C ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ C C dmset.f - part of the SLAP linear algebra library C C This library is in the public domain. C C ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ C *DECK DSDS SUBROUTINE DSDS(N, NELT, IA, JA, A, ISYM, DINV) C***BEGIN PROLOGUE DSDS C***DATE WRITTEN 890404 (YYMMDD) C***REVISION DATE 890404 (YYMMDD) C***CATEGORY NO. D2A4, D2B4 C***KEYWORDS LIBRARY=SLATEC(SLAP), C TYPE=DOUBLE PRECISION(DSDS-D), C SLAP Sparse, Diagonal C***AUTHOR Greenbaum, Anne, Courant Institute C Seager, Mark K., (LLNL) C Lawrence Livermore National Laboratory C PO BOX 808, L-300 C Livermore, CA 94550 (415) 423-3141 C seager@lll-crg.llnl.gov C***PURPOSE Diagonal Scaling Preconditioner SLAP Set Up. C Routine to compute the inverse of the diagonal of a matrix C stored in the SLAP Column format. C***DESCRIPTION C *Usage: C INTEGER N, NELT, IA(NELT), JA(NELT), ISYM C DOUBLE PRECISION A(NELT), DINV(N) C C CALL DSDS( N, NELT, IA, JA, A, ISYM, DINV ) C C *Arguments: C N :IN Integer. C Order of the Matrix. C NELT :IN Integer. C Number of elements in arrays IA, JA, and A. C IA :INOUT Integer IA(NELT). C JA :INOUT Integer JA(NELT). C A :INOUT Double Precision A(NELT). C These arrays should hold the matrix A in the SLAP Column C format. See "Description", below. C ISYM :IN Integer. C Flag to indicate symmetric storage format. C If ISYM=0, all nonzero entries of the matrix are stored. C If ISYM=1, the matrix is symmetric, and only the upper C or lower triangle of the matrix is stored. C DINV :OUT Double Precision DINV(N). C Upon return this array holds 1./DIAG(A). C C *Description C =================== S L A P Column format ================== C This routine requires that the matrix A be stored in the C SLAP Column format. In this format the non-zeros are stored C counting down columns (except for the diagonal entry, which C must appear first in each "column") and are stored in the C double precision array A. In other words, for each column C in the matrix put the diagonal entry in A. Then put in the C other non-zero elements going down the column (except the C diagonal) in order. The IA array holds the row index for C each non-zero. The JA array holds the offsets into the IA, C A arrays for the beginning of each column. That is, C IA(JA(ICOL)), A(JA(ICOL)) points to the beginning of the C ICOL-th column in IA and A. IA(JA(ICOL+1)-1), C A(JA(ICOL+1)-1) points to the end of the ICOL-th column. C Note that we always have JA(N+1) = NELT+1, where N is the C number of columns in the matrix and NELT is the number of C non-zeros in the matrix. C C Here is an example of the SLAP Column storage format for a C 5x5 Matrix (in the A and IA arrays '|' denotes the end of a C column): C C 5x5 Matrix SLAP Column format for 5x5 matrix on left. C 1 2 3 4 5 6 7 8 9 10 11 C |11 12 0 0 15| A: 11 21 51 | 22 12 | 33 53 | 44 | 55 15 35 C |21 22 0 0 0| IA: 1 2 5 | 2 1 | 3 5 | 4 | 5 1 3 C | 0 0 33 0 35| JA: 1 4 6 8 9 12 C | 0 0 0 44 0| C |51 0 53 0 55| C C With the SLAP format all of the "inner loops" of this C routine should vectorize on machines with hardware support C for vector gather/scatter operations. Your compiler may C require a compiler directive to convince it that there are C no implicit vector dependencies. Compiler directives for C the Alliant FX/Fortran and CRI CFT/CFT77 compilers are C supplied with the standard SLAP distribution. C C *Precision: Double Precision C C *Cautions: C This routine assumes that the diagonal of A is all non-zero C and that the operation DINV = 1.0/DIAG(A) will not underflow C or overflow. This is done so that the loop vectorizes. C Matricies with zero or near zero or very large entries will C have numerical difficulties and must be fixed before this C routine is called. C***REFERENCES (NONE) C***ROUTINES CALLED (NONE) C***END PROLOGUE DSDS IMPLICIT DOUBLE PRECISION(A-H,O-Z) INTEGER N, NELT, IA(NELT), JA(NELT), ISYM DOUBLE PRECISION A(NELT), DINV(N) C C Assume the Diagonal elements are the first in each column. C This loop should *VECTORIZE*. If it does not you may have C to add a compiler directive. We do not check for a zero C (or near zero) diagonal element since this would interfere C with vectorization. If this makes you nervous put a check C in! It will run much slower. C***FIRST EXECUTABLE STATEMENT DSDS 1 CONTINUE DO 10 ICOL = 1, N DINV(ICOL) = 1.0D0/A(JA(ICOL)) 10 CONTINUE C RETURN C------------- LAST LINE OF DSDS FOLLOWS ---------------------------- END *DECK DSDSCL SUBROUTINE DSDSCL( N, NELT, IA, JA, A, ISYM, X, B, DINV, JOB, $ ITOL ) C***BEGIN PROLOGUE DSDSCL C***DATE WRITTEN 890404 (YYMMDD) C***REVISION DATE 890404 (YYMMDD) C***CATEGORY NO. D2B4 C***KEYWORDS LIBRARY=SLATEC(SLAP), C TYPE=DOUBLE PRECISION(DSDSCL-D), C SLAP Sparse, Diagonal C***AUTHOR Greenbaum, Anne, Courant Institute