nag_dgesvd (f08kbc) (PDF version)
f08 Chapter Contents
f08 Chapter Introduction
NAG Library Manual

NAG Library Function Document

nag_dgesvd (f08kbc)

+ Contents

    1  Purpose
    7  Accuracy

1  Purpose

nag_dgesvd (f08kbc) computes the singular value decomposition (SVD) of a real m by n matrix A, optionally computing the left and/or right singular vectors.

2  Specification

#include <nag.h>
#include <nagf08.h>
void  nag_dgesvd (Nag_OrderType order, Nag_ComputeUType jobu, Nag_ComputeVTType jobvt, Integer m, Integer n, double a[], Integer pda, double s[], double u[], Integer pdu, double vt[], Integer pdvt, double work[], NagError *fail)

3  Description

The SVD is written as
A = UΣVT ,
where Σ is an m by n matrix which is zero except for its minm,n diagonal elements, U is an m by m orthogonal matrix, and V is an n by n orthogonal matrix. The diagonal elements of Σ are the singular values of A; they are real and non-negative, and are returned in descending order. The first minm,n columns of U and V are the left and right singular vectors of A.
Note that the function returns VT, not V.

4  References

Anderson E, Bai Z, Bischof C, Blackford S, Demmel J, Dongarra J J, Du Croz J J, Greenbaum A, Hammarling S, McKenney A and Sorensen D (1999) LAPACK Users' Guide (3rd Edition) SIAM, Philadelphia http://www.netlib.org/lapack/lug
Golub G H and Van Loan C F (1996) Matrix Computations (3rd Edition) Johns Hopkins University Press, Baltimore

