public void ReadRowsInSecondScanDoNotUseOriginalReaderForEmptyRowsArray()
{
// arrange
const int rows = 0;
const int columns = 0;
const double density = 0.01;
var originalVectors = SparseVectorHelper.GenerateSparseVectors(rows, columns, density, () => SparseVectorHelper.RandomInInterval(-100, 100, 2)).ToArray();
var originalReader = new InMemorySparseMatrixReader(originalVectors);
var originalReaderWithCounters = new SparseMatrixReaderWithMemberInvocationCounters <double>(originalReader);
using (var cachedReader = new CachedMatrixMarketReader <double>(originalReaderWithCounters))
{
// act
foreach (var row in cachedReader.ReadRows())
{
}
foreach (var row in cachedReader.ReadRows())
{
}
// assert
Assert.AreEqual(1, originalReaderWithCounters.ColumnsCountInvocations);
Assert.AreEqual(1, originalReaderWithCounters.RowsCountInvocations);
Assert.AreEqual(1, originalReaderWithCounters.ElementsCountInvocations);
Assert.AreEqual(1, originalReaderWithCounters.ReadRowsInvocations);
}
}
public void MatrixMarketReaderReadSameVectorsAsMatrixMarketWriterWrited()
{
// arrange
const int rows = 100;
const int columns = 100;
const double density = 0.01;
var originalVectors = SparseVectorHelper.GenerateSparseVectors(rows, columns, density, () => SparseVectorHelper.RandomInInterval(-100, 100, 2)).ToArray();
// act
// write
byte[] memory;
int rowsWrited;
int columnsWrited;
long elementsWrited;
using (var stream = new MemoryStream())
using (var writer = new MatrixMarketWriter <double>(stream))
{
writer.Write(originalVectors);
stream.Flush();
memory = stream.ToArray();
rowsWrited = writer.RowsCount;
columnsWrited = writer.ColumnsCount;
elementsWrited = writer.ElementsCount;
}
// read
SparseVector <double>[] readedVectors;
int rowsToRead;
int columnsToRead;
long elementToRead;
using (var stream = new MemoryStream(memory, false))
using (var reader = new MatrixMarketReader <double>(stream))
{
rowsToRead = reader.RowsCount;
columnsToRead = reader.ColumnsCount;
elementToRead = reader.ElementsCount;
readedVectors = reader.ReadRows().ToArray();
}
string text;
using (var stream = new MemoryStream(memory, false))
using (var reader = new StreamReader(stream))
{
text = reader.ReadToEnd();
}
// assert
Assert.AreEqual(rowsWrited, rowsToRead, string.Format("Rows count does not equal:\r\n{0}", text));
Assert.AreEqual(columnsWrited, columnsToRead, string.Format("Columns count does not equal:\r\n{0}", text));
Assert.AreEqual(elementsWrited, elementToRead, string.Format("Elements count does not equal:\r\n{0}", text));
Assert.IsTrue(originalVectors.SequenceEqual(readedVectors), string.Format("Vectors do not equal:\r\n{0}", text));
}
public void EqualsReturnsFalseForDifferentTypeArgument()
{
// arrange
var originalVector = SparseVectorHelper.GenerateRandomVector(1000, 0.7, () => SparseVectorHelper.RandomInInterval(-100, 100, 2));
var sv = new SparseVector <double>(originalVector);
// act
var falseResult1 = sv.Equals(string.Empty);
// assert
Assert.IsFalse(falseResult1);
}
public void EnumeratorReturnsOnlyNonZeroValuePairsOrderedByIndexAscending()
{
// arrange
var originalVector = SparseVectorHelper.GenerateRandomVector(1000, 0.7, () => SparseVectorHelper.RandomInInterval(-100, 100, 2));
var sv = new SparseVector <double>(originalVector);
var nonZeroValuesPairs = originalVector.Where(pair => pair.Value != 0.0).OrderBy(pair => pair.Key).ToArray();
// act
// assert
Assert.IsTrue(sv.SequenceEqual(nonZeroValuesPairs));
}
public void IndexedPropertyGetAccessorReturnsWhatWasSavedThroughConstructor()
{
// arrange
var originalVector = SparseVectorHelper.GenerateRandomVector(1000, 0.7, () => SparseVectorHelper.RandomInInterval(-100, 100, 2));
// act
var sv = new SparseVector <double>(originalVector);
// assert
foreach (var element in originalVector)
{
Assert.AreEqual(element.Value, sv[element.Key]);
}
}
public void GetHashCodeReturnsSameHashCodeForEqualSparseVectors()
{
// arrange
var originalVector = SparseVectorHelper.GenerateRandomVector(1000, 0.7, () => SparseVectorHelper.RandomInInterval(-100, 100, 2));
var originalVectorCopy = originalVector.ToArray();
var sv1 = new SparseVector <double>(originalVector);
var sv2 = new SparseVector <double>(originalVectorCopy);
// act
