public void Merge_ManySparse()
{
const int numToMerge = 20;
const int bits = 18;
const int cardinality = 10000;
var hlls = new HyperLogLogPlus[numToMerge];
var baseline = new HyperLogLogPlus(bits, 25);
for (int i = 0; i < numToMerge; i++)
{
hlls[i] = new HyperLogLogPlus(bits, 25);
for (int j = 0; j < cardinality; j++)
{
double val = Rnd.NextDouble();
hlls[i].OfferHashed(Hash64(val));
baseline.OfferHashed(Hash64(val));
}
}
const long expectedCardinality = numToMerge * cardinality;
var hll = hlls[0];
hlls = hlls.Skip(1).ToArray();
long mergedEstimate = hll.Merge(hlls).Cardinality();
double se = expectedCardinality * (1.04 / Math.Sqrt(Math.Pow(2, bits)));
Console.WriteLine("Expect estimate: {0} is between {1} and {2}", mergedEstimate, expectedCardinality - (3 * se), expectedCardinality + (3 * se));
Assert.That(mergedEstimate, Is.InRange(expectedCardinality - (3 * se), expectedCardinality + (3 * se)));
}
public void Merge_Sparse()
{
const int numToMerge = 4;
const int bits = 18;
const int cardinality = 4000;
var hlls = new HyperLogLogPlus[numToMerge];
var baseline = new HyperLogLogPlus(bits, 25);
for (var i = 0; i < numToMerge; ++i)
{
hlls[i] = new HyperLogLogPlus(bits, 25);
for (int j = 0; j < cardinality; ++j)
{
double val = Rnd.NextDouble();
hlls[i].OfferHashed(Hash64(val));
baseline.OfferHashed(Hash64(val));
}
}
const long expectedCardinality = numToMerge * cardinality;
HyperLogLogPlus hll = hlls[0];
hlls = hlls.Skip(1).ToArray();
long mergedEstimate = hll.Merge(hlls).Cardinality();
double se = expectedCardinality * (1.04 / Math.Sqrt(Math.Pow(2, bits)));
Console.WriteLine("Expect estimate: {0} is between {1} and {2}", mergedEstimate, expectedCardinality - (3 * se), expectedCardinality + (3 * se));
double err = Math.Abs(mergedEstimate - expectedCardinality) / (double)expectedCardinality;
Console.WriteLine("Percentage error " + err);
Assert.That(err, Is.LessThan(0.1));
Assert.That(mergedEstimate, Is.InRange(expectedCardinality - (3 * se), expectedCardinality + (3 * se)));
}
