dnbfs.f
SUBROUTINE DNBFS (ABE, LDA, N, ML, MU, V, ITASK, IND, WORK, IWORK)
C***BEGIN PROLOGUE DNBFS
C***PURPOSE Solve a general nonsymmetric banded system of linear
C equations.
C***LIBRARY SLATEC
C***CATEGORY D2A2
C***TYPE DOUBLE PRECISION (SNBFS-S, DNBFS-D, CNBFS-C)
C***KEYWORDS BANDED, LINEAR EQUATIONS, NONSYMMETRIC
C***AUTHOR Voorhees, E. A., (LANL)
C***DESCRIPTION
C
C Subroutine DNBFS solves a general nonsymmetric banded NxN
C system of double precision real linear equations using
C SLATEC subroutines DNBCO and DNBSL. These are adaptations
C of the LINPACK subroutines DGBCO and DGBSL which require
C a different format for storing the matrix elements. If
C A is an NxN double precision matrix and if X and B are
C double precision N-vectors, then DNBFS solves the equation
C
C A*X=B.
C
C A band matrix is a matrix whose nonzero elements are all
C fairly near the main diagonal, specifically A(I,J) = 0
C if I-J is greater than ML or J-I is greater than
C MU . The integers ML and MU are called the lower and upper
C band widths and M = ML+MU+1 is the total band width.
C DNBFS uses less time and storage than the corresponding
C program for general matrices (DGEFS) if 2*ML+MU .LT. N .
C
C The matrix A is first factored into upper and lower tri-
C angular matrices U and L using partial pivoting. These
C factors and the pivoting information are used to find the
C solution vector X. An approximate condition number is
C calculated to provide a rough estimate of the number of
C digits of accuracy in the computed solution.
C
C If the equation A*X=B is to be solved for more than one vector
C B, the factoring of A does not need to be performed again and
C the option to only solve (ITASK .GT. 1) will be faster for
C the succeeding solutions. In this case, the contents of A,
C LDA, N and IWORK must not have been altered by the user follow-
C ing factorization (ITASK=1). IND will not be changed by DNBFS
C in this case.
C
C
C Band Storage
C
C If A is a band matrix, the following program segment
C will set up the input.
C
C ML = (band width below the diagonal)
C MU = (band width above the diagonal)
C DO 20 I = 1, N
C J1 = MAX(1, I-ML)
C J2 = MIN(N, I+MU)
C DO 10 J = J1, J2
C K = J - I + ML + 1
C ABE(I,K) = A(I,J)
C 10 CONTINUE
C 20 CONTINUE
C
C This uses columns 1 through ML+MU+1 of ABE .
C Furthermore, ML additional columns are needed in
C ABE starting with column ML+MU+2 for elements
C generated during the triangularization. The total
C number of columns needed in ABE is 2*ML+MU+1 .
C
C Example: If the original matrix is
C
C 11 12 13 0 0 0
C 21 22 23 24 0 0
C 0 32 33 34 35 0
C 0 0 43 44 45 46
C 0 0 0 54 55 56
C 0 0 0 0 65 66
C
C then N = 6, ML = 1, MU = 2, LDA .GE. 5 and ABE should contain
C
C * 11 12 13 + , * = not used
C 21 22 23 24 + , + = used for pivoting
C 32 33 34 35 +
C 43 44 45 46 +
C 54 55 56 * +
C 65 66 * * +
C
C
C Argument Description ***
C
C ABE DOUBLE PRECISION(LDA,NC)
C on entry, contains the matrix in band storage as
C described above. NC must not be less than
C 2*ML+MU+1 . The user is cautioned to specify NC
C with care since it is not an argument and cannot
C be checked by DNBFS. The rows of the original
C matrix are stored in the rows of ABE and the
C diagonals of the original matrix are stored in
C columns 1 through ML+MU+1 of ABE .
C on return, contains an upper triangular matrix U and
C the multipliers necessary to construct a matrix L
C so that A=L*U.
C LDA INTEGER
C the leading dimension of array ABE. LDA must be great-
C er than or equal to N. (terminal error message IND=-1)
C N INTEGER
C the order of the matrix A. N must be greater
C than or equal to 1 . (terminal error message IND=-2)
C ML INTEGER
C the number of diagonals below the main diagonal.
C ML must not be less than zero nor greater than or
C equal to N . (terminal error message IND=-5)
C MU INTEGER
C the number of diagonals above the main diagonal.
C MU must not be less than zero nor greater than or
C equal to N . (terminal error message IND=-6)
C V DOUBLE PRECISION(N)
C on entry, the singly subscripted array(vector) of di-
C mension N which contains the right hand side B of a
C system of simultaneous linear equations A*X=B.
C on return, V contains the solution vector, X .
C ITASK INTEGER
C If ITASK=1, the matrix A is factored and then the
C linear equation is solved.
C If ITASK .GT. 1, the equation is solved using the existing
C factored matrix A and IWORK.
C If ITASK .LT. 1, then terminal error message IND=-3 is
C printed.
C IND INTEGER
C GT. 0 IND is a rough estimate of the number of digits
C of accuracy in the solution, X.
C LT. 0 See error message corresponding to IND below.
C WORK DOUBLE PRECISION(N)
C a singly subscripted array of dimension at least N.
C IWORK INTEGER(N)
C a singly subscripted array of dimension at least N.
C
C Error Messages Printed ***
C
C IND=-1 terminal N is greater than LDA.
C IND=-2 terminal N is less than 1.
C IND=-3 terminal ITASK is less than 1.
C IND=-4 terminal The matrix A is computationally singular.
C A solution has not been computed.
C IND=-5 terminal ML is less than zero or is greater than
C or equal to N .
C IND=-6 terminal MU is less than zero or is greater than
C or equal to N .
C IND=-10 warning The solution has no apparent significance.
C The solution may be inaccurate or the matrix
C A may be poorly scaled.
C
C Note- The above terminal(*fatal*) error messages are
C designed to be handled by XERMSG in which
C LEVEL=1 (recoverable) and IFLAG=2 . LEVEL=0
C for warning error messages from XERMSG. Unless
C the user provides otherwise, an error message
C will be printed followed by an abort.
C
C***REFERENCES J. J. Dongarra, J. R. Bunch, C. B. Moler, and G. W.
C Stewart, LINPACK Users' Guide, SIAM, 1979.
C***ROUTINES CALLED D1MACH, DNBCO, DNBSL, XERMSG
C***REVISION HISTORY (YYMMDD)
C 800812 DATE WRITTEN
C 890531 Changed all specific intrinsics to generic. (WRB)
C 890831 Modified array declarations. (WRB)
C 890831 REVISION DATE from Version 3.2
C 891214 Prologue converted to Version 4.0 format. (BAB)
C 900315 CALLs to XERROR changed to CALLs to XERMSG. (THJ)
C 900510 Convert XERRWV calls to XERMSG calls, changed GOTOs to
C IF-THEN-ELSEs. (RWC)
C 920501 Reformatted the REFERENCES section. (WRB)
C***END PROLOGUE DNBFS