C Seager, Mark K., (LLNL) C Lawrence Livermore National Laboratory C PO BOX 808, L-300 C Livermore, CA 94550 (415) 423-3141 C seager@lll-crg.llnl.gov C***PURPOSE Diagonal Scaling of system Ax = b. C This routine scales (and unscales) the system Ax = b C by symmetric diagonal scaling. The new system is: C -1/2 -1/2 1/2 -1/2 C D AD (D x) = D b C when scaling is selected with the JOB parameter. When C unscaling is selected this process is reversed. C The true solution is also scaled or unscaled if ITOL is set C appropriately, see below. C***DESCRIPTION C *Usage: C INTEGER N, NELT, IA(NELT), JA(NELT), ISYM, JOB, ITOL C DOUBLE PRECISION A(NELT), DINV(N) C C CALL DSDSCL( N, NELT, IA, JA, A, ISYM, X, B, DINV, JOB, ITOL ) C C *Arguments: C N :IN Integer C Order of the Matrix. C NELT :IN Integer. C Number of elements in arrays IA, JA, and A. C IA :IN Integer IA(NELT). C JA :IN Integer JA(NELT). C A :IN Double Precision A(NELT). C These arrays should hold the matrix A in the SLAP Column C format. See "Description", below. C ISYM :IN Integer. C Flag to indicate symmetric storage format. C If ISYM=0, all nonzero entries of the matrix are stored. C If ISYM=1, the matrix is symmetric, and only the upper C or lower triangle of the matrix is stored. C X :INOUT Double Precision X(N). C Initial guess that will be later used in the iterative C solution. C of the scaled system. C B :INOUT Double Precision B(N). C Right hand side vector. C DINV :OUT Double Precision DINV(N). C Upon return this array holds 1./DIAG(A). C JOB :IN Integer. C Flag indicating weather to scale or not. JOB nonzero means C do scaling. JOB = 0 means do unscaling. C ITOL :IN Integer. C Flag indicating what type of error estimation to do in the C iterative method. When ITOL = 11 the exact solution from C common block solblk will be used. When the system is scaled C then the true solution must also be scaled. If ITOL is not C 11 then this vector is not referenced. C C *Common Blocks: C SOLN :INOUT Double Precision SOLN(N). COMMON BLOCK /SOLBLK/ C The true solution, SOLN, is scaled (or unscaled) if ITOL is C set to 11, see above. C C *Description C =================== S L A P Column format ================== C This routine requires that the matrix A be stored in the C SLAP Column format. In this format the non-zeros are stored C counting down columns (except for the diagonal entry, which C must appear first in each "column") and are stored in the C double precision array A. In other words, for each column C in the matrix put the diagonal entry in A. Then put in the C other non-zero elements going down the column (except the C diagonal) in order. The IA array holds the row index for C each non-zero. The JA array holds the offsets into the IA, C A arrays for the beginning of each column. That is, C IA(JA(ICOL)), A(JA(ICOL)) points to the beginning of the C ICOL-th column in IA and A. IA(JA(ICOL+1)-1), C A(JA(ICOL+1)-1) points to the end of the ICOL-th column. C Note that we always have JA(N+1) = NELT+1, where N is the C number of columns in the matrix and NELT is the number of C non-zeros in the matrix. C C Here is an example of the SLAP Column storage format for a C 5x5 Matrix (in the A and IA arrays '|' denotes the end of a C column): C C 5x5 Matrix SLAP Column format for 5x5 matrix on left. C 1 2 3 4 5 6 7 8 9 10 11 C |11 12 0 0 15| A: 11 21 51 | 22 12 | 33 53 | 44 | 55 15 35 C |21 22 0 0 0| IA: 1 2 5 | 2 1 | 3 5 | 4 | 5 1 3 C | 0 0 33 0 35| JA: 1 4 6 8 9 12 C | 0 0 0 44 0| C |51 0 53 0 55| C C With the SLAP format all of the "inner loops" of this C routine should vectorize on machines with hardware support C for vector gather/scatter operations. Your compiler may C require a compiler directive to convince it that there are C no implicit vector dependencies. Compiler directives for C the Alliant FX/Fortran and CRI CFT/CFT77 compilers are C supplied with the standard SLAP distribution. C C *Precision: Double Precision C C *Cautions: C This routine assumes that the diagonal of A is all non-zero C and that the operation DINV = 1.0/DIAG(A) will not under- C flow or overflow. This is done so that the loop vectorizes. C Matricies with zero or near zero or very large entries will C have numerical difficulties and must be fixed before this C routine is called. C C *See Also: C DSDCG C***REFERENCES (NONE) C***ROUTINES CALLED (NONE) C***END PROLOGUE DSDSCL IMPLICIT DOUBLE PRECISION(A-H,O-Z) INTEGER N, NELT, IA(NELT), JA(NELT), ISYM, JOB, ITOL DOUBLE PRECISION A(NELT), X(N), B(N), DINV(N) COMMON /SOLBLK/ SOLN(1) C C SCALING... C IF( JOB.NE.0 ) THEN DO 10 ICOL = 1, N DINV(ICOL) = 1.0D0/SQRT( A(JA(ICOL)) ) 10 CONTINUE ELSE C C UNSCALING... C DO 15 ICOL = 1, N DINV(ICOL) = 1.0D0/DINV(ICOL) 15 CONTINUE ENDIF C DO 30 ICOL = 1, N JBGN = JA(ICOL) JEND = JA(ICOL+1)-1 DI = DINV(ICOL) DO 20 J = JBGN, JEND A(J) = DINV(IA(J))*A(J)*DI 20 CONTINUE 30 CONTINUE C DO 40 ICOL = 1, N B(ICOL) = B(ICOL)*DINV(ICOL) X(ICOL) = X(ICOL)/DINV(ICOL) 40 CONTINUE C C Check to see if we need to scale the "true solution" as well. C IF( ITOL.EQ.11 ) THEN DO 50 ICOL = 1, N SOLN(ICOL) = SOLN(ICOL)/DINV(ICOL) 50 CONTINUE ENDIF C RETURN END *DECK DSD2S SUBROUTINE DSD2S(N, NELT, IA, JA, A, ISYM, DINV) C***BEGIN PROLOGUE DSD2S C***DATE WRITTEN 890404 (YYMMDD) C***REVISION DATE 890404 (YYMMDD) C***CATEGORY NO. D2B4 C***KEYWORDS LIBRARY=SLATEC(SLAP), C TYPE=DOUBLE PRECISION(DSD2S-D), C SLAP Sparse, Diagonal C***AUTHOR Greenbaum, Anne, Courant Institute C Seager, Mark K., (LLNL) C Lawrence Livermore National Laboratory C PO BOX 808, L-300 C Livermore, CA 94550 (415) 423-3141 C seager@lll-crg.llnl.gov C***PURPOSE Diagonal Scaling Preconditioner SLAP Normal Eqns Set Up. C Routine to compute the inverse of the diagonal of the C matrix A*A'. Where A is stored in SLAP-Column format. C***DESCRIPTION C *Usage: C INTEGER N, NELT, IA(NELT), JA(NELT), ISYM C DOUBLE PRECISION A(NELT), DINV(N) C C CALL DSD2S( N, NELT, IA, JA, A, ISYM, DINV ) C C *Arguments: C N :IN Integer C Order of the Matrix. C NELT :IN Integer. C Number of elements in arrays IA, JA, and A. C IA :IN Integer IA(NELT). C JA :IN Integer JA(NELT). C A :IN Double Precision A(NELT). C These arrays should hold the matrix A in the SLAP Column C format. See "Description", below. C ISYM :IN Integer. C Flag to indicate symmetric storage format. C If ISYM=0, all nonzero entries of the matrix are stored. C If ISYM=1, the matrix is symmetric, and only the upper C or lower triangle of the matrix is stored. C DINV :OUT Double Precision DINV(N). C Upon return this array holds 1./DIAG(A*A'). C C *Description C =================== S L A P Column format ================== C This routine requires that the matrix A be stored in the C SLAP Column format. In this format the non-zeros are stored C counting down columns (except for the diagonal entry, which C must appear first in each "column") and are stored in the C double precision array A. In other words, for each column C in the matrix put the diagonal entry in A. Then put in the C other non-zero elements going down the column (except the C diagonal) in order. The IA array holds the row index for C each non-zero. The JA array holds the offsets into the IA, C A arrays for the beginning of each column. That is, C IA(JA(ICOL)), A(JA(ICOL)) points to the beginning of the C ICOL-th column in IA and A. IA(JA(ICOL+1)-1), C A(JA(ICOL+1)-1) points to the end of the ICOL-th column. C Note that we always have JA(N+1) = NELT+1, where N is the C number of columns in the matrix and NELT is the number of C non-zeros in the matrix. C C Here is an example of the SLAP Column storage format for a C 5x5 Matrix (in the A and IA arrays '|' denotes the end of a C column): C C 5x5 Matrix SLAP Column format for 5x5 matrix on left. C 1 2 3 4 5 6 7 8 9 10 11 C |11 12 0 0 15| A: 11 21 51 | 22 12 | 33 53 | 44 | 55 15 35 C |21 22 0 0 0| IA: 1 2 5 | 2 1 | 3 5 | 4 | 5 1 3 C | 0 0 33 0 35| JA: 1 4 6 8 9 12 C | 0 0 0 44 0| C |51 0 53 0 55| C C With the SLAP format all of the "inner loops" of this C routine should vectorize on machines with hardware support C for vector gather/scatter operations. Your compiler may C require a compiler directive to convince it that there are C no implicit vector dependencies. Compiler directives for C the Alliant FX/Fortran and CRI CFT/CFT77 compilers are C supplied with the standard SLAP distribution. C C *Precision: Double Precision C C *Cautions: C This routine assumes that the diagonal of A is all non-zero C and that the operation DINV = 1.0/DIAG(A*A') will not under- C flow or overflow. This is done so that the loop vectorizes. C Matricies with zero or near zero or very large entries will C have numerical difficulties and must be fixed before this C routine is called. C C *See Also: C DSDCGN C***REFERENCES (NONE) C***ROUTINES CALLED (NONE) C***END PROLOGUE DSD2S IMPLICIT DOUBLE PRECISION(A-H,O-Z) INTEGER N, NELT, IA(NELT), JA(NELT), ISYM DOUBLE PRECISION A(NELT), DINV(N) C C***FIRST EXECUTABLE STATEMENT DSD2S DO 10 I = 1, N DINV(I) = 0. 