is computed using a Schur decomposition, a Padé approximant and the inverse scaling and squaring method. The Padé approximant is then differentiated in order to obtain the Fréchet derivative

L (A, E)

References

Al–Mohy A H and Higham N J (2011) Improved inverse scaling and squaring algorithms for the matrix logarithm SIAM J. Sci. Comput. 34(4) C152–C169

Al–Mohy A H, Higham N J and Relton S D (2012) Computing the Fréchet derivative of the matrix logarithm and estimating the condition number MIMS EPrint 2012.72

Higham N J (2008) Functions of Matrices: Theory and Computation SIAM, Philadelphia, PA, USA

Parameters

Compulsory Input Parameters

1: $a (lda :)$ – 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: $e (lde :)$ – 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 $E$

Optional Input Parameters

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

Constraint: $n \geq 0$ .

Output Parameters

1: $a (lda :)$ – 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 logarithm, $\log (A)$ .
2: $e (lde :)$ – 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 $L (A, E)$
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:

$ifail = 1$: $A$ is singular so the logarithm cannot be computed.

$ifail = 2$: $A$ has eigenvalues on the negative real line. The principal logarithm is not defined in this case; nag_matop_complex_gen_matrix_frcht_log (f01kk) can be used to return a complex, non-principal log.

$ifail = 3$: $\log (A)$ 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: $n \geq 0$ .

$ifail = - 3$: Constraint: $lda \geq n$ .

$ifail = - 5$: Constraint: $lde \geq n$ .

$ifail = - 99$: 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

A^{T} A = A A^{T}

), the Schur decomposition is diagonal and the computation of the matrix logarithm reduces to evaluating the logarithm of the eigenvalues of

A

and then constructing

\log (A)

using the Schur vectors. This should give a very accurate result. In general, however, no error bounds are available for the algorithm. The sensitivity of the computation of

\log (A)

and

L (A, E)

is worst when

A

has an eigenvalue of very small modulus or has a complex conjugate pair of eigenvalues lying close to the negative real axis. See Al–Mohy and Higham (2011), Al–Mohy et al. (2012) and Section 11.2 of Higham (2008) for details and further discussion.

Further Comments

The cost of the algorithm is

O (n^{3})

floating-point operations. The real allocatable memory required is approximately

5 n^{2}

; see Al–Mohy et al. (2012) for further details.

If the matrix logarithm alone is required, without the Fréchet derivative, then nag_matop_real_gen_matrix_log (f01ej) should be used. If the condition number of the matrix logarithm is required then nag_matop_real_gen_matrix_cond_log (f01jj) should be used. If

A

has negative real eigenvalues then nag_matop_complex_gen_matrix_frcht_log (f01kk) can be used to return a complex, non-principal matrix logarithm and its Fréchet derivative

L (A, E)

Example

This example finds the principal matrix logarithm

\log (A)

and the Fréchet derivative

L (A, E)

, where

A = (\begin{array}{r} 4 & 2 & 0 & 2 \\ 3 & 3 & 1 & 1 \\ 3 & 2 & 1 & 0 \\ 3 & 3 & 1 & 2 \end{array}) and   ​ E = (\begin{array}{r} 1 & 2 & 2 & 2 \\ 0 & 0 & 3 & 1 \\ 1 & 2 & 1 & 2 \\ 1 & 3 & 1 & 1 \end{array}) .

Open in the MATLAB editor: f01jk_example

function f01jk_example


fprintf('f01jk example results\n\n');

% Logarithm of matrix A and Frechet derivative of log(A)E.

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

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

[loga, lae, ifail] = f01jk(a,e);

[ifail] = x04ca('General', ' ', loga, 'log(A):');
disp(' ');
[ifail] = x04ca('General', ' ', lae, 'L_log(A,E):');

f01jk example results

 log(A):
             1          2          3          4
 1      1.1165     0.5296    -0.4079     0.6962
 2      0.6996     0.2025     0.8192     0.4745
 3      1.3114     1.5867    -0.1433    -1.1720
 4      0.5272     1.2856     0.4055     0.2106
 
 L_log(A,E):
             1          2          3          4
 1     -0.1211     0.1974     0.1463     0.8268
 2     -1.2615    -4.1260     3.4035     2.4651
 3      1.2387     5.7968    -3.6489    -2.7203
 4      0.6231     3.7059    -1.9334    -1.8540

PDF version (NAG web site, 64-bit version, 64-bit version)

Chapter Contents

Chapter Introduction

NAG Toolbox