public void HybridIntersectDifferentFiltersTest()
{
var addSize = 10000;
var testData = DataGenerator.Generate().Take(addSize).ToArray();
var errorRate = 0.001F;
var size = testData.Length;
var configuration = new HybridDefaultBloomFilterConfiguration();
var bloomFilter = new InvertibleHybridBloomFilter <TestEntity, long, sbyte>(configuration);
bloomFilter.Initialize(2 * size, errorRate);
foreach (var itm in testData)
{
bloomFilter.Add(itm);
}
var bloomFilter2 = new InvertibleHybridBloomFilter <TestEntity, long, sbyte>(configuration);
bloomFilter2.Initialize(2 * size, errorRate);
foreach (var itm in testData.Skip(1000))
{
bloomFilter2.Add(itm);
}
bloomFilter.Intersect(bloomFilter2);
Assert.AreEqual(9000, bloomFilter.ItemCount);
var count = testData.Skip(1000).Count(bloomFilter.Contains);
//Note: intersect introduces a horrible error rate when utilizing XOR, so don't actually use intersect.
//There are however definitions of operations possible where the intersect would not have this horrible effect.
Assert.IsTrue(count > 6000);
Assert.IsTrue(testData.Take(1000).All(d => !bloomFilter.Contains(d)));
}
public void HybridIntersectEqualFiltersTest()
{
var addSize = 10000;
var testData = DataGenerator.Generate().Take(addSize).ToArray();
var errorRate = 0.001F;
var size = testData.Length;
var configuration = new HybridDefaultBloomFilterConfiguration();
var bloomFilter = new InvertibleHybridBloomFilter <TestEntity, long, sbyte>(configuration);
bloomFilter.Initialize(2 * size, errorRate);
foreach (var itm in testData)
{
bloomFilter.Add(itm);
}
var bloomFilter2 = new InvertibleHybridBloomFilter <TestEntity, long, sbyte>(configuration);
bloomFilter2.Initialize(2 * size, errorRate);
foreach (var itm in testData)
{
bloomFilter2.Add(itm);
}
bloomFilter.Intersect(bloomFilter2);
Assert.AreEqual(addSize, bloomFilter.ItemCount);
Assert.IsTrue(testData.All(bloomFilter.Contains));
}