10 CONTINUE C C Loop over each column. CVD$R NOCONCUR DO 40 I = 1, N KBGN = JA(I) KEND = JA(I+1) - 1 C C Add in the contributions for each row that has a non-zero C in this column. CLLL. OPTION ASSERT (NOHAZARD) CDIR$ IVDEP CVD$ NODEPCHK DO 20 K = KBGN, KEND DINV(IA(K)) = DINV(IA(K)) + A(K)**2 20 CONTINUE IF( ISYM.EQ.1 ) THEN C C Lower triangle stored by columns => upper triangle stored by C rows with Diagonal being the first entry. Loop across the C rest of the row. KBGN = KBGN + 1 IF( KBGN.LE.KEND ) THEN DO 30 K = KBGN, KEND DINV(I) = DINV(I) + A(K)**2 30 CONTINUE ENDIF ENDIF 40 CONTINUE DO 50 I=1,N DINV(I) = 1./DINV(I) 50 CONTINUE C RETURN C------------- LAST LINE OF DSD2S FOLLOWS ---------------------------- END *DECK DS2LT SUBROUTINE DS2LT( N, NELT, IA, JA, A, ISYM, NEL, IEL, JEL, EL ) C***BEGIN PROLOGUE DS2LT C***DATE WRITTEN 890404 (YYMMDD) C***REVISION DATE 890404 (YYMMDD) C***CATEGORY NO. D2A4, D2B4 C***KEYWORDS LIBRARY=SLATEC(SLAP), C TYPE=DOUBLE PRECISION(DS2LT-D), C Linear system, SLAP Sparse, Lower Triangle C***AUTHOR Greenbaum, Anne, Courant Institute C Seager, Mark K., (LLNL) C Lawrence Livermore National Laboratory C PO BOX 808, L-300 C Livermore, CA 94550 (415) 423-3141 C seager@lll-crg.llnl.gov C***PURPOSE Lower Triangle Preconditioner SLAP Set Up. C Routine to store the lower triangle of a matrix stored C in the Slap Column format. C***DESCRIPTION C *Usage: C INTEGER N, NELT, IA(NELT), JA(NELT), ISYM C INTEGER NEL, IEL(N+1), JEL(NEL), NROW(N) C DOUBLE PRECISION A(NELT), EL(NEL) C C CALL DS2LT( N, NELT, IA, JA, A, ISYM, NEL, IEL, JEL, EL ) C C *Arguments: C N :IN Integer C Order of the Matrix. C NELT :IN Integer. C Number of non-zeros stored in A. C IA :IN Integer IA(NELT). C JA :IN Integer JA(NELT). C A :IN Double Precision A(NELT). C These arrays should hold the matrix A in the SLAP Column C format. See "Description", below. C ISYM :IN Integer. C Flag to indicate symmetric storage format. C If ISYM=0, all nonzero entries of the matrix are stored. C If ISYM=1, the matrix is symmetric, and only the lower C triangle of the matrix is stored. C NEL :OUT Integer. C Number of non-zeros in the lower triangle of A. Also C coresponds to the length of the JEL, EL arrays. C IEL :OUT Integer IEL(N+1). C JEL :OUT Integer JEL(NEL). C EL :OUT Double Precision EL(NEL). C IEL, JEL, EL contain the lower triangle of the A matrix C stored in SLAP Column format. See "Description", below C for more details bout the SLAP Column format. C C *Description C =================== S L A P Column format ================== C This routine requires that the matrix A be stored in the C SLAP Column format. In this format the non-zeros are stored C counting down columns (except for the diagonal entry, which C must appear first in each "column") and are stored in the C double precision array A. In other words, for each column C in the matrix put the diagonal entry in A. Then put in the C other non-zero elements going down the column (except the C diagonal) in order. The IA array holds the row index for C each non-zero. The JA array holds the offsets into the IA, C A arrays for the beginning of each column. That is, C IA(JA(ICOL)), A(JA(ICOL)) points to the beginning of the C ICOL-th column in IA and A. IA(JA(ICOL+1)-1), C A(JA(ICOL+1)-1) points to the end of the ICOL-th column. C Note that we always have JA(N+1) = NELT+1, where N is the C number of columns in the matrix and NELT is the number of C non-zeros in the matrix. C C Here is an example of the SLAP Column storage format for a C 5x5 Matrix (in the A and IA arrays '|' denotes the end of a C column): C C 5x5 Matrix SLAP Column format for 5x5 matrix on left. C 1 2 3 4 5 6 7 8 9 10 11 C |11 12 0 0 15| A: 11 21 51 | 22 12 | 33 53 | 44 | 55 15 35 C |21 22 0 0 0| IA: 1 2 5 | 2 1 | 3 5 | 4 | 5 1 3 C | 0 0 33 0 35| JA: 1 4 6 8 9 12 C | 0 0 0 44 0| C |51 0 53 0 55| C C *Precision: Double Precision C***REFERENCES (NONE) C***ROUTINES CALLED (NONE) C***END PROLOGUE DS2LT IMPLICIT DOUBLE PRECISION(A-H,O-Z) INTEGER N, NELT, IA(NELT), JA(NELT), ISYM INTEGER NEL, IEL(NEL), JEL(NEL) DOUBLE PRECISION A(NELT), EL(NELT) C***FIRST EXECUTABLE STATEMENT DS2LT IF( ISYM.EQ.0 ) THEN C C The matrix is stored non-symmetricly. Pick out the lower C triangle. C NEL = 0 DO 20 ICOL = 1, N JEL(ICOL) = NEL+1 JBGN = JA(ICOL) JEND = JA(ICOL+1)-1 CVD$ NOVECTOR DO 10 J = JBGN, JEND IF( IA(J).GE.ICOL ) THEN NEL = NEL + 1 IEL(NEL) = IA(J) EL(NEL) = A(J) ENDIF 10 CONTINUE 20 CONTINUE JEL(N+1) = NEL+1 ELSE C C The matrix is symmetric and only the lower triangle is C stored. Copy it to IEL, JEL, EL. C NEL = NELT DO 30 I = 1, NELT IEL(I) = IA(I) EL(I) = A(I) 30 CONTINUE DO 40 I = 1, N+1 JEL(I) = JA(I) 40 CONTINUE ENDIF RETURN C------------- LAST LINE OF DS2LT FOLLOWS ---------------------------- END *DECK DSICS SUBROUTINE DSICS(N, NELT, IA, JA, A, ISYM, NEL, IEL, JEL, $ EL, D, R, IWARN ) C***BEGIN PROLOGUE DSICS C***DATE WRITTEN 890404 (YYMMDD) C***REVISION DATE 890404 (YYMMDD) C***CATEGORY NO. D2B4 C***KEYWORDS LIBRARY=SLATEC(SLAP), C TYPE=DOUBLE PRECISION(DSICS-D), C Linear system, SLAP Sparse, Iterative Precondition C Incomplete Cholesky Factorization. C***AUTHOR Greenbaum, Anne, Courant Institute C Seager, Mark K., (LLNL) C Lawrence Livermore National Laboratory C PO BOX 808, L-300 C Livermore, CA 94550 (415) 423-3141 C seager@lll-crg.llnl.gov C***PURPOSE Incompl Cholesky Decomposition Preconditioner SLAP Set Up. C Routine to generate the Incomplete