public DoubleSVDecomp SVDecomp()
{
if (MatrixSVDecomp == null)
{
var decompServer = new DoubleSVDecompServer
{
ComputeLeftVectors = true,
ComputeRightVectors = true,
ComputeFull = true
};
MatrixSVDecomp = decompServer.GetDecomp(Matrix);
}
return(MatrixSVDecomp);
}
/// <summary>
/// 奇异值分解 singular value decomposition
/// </summary>
public static void DoSvd()
{
Utility.Status("Creating SVD");
// 源文件数, 源文件中所有独特词数目, 源文件-索引, 源文件单词-索引
int totalNumberOfSourceFiles = TfDictionary.Count;
int totalDistinctTermsInAllSourceFiles = IdfDictionary.Count;
Dictionary <string, int> allSourceFilesWithIndex = TfDictionary.Keys.Select((x, index) => new { Name = x, Index = index }).ToDictionary(x => x.Name, x => x.Index);
Dictionary <string, int> allSourceWordsWithIndex = IdfDictionary.Keys.Select((x, index) => new { Name = x, Index = index }).ToDictionary(x => x.Name, x => x.Index);
// 源码矩阵 行数: 源码中单词数, 源文件数
double[,] sourceMatrix = new double[totalDistinctTermsInAllSourceFiles, totalNumberOfSourceFiles];
foreach (var fileNameWithTfDictionary in TfDictionary)
{
// 文件索引
int fileIndex = allSourceFilesWithIndex[fileNameWithTfDictionary.Key];
foreach (var fileWordWithTf in fileNameWithTfDictionary.Value)
{
sourceMatrix[allSourceWordsWithIndex[fileWordWithTf.Key], fileIndex] = fileWordWithTf.Value;
}
}
// 创建矩阵
DoubleMatrix generalMatrix = new DoubleMatrix(sourceMatrix);
// 奇异值分解 A = U S V.T
var svd = new DoubleSVDecomp(generalMatrix);
_uk = new Dictionary <int, DoubleMatrix>();
_sk = new Dictionary <int, DoubleMatrix>();
_vkTranspose = new Dictionary <int, DoubleMatrix>();
Utility.LsiKs.Where(x => x <= svd.Cols).ToList().ForEach(k =>
{
// 创建k维矩阵
Utility.Status("Creating k matrix of size " + k);
DoubleMatrix U = svd.LeftVectors;
DoubleMatrix V = svd.RightVectors;
// 通过数组创建矩阵, 行优先存储, 可再改
_uk.Add(k, Utility.GetDimMatrix(U, U.Rows, k));
_sk.Add(k, Utility.GetDiagonal(svd.SingularValues, k));
_vkTranspose.Add(k, NMathFunctions.Transpose(Utility.GetDimMatrix(V, V.Rows, k)));
});
}
public DenseMatrix(string path)
{
(Matrix, Size) = ReadMatrix(path);
MatrixSVDecomp = null;
}