5  Arguments

1:     orderNag_OrderTypeInput
On entry: the order argument specifies the two-dimensional storage scheme being used, i.e., row-major ordering or column-major ordering. C language defined storage is specified by order=Nag_RowMajor. See Section 3.2.1.3 in the Essential Introduction for a more detailed explanation of the use of this argument.
Constraint: order=Nag_RowMajor or Nag_ColMajor.
2:     jobuNag_ComputeUTypeInput
On entry: specifies options for computing all or part of the matrix U.
jobu=Nag_AllU
All m columns of U are returned in array u.
jobu=Nag_SingularVecsU
The first minm,n columns of U (the left singular vectors) are returned in the array u.
jobu=Nag_Overwrite
The first minm,n columns of U (the left singular vectors) are overwritten on the array a.
jobu=Nag_NotU
No columns of U (no left singular vectors) are computed.
Constraint: jobu=Nag_AllU, Nag_SingularVecsU, Nag_Overwrite or Nag_NotU.
3:     jobvtNag_ComputeVTTypeInput
On entry: specifies options for computing all or part of the matrix VT.
jobvt=Nag_AllVT
All n rows of VT are returned in the array vt.
jobvt=Nag_SingularVecsVT
The first minm,n rows of VT (the right singular vectors) are returned in the array vt.
jobvt=Nag_OverwriteVT
The first minm,n rows of VT (the right singular vectors) are overwritten on the array a.
jobvt=Nag_NotVT
No rows of VT (no right singular vectors) are computed.
Constraints:
  • jobvt=Nag_AllVT, Nag_SingularVecsVT, Nag_OverwriteVT or Nag_NotVT;
  • If jobu=Nag_Overwrite, jobvt cannot be Nag_OverwriteVT.
4:     mIntegerInput
On entry: m, the number of rows of the matrix A.
Constraint: m0.
5:     nIntegerInput
On entry: n, the number of columns of the matrix A.
Constraint: n0.
6:     a[dim]doubleInput/Output
Note: the dimension, dim, of the array a must be at least
  • max1,pda×n when order=Nag_ColMajor;
  • max1,m×pda when order=Nag_RowMajor.
The i,jth element of the matrix A is stored in
  • a[j-1×pda+i-1] when order=Nag_ColMajor;
  • a[i-1×pda+j-1] when order=Nag_RowMajor.
On entry: the m by n matrix A.
On exit: if jobu=Nag_Overwrite, a is overwritten with the first minm,n columns of U (the left singular vectors, stored column-wise).
If jobvt=Nag_OverwriteVT, a is overwritten with the first minm,n rows of VT (the right singular vectors, stored row-wise).
If jobuNag_Overwrite and jobvtNag_OverwriteVT, the contents of a are destroyed.
7:     pdaIntegerInput
On entry: the stride separating row or column elements (depending on the value of order) in the array a.
Constraints:
  • if order=Nag_ColMajor, pdamax1,m;
  • if order=Nag_RowMajor, pdamax1,n.
8:     s[dim]doubleOutput
Note: the dimension, dim, of the array s must be at least max1,minm,n .
On exit: the singular values of A, sorted so that s[i-1]s[i].
9:     u[dim]doubleOutput
Note: the dimension, dim, of the array u must be at least
  • max1,pdu×m when jobu=Nag_AllU;
  • max1,pdu×minm,n when jobu=Nag_SingularVecsU and order=Nag_ColMajor;
  • max1,m×pdu when jobu=Nag_SingularVecsU and order=Nag_RowMajor;
  • 1 otherwise.
The i,jth element of the matrix U is stored in
  • u[j-1×pdu+i-1] when order=Nag_ColMajor;
  • u[i-1×pdu+j-1] when order=Nag_RowMajor.
On exit: if jobu=Nag_AllU, u contains the m by m orthogonal matrix U.
If jobu=Nag_SingularVecsU, u contains the first minm,n columns of U (the left singular vectors, stored column-wise).
If jobu=Nag_NotU or Nag_Overwrite, u is not referenced.
10:   pduIntegerInput
On entry: the stride separating row or column elements (depending on the value of order) in the array u.
Constraints:
  • if order=Nag_ColMajor,
    • if jobu=Nag_AllU, pdu max1,m ;
    • if jobu=Nag_SingularVecsU, pdu max1,m ;
    • otherwise pdu1;
  • if order=Nag_RowMajor,
    • if jobu=Nag_AllU, pdumax1,m;
    • if jobu=Nag_SingularVecsU, pdumax1,minm,n;
    • otherwise pdu1.
11:   vt[dim]doubleOutput
Note: the dimension, dim, of the array vt must be at least
  • max1,pdvt×n when jobvt=Nag_AllVT;
  • max1,pdvt×n when jobvt=Nag_SingularVecsVT and order=Nag_ColMajor;
  • max1,minm,n×pdvt when jobvt=Nag_SingularVecsVT and order=Nag_RowMajor;
  • 1 otherwise.
The i,jth element of the matrix is stored in
  • vt[j-1×pdvt+i-1] when order=Nag_ColMajor;
  • vt[i-1×pdvt+j-1] when order=Nag_RowMajor.
On exit: if jobvt=Nag_AllVT, vt contains the n by n orthogonal matrix VT.
If jobvt=Nag_SingularVecsVT, vt contains the first minm,n rows of VT (the right singular vectors, stored row-wise).
If jobvt=Nag_NotVT or Nag_OverwriteVT, vt is not referenced.
12:   pdvtIntegerInput
On entry: the stride separating row or column elements (depending on the value of order) in the array vt.
Constraints:
  • if order=Nag_ColMajor,
    • if jobvt=Nag_AllVT, pdvt max1,n ;
    • if jobvt=Nag_SingularVecsVT, pdvt max1,minm,n ;
    • otherwise pdvt1;
  • if order=Nag_RowMajor,
    • if jobvt=Nag_AllVT, pdvtmax1,n;
    • if jobvt=Nag_SingularVecsVT, pdvtmax1,n;
    • otherwise pdvt1.
13:   work[minm,n]doubleOutput
On exit: if fail.code= NE_CONVERGENCE, WORK2:minm,n (using the notation described in Section 3.2.1.4 in the Essential Introduction) contains the unconverged superdiagonal elements of an upper bidiagonal matrix B whose diagonal is in s (not necessarily sorted). B satisfies A=UBVT, so it has the same singular values as A, and singular vectors related by U and VT.
14:   failNagError *Input/Output
The NAG error argument (see Section 3.6 in the Essential Introduction).