Cholesky decomposition, C L*D*L-trans, of a symmetric positive definite matrix, A, C which is stored in SLAP Column format. The unit lower C triangular matrix L is stored by rows, and the inverse of C the diagonal matrix D is stored. C***DESCRIPTION C *Usage: C INTEGER N, NELT, IA(NELT), JA(NELT), ISYM C INTEGER NEL, IEL(NEL), JEL(N+1), IWARN C DOUBLE PRECISION A(NELT), EL(NEL), D(N), R(N) C C CALL DSICS( N, NELT, IA, JA, A, ISYM, NEL, IEL, JEL, EL, D, R, C $ IWARN ) C C *Arguments: C N :IN Integer. C Order of the Matrix. C NELT :IN Integer. C Number of elements in arrays IA, JA, and A. C IA :INOUT Integer IA(NELT). C JA :INOUT Integer JA(NELT). C A :INOUT Double Precision A(NELT). C These arrays should hold the matrix A in the SLAP Column C format. See "Description", below. C ISYM :IN Integer. C Flag to indicate symmetric storage format. C If ISYM=0, all nonzero entries of the matrix are stored. C If ISYM=1, the matrix is symmetric, and only the lower C triangle of the matrix is stored. C NEL :OUT Integer. C Number of non-zeros in the lower triangle of A. Also C coresponds to the length of the JEL, EL arrays. C IEL :OUT Integer IEL(N+1). C JEL :OUT Integer JEL(NEL). C EL :OUT Double Precision EL(NEL). C IEL, JEL, EL contain the unit lower triangular factor of the C incomplete decomposition of the A matrix stored in SLAP C Row format. The Diagonal of ones *IS* stored. See C "Description", below for more details about the SLAP Row fmt. C D :OUT Double Precision D(N) C Upon return this array holds D(I) = 1./DIAG(A). C R :WORK Double Precision R(N). C Temporary double precision workspace needed for the C factorization. C IWARN :OUT Integer. C This is a warning variable and is zero if the IC factoriza- C tion goes well. It is set to the row index corresponding to C the last zero pivot found. See "Description", below. C C *Description C =================== S L A P Column format ================== C This routine requires that the matrix A be stored in the C SLAP Column format. In this format the non-zeros are stored C counting down columns (except for the diagonal entry, which C must appear first in each "column") and are stored in the C double precision array A. In other words, for each column C in the matrix put the diagonal entry in A. Then put in the C other non-zero elements going down the column (except the C diagonal) in order. The IA array holds the row index for C each non-zero. The JA array holds the offsets into the IA, C A arrays for the beginning of each column. That is, C IA(JA(ICOL)), A(JA(ICOL)) points to the beginning of the C ICOL-th column in IA and A. IA(JA(ICOL+1)-1), C A(JA(ICOL+1)-1) points to the end of the ICOL-th column. C Note that we always have JA(N+1) = NELT+1, where N is the C number of columns in the matrix and NELT is the number of C non-zeros in the matrix. C C Here is an example of the SLAP Column storage format for a C 5x5 Matrix (in the A and IA arrays '|' denotes the end of a C column): C C 5x5 Matrix SLAP Column format for 5x5 matrix on left. C 1 2 3 4 5 6 7 8 9 10 11 C |11 12 0 0 15| A: 11 21 51 | 22 12 | 33 53 | 44 | 55 15 35 C |21 22 0 0 0| IA: 1 2 5 | 2 1 | 3 5 | 4 | 5 1 3 C | 0 0 33 0 35| JA: 1 4 6 8 9 12 C | 0 0 0 44 0| C |51 0 53 0 55| C C ==================== S L A P Row format ==================== C This routine requires that the matrix A be stored in the C SLAP Row format. In this format the non-zeros are stored C counting across rows (except for the diagonal entry, which C must appear first in each "row") and are stored in the C double precision C array A. In other words, for each row in the matrix put the C diagonal entry in A. Then put in the other non-zero C elements going across the row (except the diagonal) in C order. The JA array holds the column index for each C non-zero. The IA array holds the offsets into the JA, A C arrays for the beginning of each row. That is, C JA(IA(IROW)), A(IA(IROW)) points to the beginning of the C IROW-th row in JA and A. JA(IA(IROW+1)-1), A(IA(IROW+1)-1) C points to the end of the IROW-th row. Note that we always C have IA(N+1) = NELT+1, where N is the number of rows in C the matrix and NELT is the number of non-zeros in the C matrix. C C Here is an example of the SLAP Row storage format for a 5x5 C Matrix (in the A and JA arrays '|' denotes the end of a row): C C 5x5 Matrix SLAP Row format for 5x5 matrix on left. C 1 2 3 4 5 6 7 8 9 10 11 C |11 12 0 0 15| A: 11 12 15 | 22 21 | 33 35 | 44 | 55 51 53 C |21 22 0 0 0| JA: 1 2 5 | 2 1 | 3 5 | 4 | 5 1 3 C | 0 0 33 0 35| IA: 1 4 6 8 9 12 C | 0 0 0 44 0| C |51 0 53 0 55| C C With the SLAP format some of the "inner loops" of this C routine should vectorize on machines with hardware support C for vector gather/scatter operations. Your compiler may C require a compiler directive to convince it that there