LFCCB

Computes the LU factorization of a complex matrix in band storage mode and estimate its L1 condition number.

Required Arguments

A — Complex NLCA + NUCA + 1 by N array containing the N by N matrix in band storage mode to be factored. (Input)

NLCA — Number of lower codiagonals of A. (Input)

NUCA — Number of upper codiagonals of A. (Input)

FACT — Complex 2 * NLCA + NUCA + 1 by N array containing the LU factorization of the matrix A. (Output)
If A is not needed, A can share the first (NLCA + NUCA + 1) * N locations with FACT .

IPVT — Vector of length N containing the pivoting information for the LU factorization. (Output)

RCOND — Scalar containing an estimate of the reciprocal of the L1 condition number of A. (Output)

Optional Arguments

N — Order of the matrix. (Input)
Default: N = size (A,2).

LDA — Leading dimension of A exactly as specified in the dimension statement of the calling program. (Input)
Default: LDA = size (A,1).

LDFACT — Leading dimension of FACT exactly as specified in the dimension statement of the calling program. (Input)
Default: LDFACT = size (FACT,1).

FORTRAN 90 Interface

Generic: CALL LFCCB (A, NLCA, NUCA, FACT, IPVT, RCOND [, …])

Specific: The specific interface names are S_LFCCB and D_LFCCB.

FORTRAN 77 Interface

Single: CALL LFCCB (N, A, LDA, NLCA, NUCA, FACT, LDFACT, IPVT, RCOND)

Double: The double precision name is DLFCCB.

Description

Routine LFCCB performs an LU factorization of a complex banded coefficient matrix. It also estimates the condition number of the matrix. The LU factorization is done using scaled partial pivoting. Scaled partial pivoting differs from partial pivoting in that the pivoting strategy is the same as if each row were scaled to have the same ∞-norm.

The L1 condition number of the matrix A is defined to be κ(A) = ∥A∥1∥A-1∥1 Since it is expensive to compute ∥A-1∥1, the condition number is only estimated. The estimation algorithm is the same as used by LINPACK and is described by Cline et al. (1979).

If the estimated condition number is greater than 1/ɛ (where ɛ is machine precision), a warning error is issued. This indicates that very small changes in A can cause very large changes in the solution x. Iterative refinement can sometimes find the solution to such a system.

LFCCB fails if U, the upper triangular part of the factorization, has a zero diagonal element. This can occur only if A is singular or very close to a singular matrix.

The LU factors are returned in a form that is compatible with IMSL functions LFICB, LFSCB and LFDCB. To solve systems of equations with multiple right-hand-side vectors, use LFCCB followed by either LFICB or LFSCB called once for each right-hand side. The routine LFDCB can be called to compute the determinant of the coefficient matrix after LFCCB has performed the factorization.

Let F be the matrix FACT, let ml = NLCA and let mu = NUCA. The first ml  + mu + 1 rows of F contain the triangular matrix U in band storage form. The lower ml rows of F contain the multipliers needed to reconstruct L.

LFCCB is based on the LINPACK routine CGBCO; see Dongarra et al. (1979). CGBCO uses unscaled partial pivoting.

Comments

1. Workspace may be explicitly provided, if desired, by use of L2CCB/DL2CCB. The reference is:

CALL L2CCB (N, A, LDA, NLCA, NUCA, FACT, LDFACT, IPVT, RCOND, WK)

The additional argument is

WK — Complex work vector of length N.

2. Informational errors

 

Type

Code

Description

3

1

The input matrix is algorithmically singular.

4

2

The input matrix is singular.

Example

The inverse of a 4 × 4 band matrix with one upper and one lower codiagonal is computed. LFCCB is called to factor the matrix and to check for singularity or ill-conditioning. LFICB is called to determine the columns of the inverse.

 

USE LFCCB_INT

USE UMACH_INT

USE LFICB_INT

USE WRCRN_INT

! Declare variables

INTEGER LDA, LDFACT, N, NLCA, NUCA, NOUT

PARAMETER (LDA=3, LDFACT=4, N=4, NLCA=1, NUCA=1)

INTEGER IPVT(N)

REAL RCOND

COMPLEX A(LDA,N), AINV(N,N), FACT(LDFACT,N), RJ(N), RES(N)

!

! Set values for A in band form

!

! A = ( 0.0+0.0i 4.0+0.0i -2.0+2.0i -4.0-1.0i )

! ( 0.0-3.0i -0.5+3.0i 3.0-3.0i 1.0-1.0i )

! ( 6.0+1.0i 4.0+1.0i 0.0+2.0i 0.0+0.0i )

!

DATA A/(0.0,0.0), (0.0,-3.0), (6.0,1.0), (4.0,0.0), (-0.5,3.0),&

(4.0,1.0), (-2.0,2.0), (3.0,-3.0), (0.0,2.0), (-4.0,-1.0),&

(1.0,-1.0), (0.0,0.0)/

!

CALL LFCCB (A, NLCA, NUCA, FACT, IPVT, RCOND)

! Print the reciprocal condition number

! and the L1 condition number

CALL UMACH (2, NOUT)

WRITE (NOUT,99999) RCOND, 1.0E0/RCOND

! Set up the columns of the identity

! matrix one at a time in RJ

RJ = (0.0E0,0.0E0)

DO 10 J=1, N

RJ(J) = (1.0E0,0.0E0)

! RJ is the J-th column of the identity

! matrix so the following LFICB

! reference places the J-th column of

! the inverse of A in the J-th column

! of AINV

CALL LFICB (A, NLCA, NUCA, FACT, IPVT, RJ, AINV(:,J), RES)

RJ(J) = (0.0E0,0.0E0)

10 CONTINUE

! Print results

CALL WRCRN (’AINV’, AINV)

 

!

99999 FORMAT (’ RCOND = ’,F5.3,/,’ L1 condition number = ’,F6.3)

END

Output

 

RCOND = 0.022

L1 condition number = 45.933

AINV

1 2 3 4

1 ( 0.562, 0.170) ( 0.125, 0.260) (-0.385,-0.135) (-0.239,-1.165)

2 ( 0.122, 0.421) (-0.195, 0.094) ( 0.101,-0.289) ( 0.874,-0.179)

3 ( 0.034, 0.904) (-0.437, 0.090) (-0.153,-0.527) ( 1.087,-1.172)

4 ( 0.938, 0.870) (-0.347, 0.527) (-0.679,-0.374) ( 0.415,-1.759)