Nachschlagewerk MQL5Matrizen und VektorenOpenBLASSingular Value DecompositionSingularValueDecompositionDC

SingularValueDecompositionDC

Singular Value Decomposition, "divide-and-conquer" algorithm. This algorithm is considered the fastest among other SVD algorithms (lapack function GESDD).

Computing for type matrix<double>

bool  matrix::SingularValueDecompositionDC(
   ENUM_SVD_Z      jobz,     // how to computed
   vector          S,        // vector of computed singular values
   matrix          U,        // matrix of computed left vectors U
   matrix          VT        // transposed matrix of right vectors VT
   );

Computing for type matrix<float>

bool  matrix::SingularValueDecompositionDC(
   ENUM_SVD_Z      jobz,     // how to computed
   vectorf         S,        // vector of computed singular values
   matrixf         U,        // matrix of computed left vectors U
   matrixf         VT        // transposed matrix of right vectors VT
   );

Computing for type matrix<complex>

bool  matrix::SingularValueDecompositionDC(
   ENUM_SVD_Z      jobz,     // how to computed
   vector          S,        // vector of computed singular values
   matrixc         U,        // matrix of computed left vectors U
   matrixc         VT        // transposed matrix of right vectors VT
   );

Parameters

jobz

[in] ENUM_SVD_Z enumeration value which determines the method for computing left and singular eigenvectors.

[out] Vector of singular values.

[out] Matrix of left singular vectors.

[out] Matrix of right singular vectors.

Return Value

Return true if successful, otherwise false in case of an error.

Note

Computation depends on the value of the jobz parameter.

When jobv is set to SVDZ_N, the left and right vectors are not computed. Only singular values are computed.

When jobv is set to SVDZ_A, the full matrices of the U and VT vectors are computed.

When the value is SVDZ_S, truncated matrices of vectors U and VT are computed.

ENUM_SVD_Z

An enumeration defining the way to compute left and right singular vectors.

ID	Description
SVDZ_N	Columns U or rows VT are not computed
SVDZ_A	All M columns of U or all N columns of VT are returned in arrays U and VT
SVDZ_S	The first min(M,N) columns of U or the first min(M,N) columns of VT are returned in arrays U and VT