are C no implicit vector dependencies. Compiler directives for C the Alliant FX/Fortran and CRI CFT/CFT77 compilers are C supplied with the standard SLAP distribution. C C The IC factorization is not alway exist for SPD matricies. C In the event that a zero pivot is found it is set to be 1.0 C and the factorization procedes. The integer variable IWARN C is set to the last row where the Diagonal was fudged. This C eventuality hardly ever occurs in practice C C *Precision: Double Precision C C *See Also: C SCG, DSICCG C***REFERENCES 1. Gene Golub & Charles Van Loan, "Matrix Computations", C John Hopkins University Press; 3 (1983) IBSN C 0-8018-3010-9. C***ROUTINES CALLED XERRWV C***END PROLOGUE DSICS IMPLICIT DOUBLE PRECISION(A-H,O-Z) INTEGER N, NELT, IA(NELT), JA(NELT), ISYM INTEGER NEL, IEL(NEL), JEL(NEL) DOUBLE PRECISION A(NELT), EL(NEL), D(N), R(N) C C Set the lower triangle in IEL, JEL, EL C***FIRST EXECUTABLE STATEMENT DSICS IWARN = 0 C C All matrix elements stored in IA, JA, A. Pick out the lower C triangle (making sure that the Diagonal of EL is one) and C store by rows. C NEL = 1 IEL(1) = 1 JEL(1) = 1 EL(1) = 1.0D0 D(1) = A(1) CVD$R NOCONCUR DO 30 IROW = 2, N C Put in the Diagonal. NEL = NEL + 1 IEL(IROW) = NEL JEL(NEL) = IROW EL(NEL) = 1.0D0 D(IROW) = A(JA(IROW)) C C Look in all the lower triangle columns for a matching row. C Since the matrix is symmetric, we can look across the C irow-th row by looking down the irow-th column (if it is C stored ISYM=0)... IF( ISYM.EQ.0 ) THEN ICBGN = JA(IROW) ICEND = JA(IROW+1)-1 ELSE ICBGN = 1 ICEND = IROW-1 ENDIF DO 20 IC = ICBGN, ICEND IF( ISYM.EQ.0 ) THEN ICOL = IA(IC) IF( ICOL.GE.IROW ) GOTO 20 ELSE ICOL = IC ENDIF JBGN = JA(ICOL)+1 JEND = JA(ICOL+1)-1 IF( JBGN.LE.JEND .AND. IA(JEND).GE.IROW ) THEN CVD$ NOVECTOR DO 10 J = JBGN, JEND IF( IA(J).EQ.IROW ) THEN NEL = NEL + 1 JEL(NEL) = ICOL EL(NEL) = A(J) GOTO 20 ENDIF 10 CONTINUE ENDIF 20 CONTINUE 30 CONTINUE IEL(N+1) = NEL+1 C C Sort ROWS of lower triangle into descending order (count out C along rows out from Diagonal). C DO 60 IROW = 2, N IBGN = IEL(IROW)+1 IEND = IEL(IROW+1)-1 IF( IBGN.LT.IEND ) THEN DO 50 I = IBGN, IEND-1 CVD$ NOVECTOR DO 40 J = I+1, IEND IF( JEL(I).GT.JEL(J) ) THEN JELTMP = JEL(J) JEL(J) = JEL(I) JEL(I) = JELTMP ELTMP = EL(J) EL(J) = EL(I) EL(I) = ELTMP ENDIF 40 CONTINUE 50 CONTINUE ENDIF 60 CONTINUE C C Perform the Incomplete Cholesky decomposition by looping C over the rows. C Scale the first column. Use the structure of A to pick out C the rows with something in column 1. C IRBGN = JA(1)+1 IREND = JA(2)-1 DO 65 IRR = IRBGN, IREND IR = IA(IRR) C Find the index into EL for EL(1,IR). C Hint: it's the second entry. I = IEL(IR)+1 EL(I) = EL(I)/D(1) 65 CONTINUE C DO 110 IROW = 2, N C C Update the IROW-th diagonal. C DO 66 I = 1, IROW-1 R(I) = 0.0D0 66 CONTINUE IBGN = IEL(IROW)+1 IEND = IEL(IROW+1)-1 IF( IBGN.LE.IEND ) THEN CLLL. OPTION ASSERT (NOHAZARD) CDIR$ IVDEP CVD$ NODEPCHK DO 70 I = IBGN, IEND R(JEL(I)) = EL(I)*D(JEL(I)) D(IROW) = D(IROW) - EL(I)*R(JEL(I)) 70 CONTINUE C C Check to see if we gota problem with the diagonal. C IF( D(IROW).LE.0.0D0 ) THEN IF( IWARN.EQ.0 ) IWARN = IROW D(IROW) = 1.0D0 ENDIF ENDIF C C Update each EL(IROW+1:N,IROW), if there are any. C Use the structure of A to determine the Non-zero elements C of the IROW-th column of EL. C IRBGN = JA(IROW) IREND = JA(IROW+1)-1 DO 100 IRR = IRBGN, IREND IR = IA(IRR) IF( IR.LE.IROW ) GOTO 100 C Find the index into EL for EL(IR,IROW) IBGN = IEL(IR)+1 IEND = IEL(IR+1)-1 IF( JEL(IBGN).GT.IROW ) GOTO 100 DO 90 I = IBGN, IEND IF( JEL(I).EQ.IROW ) THEN ICEND = IEND 91 IF( JEL(ICEND).GE.IROW ) THEN ICEND = ICEND - 1 GOTO 91 ENDIF C Sum up the EL(IR,1:IROW-1)*R(1:IROW-1) contributions. CLLL. OPTION ASSERT (NOHAZARD) CDIR$ IVDEP CVD$ NODEPCHK DO 80 IC = IBGN, ICEND EL(I) = EL(I) - EL(IC)*R(JEL(IC)) 80 CONTINUE EL(I) = EL(I)/D(IROW) GOTO 100 ENDIF 90 CONTINUE C C If we get here, we have real problems... CALL XERRWV('DSICS -- A and EL data structure mismatch'// $ ' in row (i1)',53,1,2,1,IROW,0,0,0.0,0.0) 100 CONTINUE 110 CONTINUE C C Replace diagonals by their inverses. C CVD$ CONCUR DO 120 I =1, N D(I) = 1.0D0/D(I) 120 CONTINUE RETURN C------------- LAST LINE OF DSICS FOLLOWS ---------------------------- END *DECK DSILUS SUBROUTINE DSILUS(N, NELT, IA, JA, A, ISYM, NL, IL, JL, $ L, DINV, NU, IU, JU, U, NROW, NCOL) C***BEGIN PROLOGUE DSILUS C***DATE WRITTEN 890404 (YYMMDD) C***REVISION DATE 890404 (YYMMDD) C***CATEGORY NO. D2A4, D2B4 C***KEYWORDS LIBRARY=SLATEC(SLAP), C TYPE=DOUBLE PRECISION(DSILUS-D), C Non-Symmetric Linear system, Sparse, C Iterative Precondition, Incomplete LU Factorization C***AUTHOR Greenbaum, Anne, Courant Institute C Seager, Mark K., (LLNL) C Lawrence Livermore National Laboratory C PO BOX 808, L-300 C Livermore, CA 94550 (415) 