MG loves string
Time Limit: 2000/1000 MS (Java/Others) Memory Limit: 262144/262144 K (Java/Others)
Total Submission(s): 131 Accepted Submission(s): 50
Problem Description
MG is a busy boy. And today he's burying himself in such a problem:
For a length of N , a random string made of lowercase letters, every time when it transforms, all the character i will turn into a[i] .
MG states that the a[i] consists of a permutation .
Now MG wants to know the expected steps the random string transforms to its own.
It's obvious that the expected steps X will be a decimal number.
You should output X∗26Nmod 1000000007 .
For a length of N , a random string made of lowercase letters, every time when it transforms, all the character i will turn into a[i] .
MG states that the a[i] consists of a permutation .
Now MG wants to know the expected steps the random string transforms to its own.
It's obvious that the expected steps X will be a decimal number.
You should output X∗26Nmod 1000000007 .
Input
The first line is an integer
T which indicates the case number.(1<=T<=10 )
And as for each case, there are 1 integer in the first line which indicate the length of random string(1<=N<=1000000000 ).
Then there are 26 lowercase letters a[i] in the next line.
And as for each case, there are 1 integer in the first line which indicate the length of random string(1<=N<=1000000000 ).
Then there are 26 lowercase letters a[i] in the next line.
Output
As for each case, you need to output a single line.
It's obvious that the expected steps X will be a decimal number.
You should output X∗26Nmod 1000000007 .
It's obvious that the expected steps X will be a decimal number.
You should output X∗26Nmod 1000000007 .
Sample Input
2
2
abcdefghijklmnpqrstuvwxyzo
1
abcdefghijklmnopqrstuvwxyz
Sample Output
5956
26
题意:给定26个小写字母x1,x2,...,x26的字符串作为“密码表”,26个密码分别对应a~z26个小写字母,一个字母进行一次变换,意味着该字母变换成对应的密码,譬如字母b下一次变换应该变成x2,字母x2会变换成x2对应的密码等等(可以知道,经过有限次的变换,每个字母最终还是会变换回来的)。
现在给定一个随机字符串的长度N,并且字符串是由任意的小写字母组成,且该字符串经过有限次变换还是会变换到自身,求任意字符串的变换期望次数。
思路:求期望次数,首先想到的是穷举每一种长度为N的小写字符串(一共26^N种可能),对于每次得到的字符串,求出变换到自身的变换次数,之后求平均。当然这样的效率是极其低下的。每个字母经过有限次变换都是可以变换到自身的,将这个过程中所有经过变换得到的字母组成一个圈,那么这个圈中每一个字母都会变换相同的次数之后变换到自身),这样一来,所有的26个字母就可以被挑拣成几个不同的圈,情况最坏的时候每个字母分别成为一个圈,一共26个圈,有点多,现在继续考虑能否将长度一样的圈挑出来成为一个group,那么这样一共会有多少group呢,最坏的情况,圈的长度从1开始变化,1+2+3+4+5+6+7=28>26,也就是说绝对不可能产生七个长度不同的group,最多6个,这么一来范围就小多了。
现在考虑长度相同的每一个group的性质,可以知道每一个group中的每个字母变换到自身的次数都相同。
那么现在长度为N的随机字符串的每一个位置,都可以由用每一个group中任意一个字符来填充。穷举group的组合情况,对于每一种group的组合情况,随机字符串中的每一位都由这些group中的字符来填充(因为每一个group都至少得派出一个字符来填充随机字符串,所以每一种group的组合情况中group的数量必须小于等于随机串长度N,否则直接排除在这种group的组合),那么由这些group中的字符组成随机串对期望变换次数的贡献=这些group构成的字符串的变换次数change_time*这些group构成的字符串的数量number/(26^N);
其中,change_time=组合中每一个group长度的最小公倍。
p=number/(26^N)则可以由容斥原理得到,举个简单的例子,设Pabc=长度为n的随机串用group a,b,c或其一或其二中字母组成的概率,显然Pabc=((size(a)+size(b)+size(c))/26)^N.那么随机串由group a,b,c组成的概率P=Pabc-Pab-Pac-Pbc+pa+pb+pc;
穷举group的组合情况并累加贡献即可得到最终期望。
AC代码:
#define _CRT_SECURE_NO_DEPRECATE #include<iostream> #include<algorithm> #include<string> #include<cmath> #include<queue> #include<set> #include<map> #include<cstring> #include<vector> using namespace std; typedef long long ll; const int N_MAX = 1000000000 + 2,INF= 1000000007; int N; string s; vector<pair<int,int> >loop;//圈的大小<->圈的个数 ll gcd(ll a,ll b) { if (b == 0)return a; return gcd(b, a%b); } ll lcm(ll a,ll b) { return a / gcd(a, b)*b; } void cal_loop() { loop.clear(); int change[26],vis[26]; memset(vis,0,sizeof(vis)); for (int i = 0; i < s.size();i++) { change[i] = s[i] - 'a'; } map<int, int>m; for (int i = 0; i < 26;i++) { if (vis[i])continue; vis[i] = true; int num = 1; int k = change[i];//k代表不断变化的字母 while (i != k) { vis[k] = true; num++;//该圈的元素个数加1 k = change[k];//!!!!!顺序 } m[num]++; } for (map<int, int>::iterator it = m.begin(); it != m.end();it++) { loop.push_back(*it); } } ll mod_pow(ll x,ll n) {//快速幂运算 ll res = 1; while (n>0) { if (n & 1)res = res*x%INF; x = x*x%INF; n >>= 1; } return res; } ll R_C(vector<int>&loop,int N) {//容斥原理,求由这些圈中的元素组成的长度为N的字符串的数量 ll permute = 1 << (loop.size()); ll ans = 0; for (int i = 0; i < permute;i++) { int num = 0; int sum = 0; int sign=-1; for (int j = 0; j < loop.size(); j++) { if (i&(1 << j)) { num++;//num记录利用到的圈的个数 sum += loop[j];//利用到的字符的总个数 } } if (num % 2 == loop.size() % 2)//!!!!!!!! sign = 1; ans =(ans+((sign*mod_pow(sum, N))%INF+INF)%INF)%INF; } return ans; } ll solve(int N) { cal_loop(); vector<int>vec; ll ans = 0; for (int i = 0; i < (1<<loop.size());i++) { ll change_time=1; vec.clear(); for (int j = 0; j < loop.size();j++) { if (i&(1 << j)) { vec.push_back(loop[j].first*loop[j].second); change_time = lcm(change_time, loop[j].first); } } if (vec.size() > N)continue;//挑选出来的圈的个数不能超过字符串长度 ll number = R_C(vec, N); ans = (ans + change_time*number) % INF; } return ans; } int main() { int T; scanf("%d",&T); while (T--) { scanf("%d",&N); cin >> s; printf("%lld\n",solve(N)); } return 0; }