public void HybridInvertibleBloomFilterEmptySetDiffTest()
{
var addSize = 1000;
var dataSet1 = DataGenerator.Generate().Take(0).ToList();
var dataSet2 = DataGenerator.Generate().Take(addSize).ToList();
var configuration = new HybridDefaultBloomFilterConfiguration();
var bloomFilter = new InvertibleHybridBloomFilter <TestEntity, long, sbyte>(configuration);
bloomFilter.Initialize(10 * addSize, 0.0001F);
foreach (var itm in dataSet1)
{
bloomFilter.Add(itm);
}
var secondBloomFilter = new InvertibleHybridBloomFilter <TestEntity, long, sbyte>(configuration);
secondBloomFilter.Initialize(10 * addSize, 0.0001F);
foreach (var itm in dataSet2)
{
secondBloomFilter.Add(itm);
}
var changed = new HashSet <long>();
var onlyInFirst = new HashSet <long>();
var onlyInSecond = new HashSet <long>();
var decoded = bloomFilter
.SubtractAndDecode(secondBloomFilter, onlyInFirst, onlyInSecond, changed);
var onlyInSet1 = dataSet1.Where(d => dataSet2.All(d2 => d2.Id != d.Id)).Select(d => d.Id).OrderBy(id => id).ToArray();
var onlyInSet2 = dataSet2.Where(d => dataSet1.All(d1 => d1.Id != d.Id)).Select(d => d.Id).OrderBy(id => id).ToArray();
var modified = dataSet1.Where(d => dataSet2.Any(d2 => d2.Id == d.Id && d2.Value != d.Value)).Select(d => d.Id).OrderBy(id => id).ToArray();
Assert.IsTrue(decoded.HasValue, "Decoding failed");
Assert.IsTrue(onlyInSet1.Length == onlyInFirst.Count, "Incorrect number of changes detected on 'only in set 1'");
Assert.IsTrue(onlyInSet2.Length == onlyInSecond.Count, "Incorrect number of changes detected on 'only in set 2'");
Assert.IsTrue(changed.Count == modified.Length, "Incorrect number of modified items detected");
}
public void HybridIbfDecodePerformance()
{
var configuration = new KeyValueLargeBloomFilterConfiguration();
var size = new[] { 1000, 10000, 100000 };
var modPercentage = new[] { 0, 0.01D, 0.1D, 0.2D, 0.5D, 1.0D };
foreach (var s in size)
{
using (
var writer =
new StreamWriter(File.Open($"hibfdecode-{s}.csv",
FileMode.Create)))
{
writer.WriteLine("timeInMs,sizeInBytes,capacity,modCount,detectedModCount,countDiff,countDiffSd,decodeSuccessRate,listADiff,listBDiff,listCDiff");
foreach (var mod in modPercentage)
{
foreach (var capacityPercentage in new[] { 0.5, 1, 2, 5, 10, 100 })
{
var sizeInBytes = new long[100];
var timeSpan = new long[50];
var countAggregate = new int[50];
var modCountResultAggregate = new int[50];
var listADiff = new int[50];
var listBDiff = new int[50];
var listCDiff = new int[50];
var decodeResult = new int[50];
for (var run = 0; run < 50; run++)
{
var dataSet1 = DataGenerator.Generate().Take(s).ToList();
var dataSet2 = DataGenerator.Generate().Take(s).ToList();
dataSet2.Modify((int)(s * mod));
var onlyInSet1 = dataSet1.Where(d => dataSet2.All(d2 => d2.Id != d.Id)).Select(d => d.Id).OrderBy(id => id).ToArray();
var onlyInSet2 = dataSet2.Where(d => dataSet1.All(d1 => d1.Id != d.Id)).Select(d => d.Id).OrderBy(id => id).ToArray();
var modified = dataSet1.Where(d => dataSet2.Any(d2 => d2.Id == d.Id && d2.Value != d.Value)).Select(d => d.Id).OrderBy(id => id).ToArray();
var idealCapacity = Math.Max(15, onlyInSet1.Count() + onlyInSet2.Count() + modified.Count());
var stopWatch = new Stopwatch();
stopWatch.Start();
var bloomFilter1 = new InvertibleHybridBloomFilter <TestEntity, long, int>(configuration);
bloomFilter1.Initialize((int)(idealCapacity * capacityPercentage), 0.01F);
foreach (var item in dataSet1)
{
bloomFilter1.Add(item);
}
var bloomFilter2 = new InvertibleHybridBloomFilter <TestEntity, long, int>(configuration);
bloomFilter2.Initialize((int)(idealCapacity * capacityPercentage), 0.01F);
foreach (var item in dataSet2)
{
bloomFilter2.Add(item);
}
var s1 = new HashSet <long>();
var s2 = new HashSet <long>();
var s3 = new HashSet <long>();
var success = bloomFilter1.SubtractAndDecode(bloomFilter2, s1, s2, s3);
stopWatch.Stop();
using (var stream = new MemoryStream())
{
_protobufTypeModel.Serialize(stream, bloomFilter1.Extract());
stream.Position = 0;
sizeInBytes[run] = stream.Length;
}
using (var stream = new MemoryStream())
{
_protobufTypeModel.Serialize(stream, bloomFilter2.Extract());
stream.Position = 0;
sizeInBytes[50 + run] = stream.Length;
}
timeSpan[run] = stopWatch.ElapsedMilliseconds;
countAggregate[run] = onlyInSet1.Count() + onlyInSet2.Count() + modified.Count();
modCountResultAggregate[run] = s1.Union(s2).Union(s3).Count(v => onlyInSet1.Contains(v) ||
onlyInSet2.Contains(v) || modified.Contains(v));
listADiff[run] = s1.Count() - onlyInSet1.Count();
listBDiff[run] = s2.Count() - onlyInSet2.Count();
listCDiff[run] = s3.Count() - modified.Count();
decodeResult[run] = success == true? 1 : 0;
}
var countAvg = (long)countAggregate.Average();
var modCountResult = (long)modCountResultAggregate.Average();
var differenceResult =
modCountResultAggregate.Select((r, i) => r - countAggregate[i]).ToArray();
var differenceSd = Math.Sqrt(differenceResult.Variance());
writer
.WriteLine($"{timeSpan.Average()},{sizeInBytes.Average()},{Math.Max(15, capacityPercentage * (int)(s * mod))},{countAvg},{modCountResult},{(long)differenceResult.Average()},{differenceSd},{1.0D * decodeResult.Sum() / 50},{listADiff.Average()},{listBDiff.Average()},{listCDiff.Average()}");
}
}
}
}
}