public void InvertibleRemoveKeyTest()
{
var addSize = 10000;
var testData = DataGenerator.Generate().Take(addSize).ToArray();
var errorRate = 0.001F;
var size = testData.Length;
var configuration = new DefaultBloomFilterConfiguration();
var bloomFilter = new InvertibleBloomFilter <TestEntity, long, sbyte>(configuration);
bloomFilter.Initialize(2 * size, errorRate);
foreach (var itm in testData)
{
bloomFilter.Add(itm);
}
var contained = testData.Count(item => bloomFilter.Contains(item));
foreach (var item in testData.Take(addSize / 2))
{
bloomFilter.RemoveKey(item.Id);
}
var containedAfterRemove = testData.Count(item => bloomFilter.Contains(item));
//tricky: assuming zero false positives.
Assert.AreEqual(contained, containedAfterRemove * 2, "Wrong item count after removal.");
}
public void ParallelInvertibleAddDifferentSizesTest()
{
var addSize = 10000;
var testData = DataGenerator.Generate().Take(addSize).ToArray();
var testData2 = DataGenerator.Generate().Skip(addSize).Take(addSize).ToArray();
var errorRate = 0.001F;
var size = testData.Length;
var configuration = new DefaultBloomFilterConfiguration();
var bloomFilter = new InvertibleBloomFilter <TestEntity, long, sbyte>(configuration);
bloomFilter.Initialize(4 * size, errorRate);
Parallel.ForEach(Partitioner.Create(testData, true), d => bloomFilter.Add(d));
var bloomFilter2 = new InvertibleBloomFilter <TestEntity, long, sbyte>(configuration);
//We have to create a foldable version.
var data = bloomFilter.Extract();
var foldFactor = configuration.FoldingStrategy.GetAllFoldFactors(data.BlockSize).Where(f => f > 1).OrderBy(f => f).First();
bloomFilter2.Initialize(addSize, data.BlockSize / foldFactor, data.HashFunctionCount);
Parallel.ForEach(Partitioner.Create(testData2, true), d => bloomFilter2.Add(d));
//add the bloom filters.
bloomFilter.Add(bloomFilter2);
var contained = testData.Union(testData2).Count(item => bloomFilter.Contains(item));
Assert.AreEqual(contained, 2 * addSize, "Not all items found in added Bloom filters");
}
public void InvertibleAddTest()
{
var addSize = 10000;
var testData = DataGenerator.Generate().Take(addSize).ToArray();
var testData2 = DataGenerator.Generate().Skip(addSize).Take(addSize).ToArray();
var errorRate = 0.001F;
var size = testData.Length;
var configuration = new DefaultBloomFilterConfiguration();
var bloomFilter = new InvertibleBloomFilter <TestEntity, long, sbyte>(configuration);
bloomFilter.Initialize(2 * size, errorRate);
foreach (var itm in testData)
{
bloomFilter.Add(itm);
}
var bloomFilter2 = new InvertibleBloomFilter <TestEntity, long, sbyte>(configuration);
bloomFilter2.Initialize(2 * size, errorRate);
foreach (var itm in testData2)
{
bloomFilter2.Add(itm);
}
bloomFilter.Add(bloomFilter2);
var contained = testData.Union(testData2).Count(item => bloomFilter.Contains(item));
Assert.AreEqual(contained, 2 * addSize, "Not all items found in added Bloom filters");
}
public void InvertibleSimpleFold()
{
var addSize = 50;
var testData = DataGenerator.Generate().Take(addSize).ToArray();
var size = testData.Length;
var configuration = new DefaultBloomFilterConfiguration();
var bloomFilter = new InvertibleBloomFilter <TestEntity, long, sbyte>(configuration);
bloomFilter.Initialize(size, 1024, (uint)3);
foreach (var itm in testData)
{
bloomFilter.Add(itm);
}
var positiveCount = DataGenerator.Generate().Take(500).Count(itm => bloomFilter.Contains(itm));
var folded = bloomFilter.Fold(4);
var positiveCountAfterFold = DataGenerator.Generate().Take(500).Count(itm => bloomFilter.Contains(itm));
Assert.AreEqual(positiveCount, positiveCountAfterFold, "False positive count different after fold");
Assert.AreEqual(256, folded.Extract().BlockSize);
Assert.IsTrue(testData.All(item => bloomFilter.Contains(item)), "False negative found");
}
public void InvertibleFalsePositiveTest()
{
var addSize = 10000;
var testData = DataGenerator.Generate().Take(addSize).ToArray();
var errorRate = 0.001F;
var size = testData.Length;
var configuration = new DefaultBloomFilterConfiguration();
var bloomFilter = new InvertibleBloomFilter <TestEntity, long, sbyte>(configuration);
bloomFilter.Initialize(size, errorRate);
foreach (var itm in testData)
{
bloomFilter.Add(itm);
}
var notFoundCount = testData.Count(itm => !bloomFilter.Contains(itm));
Assert.IsTrue(notFoundCount == 0, "False negative error rate violated");
notFoundCount = testData.Count(itm => !bloomFilter.ContainsKey(itm.Id));
Assert.IsTrue(notFoundCount == 0, "False negative error rate violated on ContainsKey");
notFoundCount = DataGenerator.Generate().Skip(addSize).Take(addSize).Count(itm => bloomFilter.Contains(itm));
Assert.IsTrue(notFoundCount <= errorRate * addSize, "False positive error rate violated");
notFoundCount = DataGenerator.Generate().Skip(addSize).Take(addSize).Count(itm => bloomFilter.ContainsKey(itm.Id));
Assert.IsTrue(notFoundCount <= errorRate * addSize, "False positive error rate violated on ContainsKey");
}
public void InvertibleCompressTest()
{
var addSize = 10000;
var errorRate = 0.001F;
var data = DataGenerator.Generate().Take(addSize).ToArray();
var filter = new InvertibleBloomFilter <TestEntity, long, sbyte>(new DefaultBloomFilterConfiguration());
filter.Initialize(50 * data.Length, errorRate);
Assert.AreEqual(filter.Capacity, 500000, "Unexpected size of Bloom filter.");
foreach (var item in data)
{
filter.Add(item);
}
//check error rate.
var notFoundCount = DataGenerator.Generate().Skip(addSize).Take(10000).Count(itm => filter.Contains(itm));
Assert.IsTrue(notFoundCount <= errorRate * addSize, "Uncompressed Bloom filter exceeded error rate.");
filter.Compress(true);
Assert.AreEqual(filter.Capacity, 12820, "Unexpected size of compressed Bloom filter.");
var compressNotFoundCount = DataGenerator.Generate().Skip(addSize).Take(10000).Count(itm => filter.Contains(itm));
Assert.IsTrue(compressNotFoundCount <= errorRate * addSize, "Compressed Bloom filter exceeded error rate.");
}