var hc1 = sv1.GetHashCode();
var hc2 = sv2.GetHashCode();
// assert
Assert.AreEqual(hc1, hc2);
}
public void EqualsReturnsTrueForSameReferenceObjects()
{
// arrange
var originalVector = SparseVectorHelper.GenerateRandomVector(1000, 0.7, () => SparseVectorHelper.RandomInInterval(-100, 100, 2));
var sv1 = new SparseVector <double>(originalVector);
SparseVector <double> sv2 = sv1;
// act
var trueResult1 = sv1.Equals(sv2);
var trueResult2 = sv1.Equals((object)sv2);
// assert
Assert.IsTrue(trueResult1);
Assert.IsTrue(trueResult2);
}
public void EqualsReturnsFalseForNullArgument()
{
// arrange
var originalVector = SparseVectorHelper.GenerateRandomVector(1000, 0.7, () => SparseVectorHelper.RandomInInterval(-100, 100, 2));
var sv1 = new SparseVector <double>(originalVector);
SparseVector <double> sv2 = null;
// act
var falseResult1 = sv1.Equals(sv2);
var falseResult2 = sv1.Equals((object)sv2);
// assert
Assert.IsFalse(falseResult1);
Assert.IsFalse(falseResult2);
}
public void EqualOperatorReturnsTrueForEqualSparseVectors()
{
// arrange
var originalVector = SparseVectorHelper.GenerateRandomVector(1000, 0.7, () => SparseVectorHelper.RandomInInterval(-100, 100, 2));
var originalVectorCopy = originalVector.ToArray();
var sv1 = new SparseVector <double>(originalVector);
var sv2 = new SparseVector <double>(originalVectorCopy);
// act
var trueResult = sv1 == sv2;
var falseResult = sv1 != sv2;
// assert
Assert.IsTrue(trueResult);
Assert.IsFalse(falseResult);
}
public void NonZeroValuesCountReturnsOnlyNonZeroElementsCount()
{
// arrange
var originalVector = SparseVectorHelper.GenerateRandomVector(1000, 0.7, () => SparseVectorHelper.RandomInInterval(-100, 100, 2));
var nonZeroElementsCount = originalVector.Count(arg => arg.Value != 0.0);
var sv = new SparseVector <double>();
// act
foreach (var element in originalVector)
{
sv[element.Key] = element.Value;
}
// assert
Assert.AreEqual(nonZeroElementsCount, sv.NonZeroValuesCount);
}
public void ReadRowsReturnSameVectorsInFirstScanForEmptyRowsArray()
{
// arrange
const int rows = 0;
const int columns = 0;
const double density = 0.01;
var originalVectors = SparseVectorHelper.GenerateSparseVectors(rows, columns, density, () => SparseVectorHelper.RandomInInterval(-100, 100, 2)).ToArray();
var originalReader = new InMemorySparseMatrixReader(originalVectors);
using (var cachedReader = new CachedMatrixMarketReader <double>(originalReader))
{
// act
var firstScanResults = originalReader.ReadRows().ZipFull(cachedReader.ReadRows(), (originalVector, cachedVector) => originalVector == cachedVector).All(b => b);
// assert
Assert.IsTrue(firstScanResults);
}
}
static void Main(string[] args)
{
const string matrixFileName = @"c:\temp\matrix.txt";
const int rows = 10000;
const int columns = 1000;
const double density = 0.01;
const int featuresCount = 10;
const int iterationsCount = 30;
Console.WriteLine("Generating matrix {0}x{1} with avg. {2} non-zero elements ({3}%)...", rows, columns, (long)((double)rows * columns * density), density * 100.0);
using (var file = new FileStream(matrixFileName, FileMode.Create))
using (var mmwriter = new MatrixMarketWriter <double>(file))
{
var vectors = SparseVectorHelper.GenerateSparseVectors(rows, columns, density, () => SparseVectorHelper.RandomInInterval(0.01, 100, 2));
mmwriter.Write(vectors);
}
Console.WriteLine("Matrix factorization (features=" + featuresCount + ", iterations=" + iterationsCount + ")...");
using (var file = new FileStream(matrixFileName, FileMode.Open))
using (var mmreader = new MatrixMarketReader <double>(file))
using (var cachedReader = new CachedMatrixMarketReader <double>(mmreader))
{
var nmf = new NMF(cachedReader);
// act
var sw = Stopwatch.StartNew();
using (var factorization = nmf.Factorize(featuresCount, iterationsCount))
{
sw.Stop();
Console.WriteLine("Factorization time: " + sw.ElapsedMilliseconds + " ms.");
Console.WriteLine("Euclidean distance calculation...");
sw = Stopwatch.StartNew();
var euclideanDistance = nmf.GetEuclideanDistance(factorization);
sw.Stop();
Console.WriteLine("Euclidean distance calculation time: " + sw.ElapsedMilliseconds + " ms.");
Console.WriteLine("Factorization Euclidean distance: " + euclideanDistance);
}
}
}