423-3141 C seager@lll-crg.llnl.gov C***PURPOSE Incomplete LU Decomposition Preconditioner SLAP Set Up. C Routine to generate the incomplete LDU decomposition of a C matrix. The unit lower triangular factor L is stored by C rows and the unit upper triangular factor U is stored by C columns. The inverse of the diagonal matrix D is stored. C No fill in is allowed. C***DESCRIPTION C *Usage: C INTEGER N, NELT, IA(NELT), JA(NELT), ISYM C INTEGER NL, IL(N+1), JL(NL), NU, IU(N+1), JU(NU) C INTEGER NROW(N), NCOL(N) C DOUBLE PRECISION A(NELT), L(NL), U(NU), DINV(N) C C CALL DSILUS( N, NELT, IA, JA, A, ISYM, NL, IL, JL, L, C $ DINV, NU, IU, JU, U, NROW, NCOL ) C C *Arguments: C N :IN Integer C Order of the Matrix. C NELT :IN Integer. C Number of elements in arrays IA, JA, and A. C IA :IN Integer IA(NELT). C JA :IN Integer JA(NELT). C A :IN Double Precision A(NELT). C These arrays should hold the matrix A in the SLAP Column C format. See "Description", below. C ISYM :IN Integer. C Flag to indicate symmetric storage format. C If ISYM=0, all nonzero entries of the matrix are stored. C If ISYM=1, the matrix is symmetric, and only the lower C triangle of the matrix is stored. C NL :OUT Integer. C Number of non-zeros in the EL array. C IL :OUT Integer IL(N+1). C JL :OUT Integer JL(NL). C L :OUT Double Precision L(NL). C IL, JL, L contain the unit ower triangular factor of the C incomplete decomposition of some matrix stored in SLAP C Row format. The Diagonal of ones *IS* stored. See C "DESCRIPTION", below for more details about the SLAP format. C NU :OUT Integer. C Number of non-zeros in the U array. C IU :OUT Integer IU(N+1). C JU :OUT Integer JU(NU). C U :OUT Double Precision U(NU). C IU, JU, U contain the unit upper triangular factor of the C incomplete decomposition of some matrix stored in SLAP C Column format. The Diagonal of ones *IS* stored. See C "Description", below for more details about the SLAP C format. C NROW :WORK Integer NROW(N). C NROW(I) is the number of non-zero elements in the I-th row C of L. C NCOL :WORK Integer NCOL(N). C NCOL(I) is the number of non-zero elements in the I-th C column of U. C C *Description C IL, JL, L should contain the unit lower triangular factor of C the incomplete decomposition of the A matrix stored in SLAP C Row format. IU, JU, U should contain the unit upper factor C of the incomplete decomposition of the A matrix stored in C SLAP Column format This ILU factorization can be computed by C the DSILUS routine. The diagonals (which is all one's) are C stored. C C =================== S L A P Column format ================== C This routine requires that the matrix A be stored in the C SLAP Column format. In this format the non-zeros are stored C counting down columns (except for the diagonal entry, which C must appear first in each "column") and are stored in the C double precision array A. In other words, for each column C in the matrix put the diagonal entry in A. Then put in the C other non-zero elements going down the column (except the C diagonal) in order. The IA array holds the row index for C each non-zero. The JA array holds the offsets into the IA, C A arrays for the beginning of each column. That is, C IA(JA(ICOL)), A(JA(ICOL)) points to the beginning of the C ICOL-th column in IA and A. IA(JA(ICOL+1)-1), C A(JA(ICOL+1)-1) points to the end of the ICOL-th column. C Note that we always have JA(N+1) = NELT+1, where N is the C number of columns in the matrix and NELT is the number of C non-zeros in the matrix. C C Here is an example of the SLAP Column storage format for a C 5x5 Matrix (in the A and IA arrays '|' denotes the end of a C column): C C 5x5 Matrix SLAP Column format for 5x5 matrix on left. C 1 2 3 4 5 6 7 8 9 10 11 C |11 12 0 0 15| A: 11 21 51 | 22 12 | 33 53 | 44 | 55 15 35 C |21 22 0 0 0| IA: 1 2 5 | 2 1 | 3 5 | 4 | 5 1 3 C | 0 0 33 0 35| JA: 1 4 6 8 9 12 C | 0 0 0 44 0| C |51 0 53 0 55| C C ==================== S L A P Row format ==================== C This routine requires that the matrix A be stored in the C SLAP Row format. In this format the non-zeros are stored C counting across rows (except for the diagonal entry, which C must appear first in each "row") and are stored in the C double precision C array A. In other words, for each row in the matrix put the C diagonal entry in A. Then put in the other non-zero C elements going across the row (except the diagonal) in C order. The JA array holds the column index for each C non-zero. The IA array holds the offsets into the JA, A C arrays for the beginning of each row. That is, C JA(IA(IROW)), A(IA(IROW)) points to the beginning of the C IROW-th row in JA and A. JA(IA(IROW+1)-1), A(IA(IROW+1)-1) C points to the end of the IROW-th row. Note that we always C have IA(N+1) = NELT+1, where N is the number of rows in C the matrix and