How to know that a triangle triple exists in our a

Question

I was stuck in solving the following interview practice question:
I have to write a function:

int triangle(int[] A);

that given a zero-indexed array A consisting of N integers returns 1 if there exists a triple (P, Q, R) such that 0 < P < Q < R < N.

A[P] + A[Q] > A[R],  
A[Q] + A[R] > A[P],  
A[R] + A[P] > A[Q].

The function should return 0 if such triple does not exist. Assume that 0 < N < 100,000. Assume that each element of the array is an integer in range [-1,000,000..1,000,000].

For example, given array A such that

A[0]=10, A[1]=2, A[2]=5, A[3]=1, A[4]=8, A[5]=20

the function should return 1, because the triple (0, 2, 4) fulfills all of the required conditions.

For array A such that

A[0]=10, A[1]=50, A[2]=5, A[3]=1

the function should return 0.

If I do a triple loop, this would be very very slow (complexity: O(n^3)). I am thinking maybe to use to store an extra copy of the array and sort it, and use a binary search for a particular number. But I don't know how to break down this problem.
Any ideas?

Answer 1

I have got another solution to count triangles. Its time complexity is O(N**3) and takes long time to process long arrays.

Private Function solution(A As Integer()) As Integer
    ' write your code in VB.NET 4.0
    Dim result, size, i, j, k As Integer
        size = A.Length
        Array.Sort(A)
        For i = 0 To Size - 3
            j = i + 1
            While j < size
                k = j + 1
                While k < size
                    If A(i) + A(j) > A(k) Then
                        result += 1
                    End If
                    k += 1
                End While
                j += 1
            End While
        Next
        Return result
End Function

Answer 2

Here's my solution in C#, which gets 90% correctness (the wrong answer is returned for test extreme_arith_overflow1 -overflow test, 3 MAXINTs-) and 100% performance.

private struct Pair
{
    public int Value;
    public int OriginalIndex;
}

private static bool IsTriplet(Pair a, Pair b, Pair c)
{
    if (a.OriginalIndex < b.OriginalIndex && b.OriginalIndex < c.OriginalIndex)
    {
        return ((long)(a.Value + b.Value) > c.Value
                && (long)(b.Value + c.Value) > a.Value
                && (long)(c.Value + a.Value) > b.Value);
    }
    else if (b.OriginalIndex < c.OriginalIndex && c.OriginalIndex < a.OriginalIndex)
    {
        return ((long)(b.Value + c.Value) > a.Value
                    &&(long)(c.Value + a.Value) > b.Value
                    && (long)(a.Value + b.Value) > c.Value);
    }
    // c.OriginalIndex < a.OriginalIndex && a.OriginalIndex < b.OriginalIndex
    return ((long)(c.Value + a.Value) > b.Value
                && (long)(a.Value + b.Value) > c.Value
                && (long)(b.Value + c.Value) > a.Value);
}

public static int Triplets(int[] A)
{
    const int IS_TRIPLET = 1;
    const int IS_NOT_TRIPLET = 0;

    Pair[] copy = new Pair[A.Length];

    // O (N)
    for (var i = 0; i < copy.Length; i++)
    {
        copy[i].Value = A[i];
        copy[i].OriginalIndex = i;
    }

    // O (N log N)
    Array.Sort(copy, (a, b) => a.Value > b.Value ? 1 : (a.Value == b.Value) ? 0 : -1);

    // O (N)
    for (var i = 0; i < copy.Length - 2; i++)
    {
        if (IsTriplet(copy[i], copy[i + 1], copy[i + 2]))
        {
            return IS_TRIPLET;
        }
    }

    return IS_NOT_TRIPLET;
}

Answer 3

In Java:

public int triangle2(int[] A) {

    if (null == A)
        return 0;
    if (A.length < 3)
        return 0;

    Arrays.sort(A);

    for (int i = 0; i < A.length - 2 && A[i] > 0; i++) {
        if (A[i] + A[i + 1] > A[i + 2])
            return 1;
    }

    return 0;

}

Answer 4

First of all, you can sort your sequence. For the sorted sequence it's enough to check that A[i] + A[j] > A[k] for i < j < k, because A[i] + A[k] > A[k] > A[j] etc., so the other 2 inequalities are automatically true.

(From now on, i < j < k.)

Next, it's enough to check that A[i] + A[j] > A[j+1], because other A[k] are even bigger (so if the inequality holds for some k, it holds for k = j + 1 as well).

Next, it's enough to check that A[j-1] + A[j] > A[j+1], because other A[i] are even smaller (so if inequality holds for some i, it holds for i = j - 1 as well).

So, you have just a linear check: you need to check whether for at least one j A[j-1] + A[j] > A[j+1] holds true.

Altogether O(N log N) {sorting} + O(N) {check} = O(N log N).

---

Addressing the comment about negative numbers: indeed, this is what I didn't consider in the original solution. Considering the negative numbers doesn't change the solution much, since no negative number can be a part of triangle triple. Indeed, if A[i], A[j] and A[k] form a triangle triple, then A[i] + A[j] > A[k], A[i] + A[k] > A[j], which implies 2 * A[i] + A[j] + A[k] > A[k] + A[j], hence 2 * A[i] > 0, so A[i] > 0 and by symmetry A[j] > 0, A[k] > 0.

This means that we can safely remove negative numbers and zeroes from the sequence, which is done in O(log n) after sorting.

Answer 5

public static int solution(int[] A) {
    // write your code in Java SE 8
    long p, q, r;
    int isTriangle = 0;

    Arrays.sort(A);
    for (int i = 0; i < A.length; i += 1) {

        if (i + 2 < A.length) {
            p = A[i];
            q = A[i + 1];
            r = A[i + 2];
            if (p >= 0) {
                if (Math.abs(p) + Math.abs(q) > Math.abs(r) && Math.abs(q) + Math.abs(r) > Math.abs(p) && Math.abs(r) + Math.abs(p) > Math.abs(q))
                    return 1;
            }
        } else return 0;


    }

    return isTriangle;

}

The above implementation is Linear in time complexity. The concept is simple use the formaula they gave extracting a series of triplets of sorted elements.

Answer 6

A 100/100 php solution: http://www.rationalplanet.com/php-related/maxproductofthree-demo-task-at-codility-com.html

function solution($A) {
    $cnt = count($A);
    sort($A, SORT_NUMERIC);
    return max($A[0]*$A[1]*$A[$cnt-1], $A[$cnt-1]*$A[$cnt-2]*$A[$cnt-3]);
}