f04ydc estimates the -norm of a real rectangular matrix without accessing the matrix explicitly. It uses reverse communication for evaluating matrix products. The function may be used for estimating condition numbers of square matrices.
The function may be called by the names: f04ydc or nag_linsys_real_gen_norm_rcomm.
3Description
f04ydc computes an estimate (a lower bound) for the -norm
(1)
of an real matrix . The function regards the matrix as being defined by a user-supplied ‘Black Box’ which, given an matrix (with ) or an matrix , can return or . A reverse communication interface is used; thus control is returned to the calling program whenever a matrix product is required.
Note: this function is not recommended for use when the elements of are known explicitly; it is then more efficient to compute the -norm directly from formula (1) above.
The main use of the function is for estimating for a square matrix, , and hence the condition number , without forming explicitly ( above).
If, for example, an factorization of is available, the matrix products and required by f04ydc may be computed by back- and forward-substitutions, without computing .
The function can also be used to estimate -norms of matrix products such as and , without forming the products explicitly. Further applications are described by Higham (1988).
Since , f04ydc can be used to estimate the -norm of by working with instead of .
Higham N J (1988) FORTRAN codes for estimating the one-norm of a real or complex matrix, with applications to condition estimation ACM Trans. Math. Software14 381–396
Higham N J and Tisseur F (2000) A block algorithm for matrix -norm estimation, with an application to -norm pseudospectra SIAM J. Matrix. Anal. Appl.21 1185–1201
5Arguments
Note: this function uses reverse communication. Its use involves an initial entry, intermediate exits and re-entries, and a final exit, as indicated by the argument irevcm. Between intermediate exits and re-entries, all arguments other thanx and y must remain unchanged.
1: – Integer *Input/Output
On initial entry: must be set to .
On intermediate exit:
or , and x contains the matrix and y contains the matrix . The calling program must
(a)if , evaluate and store the result in y or if , evaluate and store the result in x,
(b)call f04ydc once again, with all the other arguments unchanged.
On intermediate re-entry: irevcm must be unchanged.
On final exit: .
Note: any values you return to f04ydc as part of the reverse communication procedure should not include floating-point NaN (Not a Number) or infinity values, since these are not handled by f04ydc. If your code inadvertently does return any NaNs or infinities, f04ydc is likely to produce unexpected results.
2: – IntegerInput
On entry: the number of rows of the matrix .
Constraint:
.
3: – IntegerInput
On entry: , the number of columns of the matrix .
Constraint:
.
4: – doubleInput/Output
Note: the dimension, dim, of the array x
must be at least
.
The th element of the matrix is stored in .
On initial entry: need not be set.
On intermediate exit:
if , contains the current matrix .
On intermediate re-entry: if , must contain .
On final exit: the array is undefined.
5: – IntegerInput
On entry: the stride separating matrix row elements in the array x.
Constraint:
.
6: – doubleInput/Output
Note: the dimension, dim, of the array y
must be at least
.
The th element of the matrix is stored in .
On initial entry: need not be set.
On intermediate exit:
if , contains the current matrix .
On intermediate re-entry: if , must contain .
On final exit: the array is undefined.
7: – IntegerInput
On entry: the stride separating matrix row elements in the array y.
Constraint:
.
8: – double *Input/Output
On initial entry: need not be set.
On intermediate re-entry: must not be changed.
On final exit: an estimate (a lower bound) for .
9: – IntegerInput
On entry: the number of columns of the matrices and . This is an argument that can be used to control the accuracy and reliability of the estimate and corresponds roughly to the number of columns of that are visited during each iteration of the algorithm.
If then a partly random starting matrix is used in the algorithm.
Suggested value:
.
Constraint:
.
10: – IntegerInput
On entry: the seed used for random number generation.
The NAG error argument (see Section 7 in the Introduction to the NAG Library CL Interface).
6Error Indicators and Warnings
NE_ALLOC_FAIL
Dynamic memory allocation failed.
See Section 3.1.2 in the Introduction to the NAG Library CL Interface for further information.
NE_BAD_PARAM
On entry, argument had an illegal value.
NE_INT
On entry, . Constraint: , or .
On entry, . Constraint: .
On entry, . Constraint: .
On initial entry, . Constraint: .
NE_INT_2
On entry, and . Constraint: .
On entry, and . Constraint: .
On entry, and . Constraint: .
On entry, and . Constraint: if , .
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.
See Section 7.5 in the Introduction to the NAG Library CL Interface for further information.
NE_NO_LICENCE
Your licence key may have expired or may not have been installed correctly.
See Section 8 in the Introduction to the NAG Library CL Interface for further information.
7Accuracy
In extensive tests on random matrices of size up to the estimate estnrm has been found always to be within a factor two of ; often the estimate has many correct figures. However, matrices exist for which the estimate is smaller than by an arbitrary factor; such matrices are very unlikely to arise in practice. See Higham and Tisseur (2000) for further details.
8Parallelism and Performance
Background information to multithreading can be found in the Multithreading documentation.
f04ydc is threaded by NAG for parallel execution in multithreaded implementations of the NAG Library.
Please consult the X06 Chapter Introduction for information on how to control and interrogate the OpenMP environment used within this function. Please also consult the Users' Note for your implementation for any additional implementation-specific information.
9Further Comments
9.1Timing
For most problems the time taken during calls to f04ydc will be negligible compared with the time spent evaluating matrix products between calls to f04ydc.
The number of matrix products required depends on the matrix . At most six products of the form and five products of the form will be required. The number of iterations is independent of the choice of .
9.2Overflow
It is your responsibility to guard against potential overflows during evaluation of the matrix products. In particular, when estimating using a triangular factorization of , f04ydc should not be called if one of the factors is exactly singular – otherwise division by zero may occur in the substitutions.
9.3Choice of
The argument controls the accuracy and reliability of the estimate. For , the algorithm behaves similarly to the LAPACK estimator xLACON. Increasing typically improves the estimate, without increasing the number of iterations required.
For , random matrices are used in the algorithm, so for repeatable results the same value of seed should be used each time.
A value of is recommended for new users.
9.4Use in Conjunction with NAG Library Functions
To estimate the -norm of the inverse of a matrix , the following
skeleton code can normally be used:
do {
f04ydc(&irevcm,m,n,x,pdx,y,pdy,&estnrm,t,seed,work,iwork,&fail);
if (irevcm == 1){
.. Code to compute y = A^(-1) x ..
}
else if (irevcm == 2){
.. Code to compute x = A^(-T) y ..
}
} (while irevcm != 0)
To compute or , solve the equation or , storing the result in y or x respectively.
The factorization will normally have been performed by a suitable function from Chapters F01, F03 or F07. Note also that many of the ‘Black Box’ functions in Chapter F04 for solving systems of equations also return a factorization of the matrix.
It is straightforward to use f04ydc for the following other types of matrix, using the named functions for factorization and solution:
nonsymmetric sparse (f11mecandf11mfc; note that f11mgc can also be used here).
For upper or lower triangular matrices, no factorization function is needed: and may be computed by calls to f16pjc (or f16pkc if the matrix is banded, or f16plc if the matrix is stored in packed form).
10Example
This example estimates the condition number of the matrix given by