public void ShouldCorrectlyCalculateHammingDistanceBetweenLongs()
{
var length = 50000;
for (int run = 0; run < 1000; run++)
{
var x = GenerateByteArray(length);
var a = hashConverter.ToInts(x, length / 4);
var y = GenerateByteArray(length);
var b = hashConverter.ToInts(y, length / 4);
var byteSimilarity = similarityUtility.CalculateHammingSimilarity(x, y);
var longSimilarity = similarityUtility.CalculateHammingSimilarity(a, b, 4);
Assert.AreEqual(byteSimilarity, longSimilarity);
}
}
/// <summary>
/// Compute LSH hash buckets which will be inserted into hash tables.
/// Each fingerprint will have a candidate in each of the hash tables.
/// </summary>
/// <param name = "minHashes">Min Hashes gathered from every fingerprint [N = 100]</param>
/// <param name = "numberOfHashTables">Number of hash tables [L = 25]</param>
/// <param name = "numberOfHashesPerTable">Number of min hashes per key [N = 4]</param>
/// <param name = "hashBucketsCount">Max number of hash buckets per hash table</param>
/// <returns>Collection of Pairs with Key = Hash table index, Value = Hash bin</returns>
protected virtual int[] GroupIntoHashTables(byte[] minHashes, int numberOfHashTables, int numberOfHashesPerTable, int hashBucketsCount)
{
int[] hashes = hashConverter.ToInts(minHashes, numberOfHashTables);
if (hashBucketsCount == 0)
{
return(hashes);
}
for (int i = 0; i < hashes.Length; ++i)
{
hashes[i] = System.Math.Abs(hashes[i] * LargePrime % hashBucketsCount);
}
return(hashes);
}