Question

How likely is it to get a HashCode collision with the function below in following scenarios.

1. With random int values for key[0],key[1], key[2], key[3]
2. With random key values with the following constraints
   • key[0] <1,000,000
   • key[1] <10,000
   • key[2] <1,000
   • key[3] <1,000

Assume we have 10 Million objects.

int[] key=new int[4];    
public override int GetHashCode()
{
    // Use large prime multiples to create a unique hash key
    // Create the hash offsets using a "even powers of 2 minus 1" method, which gives 
    // primes most of the time.  
    int hashKey = 0;
    hashKey += 2047 * key[0];
    hashKey += 8191 * key[1];
    hashKey += 32767 * key[2];
    hashKey += 131071 * key[3];
    return hashKey;
}

Answer 1

I wrote a quick script to test this.

import random

def hash(key):
    hashKey = 0
    hashKey += 2047 * key[0]
    hashKey += 8191 * key[1]
    hashKey += 32767 * key[2]
    hashKey += 131071 * key[3]
    return hashKey

seen = set()
collisions = 0
for i in range(0,10000000):
    x = hash([random.randint(0,1000000), random.randint(0,10000), random.randint(0,1000), random.randint(0,1000)])
    if x in seen:
        collisions += 1
    else:
        seen.add(x)

print collisions

When I ran it, it told me I got 23735 collisions.
I also tried it on one million elements, and I got 247 collisions. Both numbers are averages over 4 runs.