5 + 10;  
typedef double DB;  
const DB PI = acos(-1.0);  
   
struct Virt{  
    DB r,i;  
    Virt(){}  
    Virt(DB r,DB i): r(r),i(i){}  
    Virt operator + (const Virt &b) {return Virt(r + b.r,i + b.i);}  
    Virt operator - (const Virt &b) {return Virt(r - b.r,i - b.i);}  
    Virt operator * (const Virt &b) {return Virt(r*b.r - i*b.i,r*b.i + i*b.r);}  
    Virt operator * (const DB &t) {return Virt(r*t,i*t);}  
    Virt operator / (const DB &t) {return Virt(r/t,i/t);}  
}F[maxn],G[maxn],Fr[maxn],Gr[maxn],A[maxn],B[maxn],C[maxn],D[maxn];  
   
int n,m;  
DB p,q; 
bool flag; 
   
void Rader(Virt *a,int len)  
{  
    int j = len >> 1;  
    for (int i = 1; i < len - 1; i++) {  
        if (i < j) swap(a[i],a[j]);  
        int k = len >> 1;  
        while (j >= k) {  
            j -= k;  
            k >>= 1;  
        }  
        j += k;  
    }  
}  
   
void FFT(Virt *a,int len,int on)  
{  
    Rader(a,len);  
    DB T = 2.00*PI*(DB)(on);  
    for (int k = 2; k <= len; k <<= 1) {  
        Virt wn = Virt(cos(T/(DB)(k)),sin(T/(DB)(k)));  
        for (int i = 0; i < len; i += k) {  
            Virt w = Virt(1.00,0.00);  
            for (int j = i; j < i + (k>>1); j++) {  
                Virt u = a[j];  
                Virt t = w*a[j + (k>>1)];  
                a[j] = u + t;  
                a[j + (k>>1)] = u - t;  
                w = w*wn;  
            }  
        }  
    }  
    if (on == -1)  
        for (int i = 0; i < len; i++)  
            a[i] = a[i]/(DB)(len);  
}  
   
void Multi(int len)
{
    if (!flag) {
        for (int i = 0; i <= len; i++)
            A[i] = F[i],B[i] = G[i];
        flag = 1;
        return;
    }
    int N = len << 2;
    FFT(A,N,1); FFT(B,N,1);
    FFT(F,N,1); FFT(G,N,1);
    FFT(Fr,N,1); FFT(Gr,N,1);
    for (int i = 0; i < N; i++) {
        Virt a = A[i],b = B[i];
        A[i] = a*F[i] + b*Gr[i];
        B[i] = a*G[i] + b*Fr[i];
    }
    FFT(A,N,-1); FFT(B,N,-1);
    FFT(F,N,-1); FFT(G,N,-1);
    FFT(Fr,N,-1); FFT(Gr,N,-1);
}
   
void Solve(int y)  
{  
    F[1] = Fr[0] = Virt(p,0.00);  
    G[0] = Gr[1] = Virt(q,0.00);  
    int N = 1;  
    for (; y; y >>= 1) {
        if (y & 1) Multi(N);
        int Now = N << 2;
        FFT(F,Now,1); FFT(Fr,Now,1);  
        FFT(G,Now,1); FFT(Gr,Now,1);
        for (int i = 0; i < Now; i++) {
            C[i] = F[i]*F[i] + G[i]*Gr[i];
            D[i] = F[i]*G[i] + G[i]*Fr[i];
        }
        FFT(C,Now,-1); FFT(D,Now,-1);
        N <<= 1;
        for (int i = 0; i <= N; i++) {  
            Fr[N - i] = F[i] = C[i];  
            Gr[N - i] = G[i] = D[i];  
        }  
        for (int i = N + 1; i < Now; i++) {
            F[i] = Fr[i] = Virt(0.00,0.00);
            G[i] = Gr[i] = Virt(0.00,0.00);
        }
    }
}  
   
void Clear(int len)  
{  
    int N = len << 2;
    for (int i = 0; i < N; i++) {  
        F[i] = Fr[i] = G[i] = Gr[i] = Virt(0.00,0.00);  
        A[i] = B[i] = C[i] = D[i] = Virt(0.00,0.00);  
    }  
}  
   
int main()  
{  
    #ifdef DMC  
        freopen("DMC.txt","r",stdin);  
    #endif  
       
    while (scanf("%d%d%lf",&n,&m,&p) != EOF) {  
        q = 1.00 - p;  
        Solve(n);  
        DB ans = 0;  
        flag = 0;
        for (int i = 0; i < m; i++)  
            ans += A[i].r + B[i].r;  
        if (fabs(ans) < 1e-6) puts("0.00000");
        else printf("%.5f\n",ans);
        Clear(n);  
    }  
    return 0;  
}