hide long namesshow long names
hide short namesshow short names
Integer type:  int32  int64  nag_int  show int32  show int32  show int64  show int64  show nag_int  show nag_int

PDF version (NAG web site, 64-bit version, 64-bit version)
Chapter Contents
Chapter Introduction
NAG Toolbox

NAG Toolbox: nag_matop_real_gen_matrix_frcht_pow (f01jf)

 Contents

    1  Purpose
    2  Syntax
    7  Accuracy
    9  Example

Purpose

nag_matop_real_gen_matrix_frcht_pow (f01jf) computes the Fréchet derivative LA,E of the pth power (where p is real) of the real n by n matrix A applied to the real n by n matrix E. The principal matrix power Ap is also returned.

Syntax

[a, e, ifail] = f01jf(a, e, p, 'n', n)
[a, e, ifail] = nag_matop_real_gen_matrix_frcht_pow(a, e, p, 'n', n)

Description

For a matrix A with no eigenvalues on the closed negative real line, Ap (p) can be defined as
Ap= expplogA  
where logA is the principal logarithm of A (the unique logarithm whose spectrum lies in the strip z:-π<Imz<π).
The Fréchet derivative of the matrix pth power of A is the unique linear mapping ELA,E such that for any matrix E 
A+Ep - Ap - LA,E = oE .  
The derivative describes the first-order effect of perturbations in A on the matrix power Ap.
nag_matop_real_gen_matrix_frcht_pow (f01jf) uses the algorithms of Higham and Lin (2011) and Higham and Lin (2013) to compute Ap and LA,E. The real number p is expressed as p=q+r where q-1,1 and r. Then Ap=AqAr. The integer power Ar is found using a combination of binary powering and, if necessary, matrix inversion. The fractional power Aq is computed using a Schur decomposition, a Padé approximant and the scaling and squaring method. The Padé approximant is differentiated in order to obtain the Fréchet derivative of Aq and LA,E is then computed using a combination of the chain rule and the product rule for Fréchet derivatives.

References

Higham N J (2008) Functions of Matrices: Theory and Computation SIAM, Philadelphia, PA, USA
Higham N J and Lin L (2011) A Schur–Padé algorithm for fractional powers of a matrix SIAM J. Matrix Anal. Appl. 32(3) 1056–1078
Higham N J and Lin L (2013) An improved Schur–Padé algorithm for fractional powers of a matrix and their Fréchet derivatives MIMS Eprint 2013.1 Manchester Institute for Mathematical Sciences, School of Mathematics, University of Manchester http://eprints.ma.man.ac.uk/

Parameters

Compulsory Input Parameters

1:     alda: – double array
The first dimension of the array a must be at least n.
The second dimension of the array a must be at least n.
The n by n matrix A.
2:     elde: – double array
The first dimension of the array e must be at least n.
The second dimension of the array e must be at least n.
The n by n matrix E.
3:     p – double scalar
The required power of A.

Optional Input Parameters

1:     n int64int32nag_int scalar
Default: the first dimension of the arrays a, e and the second dimension of the array a. (An error is raised if these dimensions are not equal.)
n, the order of the matrix A.
Constraint: n0.

Output Parameters

1:     alda: – double array
The first dimension of the array a will be n.
The second dimension of the array a will be n.
The n by n principal matrix pth power, Ap.
2:     elde: – double array
The first dimension of the array e will be n.
The second dimension of the array e will be n.
The Fréchet derivative LA,E.
3:     ifail int64int32nag_int scalar
ifail=0 unless the function detects an error (see Error Indicators and Warnings).

Error Indicators and Warnings

Errors or warnings detected by the function:
   ifail=1
A has eigenvalues on the negative real line. The principal pth power is not defined in this case; nag_matop_complex_gen_matrix_frcht_pow (f01kf) can be used to find a complex, non-principal pth power.
   ifail=2
A is singular so the pth power cannot be computed.
   ifail=3
Ap has been computed using an IEEE double precision Padé approximant, although the arithmetic precision is higher than IEEE double precision.
   ifail=4
An unexpected internal error occurred. This failure should not occur and suggests that the function has been called incorrectly.
   ifail=-1
Constraint: n0.
   ifail=-3
Constraint: ldan.
   ifail=-5
Constraint: lden.
   ifail=-99
An unexpected error has been triggered by this routine. Please contact NAG.
   ifail=-399
Your licence key may have expired or may not have been installed correctly.
   ifail=-999
Dynamic memory allocation failed.

Accuracy

For a normal matrix A (for which ATA=AAT), the Schur decomposition is diagonal and the computation of the fractional part of the matrix power reduces to evaluating powers of the eigenvalues of A and then constructing Ap using the Schur vectors. This should give a very accurate result. In general, however, no error bounds are available for the algorithm. See Higham and Lin (2011) and Higham and Lin (2013) for details and further discussion.
If the condition number of the matrix power is required then nag_matop_real_gen_matrix_cond_pow (f01je) should be used.

Further Comments

The real allocatable memory required by the algorithm is approximately 6×n2.
The cost of the algorithm is On3 floating-point operations; see Higham and Lin (2011) and Higham and Lin (2013).
If the matrix pth power alone is required, without the Fréchet derivative, then nag_matop_real_gen_matrix_pow (f01eq) should be used. If the condition number of the matrix power is required then nag_matop_real_gen_matrix_cond_pow (f01je) should be used. If A has negative real eigenvalues then nag_matop_complex_gen_matrix_frcht_pow (f01kf) can be used to return a complex, non-principal pth power and its Fréchet derivative LA,E.

Example

This example finds Ap and the Fréchet derivative of the matrix power LA,E, where p=0.2,
A = 3 3 2 1 3 1 0 2 1 1 4 3 3 0 3 1   and   E = 1 0 2 1 0 4 5 2 1 0 0 0 2 3 3 0 .  
function f01jf_example


fprintf('f01jf example results\n\n');

% Principal power p of matrix A and Frechet derivative of A^pE.

a = [ 3 3 2 1;
      3 1 0 2;
      1 1 4 3;
      3 0 3 1];

e = [ 1 0 2 1;
      0 4 5 2;
      1 0 0 0;
      2 3 3 0];

p = 0.2;

[pa, lpae, ifail] = f01jf(a,e,p);

disp('A^p:');
disp(pa);

disp('L_p(A,E):');
disp(lpae);


f01jf example results

A^p:
    1.2446    0.2375    0.2172   -0.1359
    0.0925    1.1239   -0.1453    0.3731
   -0.0769    0.1972    1.3131    0.1837
    0.3985   -0.2902    0.1085    1.1560

L_p(A,E):
    0.2189   -0.2004    0.0509    0.0290
   -0.3177    0.4143    0.3044    0.0760
   -0.0033   -0.1335   -0.2789    0.2699
    0.1972    0.3333    0.5379   -0.5228


PDF version (NAG web site, 64-bit version, 64-bit version)
Chapter Contents
Chapter Introduction
NAG Toolbox

© The Numerical Algorithms Group Ltd, Oxford, UK. 2009–2015