SUBROUTINE DPBCO(ABD,LDA,N,M,RCOND,Z,INFO) C***BEGIN PROLOGUE DPBCO C***DATE WRITTEN 780814 (YYMMDD) C***REVISION DATE 820801 (YYMMDD) C***CATEGORY NO. D2B2 C***KEYWORDS BANDED,CONDITION,DOUBLE PRECISION,FACTOR,LINEAR ALGEBRA, C LINPACK,MATRIX,POSITIVE DEFINITE C***AUTHOR MOLER, C. B., (U. OF NEW MEXICO) C***PURPOSE Factors a d.p. SYMMETRIC POSITIVE DEFINITE matrix stored in C band form and estimates the condition of the matrix. C***DESCRIPTION C C DPBCO factors a double precision symmetric positive definite C matrix stored in band form and estimates the condition of the C matrix. C C If RCOND is not needed, DPBFA is slightly faster. C To solve A*X = B , follow DPBCO by DPBSL. C To compute INVERSE(A)*C , follow DPBCO by DPBSL. C To compute DETERMINANT(A) , follow DPBCO by DPBDI. C C On Entry C C ABD DOUBLE PRECISION(LDA, N) C the matrix to be factored. The columns of the upper C triangle are stored in the columns of ABD and the C diagonals of the upper triangle are stored in the C rows of ABD . See the comments below for details. C C LDA INTEGER C the leading dimension of the array ABD . C LDA must be .GE. M + 1 . C C N INTEGER C the order of the matrix A . C C M INTEGER C the number of diagonals above the main diagonal. C 0 .LE. M .LT. N . C C On Return C C ABD an upper triangular matrix R , stored in band C form, so that A = TRANS(R)*R . C If INFO .NE. 0 , the factorization is not complete. C C RCOND DOUBLE PRECISION C an estimate of the reciprocal condition of A . C For the system A*X = B , relative perturbations C in A and B of size EPSILON may cause C relative perturbations in X of size EPSILON/RCOND . C If RCOND is so small that the logical expression C 1.0 + RCOND .EQ. 1.0 C is true, then A may be singular to working C precision. In particular, RCOND is zero if C exact singularity is detected or the estimate C underflows. If INFO .NE. 0 , RCOND is unchanged. C C Z DOUBLE PRECISION(N) C a work vector whose contents are usually unimportant. C If A is singular to working precision, then Z is C an approximate null vector in the sense that C NORM(A*Z) = RCOND*NORM(A)*NORM(Z) . C If INFO .NE. 0 , Z is unchanged. C C INFO INTEGER C = 0 for normal return. C = K signals an error condition. The leading minor C of order K is not positive definite. C C Band Storage C C If A is a symmetric positive definite band matrix, C the following program segment will set up the input. C C M = (band width above diagonal) C DO 20 J = 1, N C I1 = MAX0(1, J-M) C DO 10 I = I1, J C K = I-J+M+1 C ABD(K,J) = A(I,J) C 10 CONTINUE C 20 CONTINUE C C This uses M + 1 rows of A , except for the M by M C upper left triangle, which is ignored. C C Example: If the original matrix is C C 11 12 13 0 0 0 C 12 22 23 24 0 0 C 13 23 33 34 35 0 C 0 24 34 44 45 46 C 0 0 35 45 55 56 C 0 0 0 46 56 66 C C then N = 6 , M = 2 and ABD should contain C C * * 13 24 35 46 C * 12 23 34 45 56 C 11 22 33 44 55 66 C C LINPACK. This version dated 08/14/78 . C Cleve Moler, University of New Mexico, Argonne National Lab. C C Subroutines and Functions C C LINPACK DPBFA C BLAS DAXPY,DDOT,DSCAL,DASUM C Fortran DABS,DMAX1,MAX0,MIN0,DREAL,DSIGN C***REFERENCES DONGARRA J.J., BUNCH J.R., MOLER C.B., STEWART G.W., C *LINPACK USERS GUIDE*, SIAM, 1979. C***ROUTINES CALLED DASUM,DAXPY,DDOT,DPBFA,DSCAL C***END PROLOGUE DPBCO