Codeforces Round #289 (Div. 2, ACM ICPC Rules) E. Pretty Song

E. Pretty Song

time limit per test

1 second

memory limit per test

256 megabytes

input

standard input

output

standard output

When Sasha was studying in the seventh grade, he started listening to music a lot. In order to evaluate which songs he likes more, he introduced the notion of the song's prettiness. The title of the song is a word consisting of uppercase Latin letters. The prettiness of the song is the prettiness of its title.

Let's define the simple prettiness of a word as the ratio of the number of vowels in the word to the number of all letters in the word.

Let's define the prettiness of a word as the sum of simple prettiness of all the substrings of the word.

More formally, let's define the function vowel(c) which is equal to 1, if c is a vowel, and to 0 otherwise. Let si be the i-th character of string s, and si..j be the substring of word s, staring at the i-th character and ending at the j-th character (sisi + 1... sj, i ≤ j).

Then the simple prettiness of s is defined by the formula:

The prettiness of s equals

Find the prettiness of the given song title.

We assume that the vowels are I, E, A, O, U, Y.

Input

The input contains a single string s (1 ≤ |s| ≤ 5·105) — the title of the song.

Output

Print the prettiness of the song with the absolute or relative error of at most 10 - 6.

Sample test(s)

input

IEAIAIO

output

28.0000000

input

BYOB

output

5.8333333

input

YISVOWEL

output

17.0500000

Note

In the first sample all letters are vowels. The simple prettiness of each substring is 1. The word of length 7 has 28 substrings. So, theprettiness of the song equals to 28.

分析：

统计时不能盲目的统计，要以单个字符的对应分值累加，这样，分子可以保证全是1

首先对于每个字符，在字符串的子串中出现只有3种形式（如果将“单个字符为一个子串”这种特殊情况

分出来也可以说是4种）：分别是  以该字符开头，以该字符结尾，该字符在子串中间。

该字符的位置为i，字符串从0~len-1。长度为len。

1、以该字符结尾

     1/1+1/2+1/3+1/4+.....+1/(i+1)

2、以该字符开头

     1/1+1/2+1/3+1/4+.....+1/(len-i)

这里，单个字符的情况在1、2中重复了，故应-1去重

3、该字符在中间的情况

下标     0，1，2，.... ，i-1，i，i+1，....，len-1

以i-1开头，结尾>=i+1:    1/3+1/4+....+1/(len-i+1)  ，即sum[len-i+1]-sum[2]

以i-2开头，结尾>=i+1:    1/4+1/5+....+1/(len-i+2)  ，即sum[len-i+2]-sum[3]

................

以0开头，结尾>=i+1:      1/(i+2)+1/(i+3)+....+1/len  ，即sum[len]-sum[i+1]

上式累加得：(sum[len-i+1]+sum[len-i+2]+...+sum[len]) - (sum[2]+sum[3]+...+sum[i+1])

即t[len] - t[len-i] - ( t[i+1] - t[1] )

#include<cstdio>
#include<cstring>
#include<iostream>
#include<algorithm>
#include<cstdlib>
#include<cmath>

using namespace std;

typedef long long ll;

char str[500010];
double sum[500010],t[500010];

void init()
{
    memset(sum,0,sizeof(sum));
    memset(t,0,sizeof(t));
    for(int i=1;i<=500000;i++)
        sum[i] = 1.0/i+sum[i-1];
    for(int i=1;i<=500000;i++)
        t[i] = sum[i]+t[i-1];
}

bool check(char c)
{
    if(c=='A' || c=='E' || c=='I' || c=='O' || c=='U' || c=='Y')
        return true;
    return false;
}

int main()
{
    init();
    while(~scanf("%s",str))
    {
        int len = strlen(str);
        double ans = 0.0;
        for(int i=0;i<len;i++)
        {
            if(check(str[i]))
            {
                ans += sum[i+1]+sum[len-i]-1;
                ans += t[len]-t[len-i]-t[i+1]+t[1];
            }
        }
        printf("%.7lf\n",ans);
    }
    return 0;
}