1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
#pragma GCC optimize ("Os")
#pragma GCC optimize ("O3")
#include <iostream>
#include <algorithm>
#include <vector>
#include <complex>
#include <cmath>
#include <cstring>
//#include <cstdint>

// 確認有 uint32 這個 type
#ifndef uint32_t
#define uint32_t unsigned long int
#endif

namespace FFT  // WARNING!!! do not reveal this namespace
{
    using namespace std; // be careful
    const long double M_PIl = acos(-1.);
    int NumberOfBitsNeeded(int PowerOfTwo) {
        for (int i = 0;; ++i) {
            if (PowerOfTwo & (1 << i)) {
                return i;
            }
        }
    }

    // 參考 morris + 掛長 的 blog 的快速翻轉 bit (MSB->LSB; LSB->MSB)
    inline uint32_t FastReverseBits(uint32_t a, int NumBits) {
        a = ( ( a & 0x55555555U ) << 1 ) | ( ( a & 0xAAAAAAAAU ) >> 1 ) ;
        a = ( ( a & 0x33333333U ) << 2 ) | ( ( a & 0xCCCCCCCCU ) >> 2 ) ;
        a = ( ( a & 0x0F0F0F0FU ) << 4 ) | ( ( a & 0xF0F0F0F0U ) >> 4 ) ;
        a = ( ( a & 0x00FF00FFU ) << 8 ) | ( ( a & 0xFF00FF00U ) >> 8 ) ;
        a = ( ( a & 0x0000FFFFU ) << 16 ) | ( ( a & 0xFFFF0000U ) >> 16 ) ;
        return a >> (32 - NumBits);
    }

    // 嘗試打開 fast-math 優化選項，如果 Judge 機不支援，請記得把這行註解掉
    void FFT(bool, const vector<complex<long double> >&, vector<complex<long double> >&) __attribute__((optimize("fast-math")));
    // FFT 本體， In 是輸入的向量（訊號），Out 是輸出的向量（訊號）
    // 這裏其實不太重要，主要會用得的部分是下方的卷積
    // NOTE:::::::::::: 兩個向量長度必須是 2^k，等長
    void FFT(bool InverseTransform, const vector<complex<long double> >& In, vector<complex<long double> >& Out) {
        // simultaneous data copy and bit-reversal ordering into outputs
        int NumSamples = In.size();
        int NumBits = NumberOfBitsNeeded(NumSamples);
        for (int i = 0; i < NumSamples; ++i) {
            Out[FastReverseBits((uint32_t)i, NumBits)] = In[i];
        }
        // the FFT process
        long double angle_numerator = M_PIl * (InverseTransform ? -2 : 2);
        for (int BlockEnd = 1, BlockSize = 2; BlockSize <= NumSamples; BlockSize <<= 1) {
            long double ndelta_angle = -(angle_numerator / BlockSize);
            long double sin1 = sin(ndelta_angle);
            long double cos1 = cos(ndelta_angle);
            long double sin2 = sin(ndelta_angle*2); // 注意：用合角公式計算有可能放大誤差，這裡改用較保險的做法
            long double cos2 = cos(ndelta_angle*2);
            for (int i = 0; i < NumSamples; i += BlockSize) {
                complex<long double> a1(cos1, sin1), a2(cos2, sin2);
                for (int j = i, n = 0; n < BlockEnd; ++j, ++n) {
                    complex<long double> a0(2 * cos1 * a1.real() - a2.real(), 2 * cos1 * a1.imag() - a2.imag());
                    a2 = a1;
                    a1 = a0;
                    complex<long double> a = a0 * Out[j + BlockEnd];
                    Out[j + BlockEnd] = Out[j] - a;
                    Out[j] += a;
                }
            }
            BlockEnd = BlockSize;
        }
        // normalize if inverse transform
        if (InverseTransform) {
            for (int i = 0; i < NumSamples; ++i) {
                Out[i] /= NumSamples;
            }
        }
    }

    // 卷積，輸入"""等長"""的 a, b 兩向量（長度必須是 2^k），會得到 a * b （a卷積b）的結果
    // 下面的流程 A*B -> a=FFT(A); b=FFT(B) -> c = a dot b -> ANS = IFFT(c)
    template<class T> // a, b 兩向量的輸入型態
        void convolution(const vector<T> &a, vector<T> &ret) {
            int n = a.size();
            vector<complex<long double> > s(n), d(n), y(n);
            for (int i = 0; i < n; ++i) {
                s[i] = complex<long double>((long double)a[i], 0);
            }
            FFT(false, s, d); // d1 = FFT(a)
            for (int i = 0; i < n; ++i) { // y = d1 dot d2
                y[i] = d[i] * d[i];
            }
            FFT(true, y, s);   // ANS = IFFT(y)
            ret.resize(n,0);
            for (int i = 0; i < n; ++i) {
                ret[i] = (s[i].real()+0.5L);
            }
        }
}; // namespace FFT

#include <iostream>
#include <iomanip>
using namespace std;
vector<int>input;
vector<long long>num;
vector<long long>res;

int main(void) {
    int T;
    scanf("%d", &T);
    while(T--) {
        int n;
        scanf("%d",&n);
        input.resize(n);
        for (int i=0; i<n; ++i) {
            int t;
            scanf("%d", &input[i]);
        }
        sort(input.begin(), input.end());
        int d=2; for (; (input[input.size()-1]+1)*2>d; d<<=1);
        num.resize(d,0);
        fill(num.begin(), num.end(), 0);
        for (int i=0; i<input.size(); ++i) ++num[ input[i] ];
        FFT::convolution<long long>(num,res);
        num.clear();
        for (int i=0; i<n; ++i) --res[ input[i]+input[i] ];
        for (int i=0; i<res.size(); ++i) res[i]/=2LL;
        res[0]=0LL;
        for (int i=1; i<res.size(); ++i) res[i]+=res[i-1];
        double cnt = 0;
        for (int i=0; i<n; ++i) {
            cnt += res[res.size()-1]-res[input[i]];
            cnt -= n-1;
            cnt -= (double)i*(n-1-i);
            cnt -= (double)(n-1-i)*(n-2-i)/2;
        }
        double ans = 6.0*cnt/n/(n-1.0)/(n-2.0);
        printf("%.7lf\n", ans);
        input.clear(); num.clear();
    }
    return 0;
}