NELT is the number of non-zeros in the C matrix. C C Here is an example of the SLAP Row storage format for a 5x5 C Matrix (in the A and JA arrays '|' denotes the end of a row): C C 5x5 Matrix SLAP Row format for 5x5 matrix on left. C 1 2 3 4 5 6 7 8 9 10 11 C |11 12 0 0 15| A: 11 12 15 | 22 21 | 33 35 | 44 | 55 51 53 C |21 22 0 0 0| JA: 1 2 5 | 2 1 | 3 5 | 4 | 5 1 3 C | 0 0 33 0 35| IA: 1 4 6 8 9 12 C | 0 0 0 44 0| C |51 0 53 0 55| C C *Precision: Double Precision C *See Also: C SILUR C***REFERENCES 1. Gene Golub & Charles Van Loan, "Matrix Computations", C John Hopkins University Press; 3 (1983) IBSN C 0-8018-3010-9. C***ROUTINES CALLED (NONE) C***END PROLOGUE DSILUS IMPLICIT DOUBLE PRECISION(A-H,O-Z) INTEGER N, NELT, IA(NELT), JA(NELT), ISYM, NL, IL(N+1), JL(NELT+1) INTEGER NU, IU(NELT+1), JU(N+1), NROW(N), NCOL(N) DOUBLE PRECISION A(NELT), L(NL), DINV(N), U(NU) C C Count number of elements in each row of the lower triangle. C***FIRST EXECUTABLE STATEMENT DSILUS DO 10 I=1,N NROW(I) = 0 NCOL(I) = 0 10 CONTINUE CVD$R NOCONCUR CVD$R NOVECTOR DO 30 ICOL = 1, N JBGN = JA(ICOL)+1 JEND = JA(ICOL+1)-1 IF( JBGN.LE.JEND ) THEN DO 20 J = JBGN, JEND IF( IA(J).LT.ICOL ) THEN NCOL(ICOL) = NCOL(ICOL) + 1 ELSE NROW(IA(J)) = NROW(IA(J)) + 1 IF( ISYM.NE.0 ) NCOL(IA(J)) = NCOL(IA(J)) + 1 ENDIF 20 CONTINUE ENDIF 30 CONTINUE JU(1) = 1 IL(1) = 1 DO 40 ICOL = 1, N IL(ICOL+1) = IL(ICOL) + NROW(ICOL) JU(ICOL+1) = JU(ICOL) + NCOL(ICOL) NROW(ICOL) = IL(ICOL) NCOL(ICOL) = JU(ICOL) 40 CONTINUE C C Copy the matrix A into the L and U structures. DO 60 ICOL = 1, N DINV(ICOL) = A(JA(ICOL)) JBGN = JA(ICOL)+1 JEND = JA(ICOL+1)-1 IF( JBGN.LE.JEND ) THEN DO 50 J = JBGN, JEND IROW = IA(J) IF( IROW.LT.ICOL ) THEN C Part of the upper triangle. IU(NCOL(ICOL)) = IROW U(NCOL(ICOL)) = A(J) NCOL(ICOL) = NCOL(ICOL) + 1 ELSE C Part of the lower triangle (stored by row). JL(NROW(IROW)) = ICOL L(NROW(IROW)) = A(J) NROW(IROW) = NROW(IROW) + 1 IF( ISYM.NE.0 ) THEN C Symmetric...Copy lower triangle into upper triangle as well. IU(NCOL(IROW)) = ICOL U(NCOL(IROW)) = A(J) NCOL(IROW) = NCOL(IROW) + 1 ENDIF ENDIF 50 CONTINUE ENDIF 60 CONTINUE C C Sort the rows of L and the columns of U. DO 110 K = 2, N JBGN = JU(K) JEND = JU(K+1)-1 IF( JBGN.LT.JEND ) THEN DO 80 J = JBGN, JEND-1 DO 70 I = J+1, JEND IF( IU(J).GT.IU(I) ) THEN ITEMP = IU(J) IU(J) = IU(I) IU(I) = ITEMP TEMP = U(J) U(J) = U(I) U(I) = TEMP ENDIF 70 CONTINUE 80 CONTINUE ENDIF IBGN = IL(K) IEND = IL(K+1)-1 IF( IBGN.LT.IEND ) THEN DO 100 I = IBGN, IEND-1 DO 90 J = I+1, IEND IF( JL(I).GT.JL(J) ) THEN JTEMP = JU(I) JU(I) = JU(J) JU(J) = JTEMP TEMP = L(I) L(I) = L(J) L(J) = TEMP ENDIF 90 CONTINUE 100 CONTINUE ENDIF 110 CONTINUE C C Perform the incomplete LDU decomposition. DO 300 I=2,N C C I-th row of L INDX1 = IL(I) INDX2 = IL(I+1) - 1 IF(INDX1 .GT. INDX2) GO TO 200 DO 190 INDX=INDX1,INDX2 IF(INDX .EQ. INDX1) GO TO 180 INDXR1 = INDX1 INDXR2 = INDX - 1 INDXC1 = JU(JL(INDX)) INDXC2 = JU(JL(INDX)+1) - 1 IF(INDXC1 .GT. INDXC2) GO TO 180 160 KR = JL(INDXR1) 170 KC = IU(INDXC1) IF(KR .GT. KC) THEN INDXC1 = INDXC1 + 1 IF(INDXC1 .LE. INDXC2) GO TO 170 ELSEIF(KR .LT. KC) THEN INDXR1 = INDXR1 + 1 IF(INDXR1 .LE. INDXR2) GO TO 160 ELSEIF(KR .EQ. KC) THEN L(INDX) = L(INDX) - L(INDXR1)*DINV(KC)*U(INDXC1) INDXR1 = INDXR1 + 1 INDXC1 = INDXC1 + 1 IF(INDXR1 .LE. INDXR2 .AND. INDXC1 .LE. INDXC2) GO TO 160 ENDIF 180 L(INDX) = L(INDX)/DINV(JL(INDX)) 190 CONTINUE C C ith column of u 200 INDX1 = JU(I) INDX2 = JU(I+1) - 1 IF(INDX1 .GT. INDX2) GO TO 260 DO 250 INDX=INDX1,INDX2 IF(INDX .EQ. INDX1) GO TO 240 INDXC1 = INDX1 INDXC2 = INDX - 1 INDXR1 = IL(IU(INDX)) INDXR2 = IL(IU(INDX)+1) - 1 IF(INDXR1 .GT. INDXR2) GO TO 240 210 KR = JL(INDXR1) 220 KC = IU(INDXC1) IF(KR .GT. KC) THEN INDXC1 = INDXC1 + 1 IF(INDXC1 .LE. INDXC2) GO TO 220 ELSEIF(KR .LT. KC) THEN INDXR1 = INDXR1 + 1 IF(INDXR1 .LE. INDXR2) GO TO 210 ELSEIF(KR .EQ. KC) THEN U(INDX) = U(INDX) - L(INDXR1)*DINV(KC)*U(INDXC1) INDXR1 = INDXR1 + 1 INDXC1 = INDXC1 + 1 IF(INDXR1 .LE. INDXR2 .AND. INDXC1 .LE. INDXC2) GO TO 210 ENDIF 240 U(INDX) = U(INDX)/DINV(IU(INDX)) 250 CONTINUE C C ith diagonal element 260 INDXR1 = IL(I) INDXR2 = IL(I+1) - 1 IF(INDXR1 .GT. INDXR2) GO TO 300 INDXC1 = JU(I) INDXC2 = JU(I+1) - 1 IF(INDXC1 .GT. INDXC2) GO TO 300 270 KR = JL(INDXR1) 280 KC = IU(INDXC1) IF(KR .GT. KC) THEN INDXC1 = INDXC1 + 1 IF(INDXC1 .LE. INDXC2) GO TO 280 ELSEIF(KR .LT. KC) THEN INDXR1 = INDXR1 + 1 IF(INDXR1 .LE. INDXR2) GO TO 270 ELSEIF(KR .EQ. KC) THEN DINV(I) = DINV(I) - L(INDXR1)*DINV(KC)*U(INDXC1) INDXR1 = INDXR1 + 1 INDXC1 = INDXC1 + 1 IF(INDXR1 .LE. INDXR2 .AND. INDXC1 .LE. INDXC2) GO TO 270 ENDIF C 300 CONTINUE C C replace diagonal lts by their inverses. CVD$ VECTOR DO 430 I=1,N DINV(I) = 1./DINV(I) 430 CONTINUE C RETURN C------------- LAST LINE OF DSILUS FOLLOWS ---------------------------- END