public void IterateCollect()
{
QuantileAggregation quantiles = new QuantileAggregation(0.5);
ExponentialHistogramAggregator aggregator = new ExponentialHistogramAggregator(quantiles);
aggregator.Update(1);
aggregator.Update(2);
aggregator.Update(3);
aggregator.Update(4);
aggregator.Update(5);
var stats = (HistogramStatistics)aggregator.Collect();
Assert.NotNull(stats);
Assert.Equal(1, stats.Quantiles.Length);
Assert.Equal(0.5, stats.Quantiles[0].Quantile);
Assert.Equal(3, stats.Quantiles[0].Value);
aggregator.Update(9);
aggregator.Update(8);
aggregator.Update(7);
aggregator.Update(6);
aggregator.Update(5);
stats = (HistogramStatistics)aggregator.Collect();
Assert.NotNull(stats);
Assert.Equal(1, stats.Quantiles.Length);
Assert.Equal(0.5, stats.Quantiles[0].Quantile);
Assert.Equal(7, stats.Quantiles[0].Value);
}
public void DifferencesGreaterThanErrorBound()
{
QuantileAggregation quantiles = new QuantileAggregation(0.5);
quantiles.MaxRelativeError = 0.0001;
ExponentialHistogramAggregator aggregator = new ExponentialHistogramAggregator(quantiles);
aggregator.Update(90.01);
aggregator.Update(90.01);
aggregator.Update(90.02);
aggregator.Update(90.03);
aggregator.Update(90.04);
aggregator.Update(100.01);
aggregator.Update(100.01);
aggregator.Update(100.02);
aggregator.Update(100.03);
aggregator.Update(100.04);
var stats = (HistogramStatistics)aggregator.Collect();
Assert.Equal(0.5, stats.Quantiles[0].Quantile);
//At default error of 0.001 result of 100 would be acceptable, but with higher precision it is not
Assert.True(100 < stats.Quantiles[0].Value);
Assert.True(Math.Abs(100.01 - stats.Quantiles[0].Value) <= 100.01 * quantiles.MaxRelativeError);
}
public void NoUpdatesInSomeIntervals()
{
QuantileAggregation quantiles = new QuantileAggregation(0.5);
ExponentialHistogramAggregator aggregator = new ExponentialHistogramAggregator(quantiles);
aggregator.Update(1);
aggregator.Update(2);
aggregator.Update(3);
aggregator.Update(4);
aggregator.Update(5);
var stats = (HistogramStatistics)aggregator.Collect();
stats = (HistogramStatistics)aggregator.Collect();
Assert.NotNull(stats);
Assert.Equal(0, stats.Quantiles.Length);
}
public void OneUpdate()
{
QuantileAggregation quantiles = new QuantileAggregation(0.5);
ExponentialHistogramAggregator aggregator = new ExponentialHistogramAggregator(quantiles);
aggregator.Update(99);
var stats = (HistogramStatistics)aggregator.Collect();
Assert.NotNull(stats);
Assert.Equal(1, stats.Quantiles.Length);
Assert.Equal(0.5, stats.Quantiles[0].Quantile);
Assert.Equal(99, stats.Quantiles[0].Value);
}
public void UnorderedQuantiles()
{
QuantileAggregation quantiles = new QuantileAggregation(0.9, 0.1);
ExponentialHistogramAggregator aggregator = new ExponentialHistogramAggregator(quantiles);
aggregator.Update(1);
aggregator.Update(2);
aggregator.Update(3);
aggregator.Update(4);
aggregator.Update(5);
var stats = (HistogramStatistics)aggregator.Collect();
Assert.Equal(0.1, stats.Quantiles[0].Quantile);
Assert.Equal(1, stats.Quantiles[0].Value);
Assert.Equal(0.9, stats.Quantiles[1].Quantile);
Assert.Equal(5, stats.Quantiles[1].Value);
}
public void OutOfBoundsQuantiles()
{
QuantileAggregation quantiles = new QuantileAggregation(-3, 100);
ExponentialHistogramAggregator aggregator = new ExponentialHistogramAggregator(quantiles);
aggregator.Update(1);
aggregator.Update(2);
aggregator.Update(3);
aggregator.Update(4);
aggregator.Update(5);
var stats = (HistogramStatistics)aggregator.Collect();
Assert.Equal(-3, stats.Quantiles[0].Quantile);
Assert.Equal(1, stats.Quantiles[0].Value);
Assert.Equal(100, stats.Quantiles[1].Quantile);
Assert.Equal(5, stats.Quantiles[1].Value);
}
public void MixedValues()
{
QuantileAggregation quantiles = new QuantileAggregation(0.5);
ExponentialHistogramAggregator aggregator = new ExponentialHistogramAggregator(quantiles);
aggregator.Update(19);
aggregator.Update(-4);
aggregator.Update(100_000);
aggregator.Update(-0.5);
aggregator.Update(-189.3231);
var stats = (HistogramStatistics)aggregator.Collect();
Assert.NotNull(stats);
Assert.Equal(1, stats.Quantiles.Length);
Assert.Equal(0.5, stats.Quantiles[0].Quantile);
Assert.Equal(-0.5, stats.Quantiles[0].Value);
}
public void FilterNaNAndInfinities()
{
QuantileAggregation quantiles = new QuantileAggregation(0, 1);
ExponentialHistogramAggregator aggregator = new ExponentialHistogramAggregator(quantiles);
aggregator.Update(double.NaN);
aggregator.Update(-double.NaN);
aggregator.Update(double.PositiveInfinity);
aggregator.Update(double.NegativeInfinity);
aggregator.Update(100);
var stats = (HistogramStatistics)aggregator.Collect();
Assert.NotNull(stats);
Assert.Equal(2, stats.Quantiles.Length);
Assert.Equal(0, stats.Quantiles[0].Quantile);
Assert.Equal(100, stats.Quantiles[0].Value);
Assert.Equal(1, stats.Quantiles[1].Quantile);
Assert.Equal(100, stats.Quantiles[1].Value);
}
public void DifferencesLessThanErrorBound()
{
QuantileAggregation quantiles = new QuantileAggregation(0.5);
ExponentialHistogramAggregator aggregator = new ExponentialHistogramAggregator(quantiles);
aggregator.Update(90.01);
aggregator.Update(90.01);
aggregator.Update(90.02);
aggregator.Update(90.03);
aggregator.Update(90.04);
aggregator.Update(100.01);
aggregator.Update(100.01);
aggregator.Update(100.02);
aggregator.Update(100.03);
aggregator.Update(100.04);
var stats = (HistogramStatistics)aggregator.Collect();
Assert.Equal(0.5, stats.Quantiles[0].Quantile);
Assert.Equal(100, stats.Quantiles[0].Value);
Assert.True(Math.Abs(100.01 - stats.Quantiles[0].Value) <= 100.01 * quantiles.MaxRelativeError);
}