6  Error Indicators and Warnings

NE_ALLOC_FAIL
Dynamic memory allocation failed.
NE_BAD_PARAM
On entry, argument value had an illegal value.
NE_CONVERGENCE
If nag_dgesvd (f08kbc) did not converge, fail.errnum specifies how many superdiagonals of an intermediate bidiagonal form did not converge to zero.
NE_ENUM_INT_2
On entry, jobu=value, pdu=value and m=value.
Constraint: if jobu=Nag_AllU, pdu max1,m ;
if jobu=Nag_SingularVecsU, pdu max1,m ;
otherwise pdu1.
NE_ENUM_INT_3
On entry, jobu=value, pdu=value, m=value and n=value.
Constraint: if jobu=Nag_AllU, pdumax1,m;
if jobu=Nag_SingularVecsU, pdumax1,minm,n;
otherwise pdu1.
On entry, jobvt=value, pdvt=value, m=value and n=value.
Constraint: if jobvt=Nag_AllVT, pdvt max1,n ;
if jobvt=Nag_SingularVecsVT, pdvt max1,minm,n ;
otherwise pdvt1.
On entry, jobvt=value, pdvt=value, m=value and n=value.
Constraint: if jobvt=Nag_AllVT, pdvtmax1,n;
if jobvt=Nag_SingularVecsVT, pdvtmax1,n;
otherwise pdvt1.
NE_INT
On entry, m=value.
Constraint: m0.
On entry, n=value.
Constraint: n0.
On entry, pda=value.
Constraint: pda>0.
On entry, pdu=value.
Constraint: pdu>0.
On entry, pdvt=value.
Constraint: pdvt>0.
NE_INT_2
On entry, pda=value and m=value.
Constraint: pdamax1,m.
On entry, pda=value and n=value.
Constraint: pdamax1,n.
NE_INTERNAL_ERROR
An internal error has occurred in this function. Check the function call and any array sizes. If the call is correct then please contact NAG for assistance.

7  Accuracy

The computed singular value decomposition is nearly the exact singular value decomposition for a nearby matrix A+E , where
E2 = Oε A2 ,
and ε  is the machine precision. In addition, the computed singular vectors are nearly orthogonal to working precision. See Section 4.9 of Anderson et al. (1999) for further details.

8  Parallelism and Performance

nag_dgesvd (f08kbc) is threaded by NAG for parallel execution in multithreaded implementations of the NAG Library.
nag_dgesvd (f08kbc) makes calls to BLAS and/or LAPACK routines, which may be threaded within the vendor library used by this implementation. Consult the documentation for the vendor library for further information.
Please consult the Users' Note for your implementation for any additional implementation-specific information.

9  Further Comments

The total number of floating-point operations is approximately proportional to mn2  when m>n and m2n  otherwise.
The singular values are returned in descending order.
The complex analogue of this function is nag_zgesvd (f08kpc).

10  Example

This example finds the singular values and left and right singular vectors of the 6 by 4 matrix
A = 2.27 -1.54 1.15 -1.94 0.28 -1.67 0.94 -0.78 -0.48 -3.09 0.99 -0.21 1.07 1.22 0.79 0.63 -2.35 2.93 -1.45 2.30 0.62 -7.39 1.03 -2.57 ,
together with approximate error bounds for the computed singular values and vectors.
The example program for nag_dgesdd (f08kdc) illustrates finding a singular value decomposition for the case mn.

10.1  Program Text

Program Text (f08kbce.c)

10.2  Program Data

Program Data (f08kbce.d)

10.3  Program Results

Program Results (f08kbce.r)


nag_dgesvd (f08kbc) (PDF version)
f08 Chapter Contents
f08 Chapter Introduction
NAG Library Manual

© The Numerical Algorithms Group Ltd, Oxford, UK. 2014