All use is subject to licence, see For any commercial application, a separate licence must be signed.

Size: px
Start display at page:

Download "All use is subject to licence, see For any commercial application, a separate licence must be signed."

Transcription

1 HSL MI6 PAKAGE SPEIFIATION HSL SUMMARY This routine uses the BiGStab (Bionjugate Gradient Stabilized) method to solve the n n unsymmetric linear system Ax = b, optionally using preconditioning. If P L, PR are the preconditioning matrices, the routine actually solves the preconditioned system Ax = b, with A = PLAPR and b = PLb and recovers the solution x = PRx. If P L = I, preconditioning is said to be from the right, if P R = I, it is said to be from the left, and otherwise it is from both sides. Reverse communication is used for preconditioning operations and matrix-vector products of the form Az. ATTRIBUTES Version: Types: Real (single, double). alls: FD15, _OPY, _DOT, _NRM, _SAL, _AXPY. Language: Fortran 77. Original date: March 001. Origin: N.I.M. Gould and J.A. Scott, Rutherford Appleton Laboratory. Remark: This is a threadsafe version of MI06A and supersedes it. HOW TO USE THE PAKAGE.1 Argument lists and calling sequence There are two entries: (a) MI6I/ID sets default values for control parameters. It should normally be called once prior to calls to MI6A/AD. (b) MI6A/AD uses the Bionjugate Gradient Stabilized method to solve Ax = b, optionally using preconditioning. MI6A/AD uses reverse communication for preconditioning operations and matrix-vector products..1.1 To set default values for the control parameters The single precision version ALL MI6I(INTL,NTL,ISAVE,RSAVE) The double precision version ALL MI6ID(INTL,NTL,ISAVE,RSAVE) INTL is an INTEGER array of length 8 that need not be set by the user. On return it contains default values (see Section. for details). NTL is a REAL (DOUBLE PREISION in the D version) array of length 5 that need not be set by the user. On return it contains default values (see Section. for details). ISAVE is an INTEGER array of length 14 used by MI6 as private workspace and must not be altered by the user. RSAVE is a REAL (DOUBLE PREISION in the D version) array of length 9 used by MI6 as private workspace and must not be altered by the user. HSL 007 MI6 Version Documentation date: 1st August 007 1

2 MI6 HSL To solve Ax = b The single precision version ALL MI6A(IAT,N,W,LDW,LOY,LOZ,RESID,INTL,NTL,INFO,ISAVE,RSAVE) The double precision version ALL MI6AD(IAT,N,W,LDW,LOY,LOZ,RESID,INTL,NTL,INFO,ISAVE,RSAVE) IAT is an INTEGER variable. Prior to the first call to MI6A/AD, IAT must be set by the user to 0. On each exit, IAT indicates the action required by the user. Possible values of IAT and the action required are as follows: N W LDW 1 An error has occurred and the user must terminate the computation (see INFO(1)). 1 If INTL(4) = 0 (the default), convergence has been achieved and the user should terminate the computation. If INTL(4) is nonzero, the user may test for convergence. If the user does not wish to test for convergence (we do not recommend the user tests for convergence each time IAT = 1 is returned) or if convergence has not been achieved, the user must recall MI6A/AD without changing any of the arguments. The user must perform the matrix-vector product y := Az and recall MI6A/AD. The vectors y and z are held in columns LOY and LOZ of array W, respectively. The vector z must be unchanged by the user. 3 The user must perform the preconditioning operation y := PLPRz, (.1) where P L, PR are the preconditioning matrices, and recall MI6A/AD. The vectors y and z are held in columns LOY and LOZ of array W, respectively. The vector z must be unchanged by the user. is an INTEGER variable that must be set by the user to n, the order of the matrix A. This variable must be preserved by the user between calls to MI6A/AD. This argument is not altered by the routine. Restriction: N 1. is a REAL (DOUBLE PREISION in the D version) two-dimensional array with dimensions (LDW, 8). Prior to the first call, column 1 must hold the right-hand side vector b and, if INTL(5) is nonzero, column must hold the initial estimate of the solution vector x. On exit with IAT = 1, column 1 holds the current residual vector r = b Ax, and the current estimate of the solution x is held in column. On exit with IAT > 1, the user is required to calculate y in column LOY of W (see argument IAT). The remaining columns of W must not be altered by the user between calls to MI6A/AD. is an INTEGER variable that must be set by the user to the first dimension of the array W. This argument is not altered by the routine. Restriction: LDW N. LOY,LOZ are INTEGER variables that need not be set by the user. On exit with IAT > 1, they indicate which columns of W contain the vectors y and z (see argument IAT). These arguments must not be altered by the user between calls to MI6A/AD. RESID is a REAL (DOUBLE PREISION in the D version) variable that need not be set by the user. On exit with IAT = 1, RESID holds the -norm of the current residual vector b Ax, where x is the current estimate of the solution. INTL is an INTEGER array of length 8 that contains control parameters. Default values for the components may be set by a call to MI6I/ID. Details of the control parameters are given in Section.. This argument is not altered by the routine. NTL is a REAL (DOUBLE PREISION in the D version) array of length 5 that contains control parameters and must be (.) MI6 Version Documentation date: 1st August 007 HSL 007

3 HSL 007 MI6 set by the user. Default values for the components may be set by a call to MI6I/ID. Details of the control parameters are given in Section.. This argument is not altered by the routine. INFO is an INTEGER array of length 4 that need not be set by the user. It is used to hold information about the execution of the subroutine. On exit from MI6A/AD, a value for INFO(1) of zero indicates that the subroutine has performed successfully. For nonzero values, see Section.3. INFO() holds the number of iterations performed.. INFO(3) and INFO(4) are not currently used. ISAVE is an INTEGER array of length 14 used by MI6 as private workspace and must not be altered by the user. RSAVE is a REAL (DOUBLE PREISION in the D version) array of length 9 used by MI6 as private workspace and must not be altered by the user.. ontrol parameters The elements of the arrays INTL and NTL control the action of MI6A/AD. Default values may be set by calling MI6I/ID. INTL(1) is the stream number for error messages and has the default value 6. Printing of error messages is suppressed if INTL(1) 0. INTL() is the stream number for warning messages and has the default value 6. Printing of warning messages is suppressed if INTL() 0. INTL(3) controls whether the user wishes to use preconditioning. It has default value 0 and in this case no preconditioning is used. If INTL(3) is nonzero, the user will be expected to perform preconditioning. INTL(4) controls whether the convergence test offered by MI6A/AD is to be used. It has default value 0 and in this case the computed solution x is accepted if b Ax is less than or equal to max( b Ax *NTL(1), NTL()), where x is the initial estimate of the solution. If the user does not want to use this test for convergence, INTL(4) should be nonzero. In this case, the user may test for convergence when IAT = 1 is returned. INTL(5) controls whether the user wishes to supply an initial estimate of the solution vector x. It has default value T 0 and in this case x = (0, 0,..., 0) is used as the initial estimate. If the user wishes to supply an initial estimate, INTL(5) should be nonzero and the initial estimate placed in the first N entries of column of the array W prior to the first call to MI6A/AD. INTL(6) determines the maximum number of iterations allowed. It has default value 1 and in this case the maximum number of iterations allowed is n. If the user does not want the maximum to be n, INTL(6) should be set to the maximum number of iterations the user does wish to allow. Values of INTL(6) which are less than or equal to zero are treated as if they were the default 1. INTRL(7) and INTRL(8) are set to zero by MI6I/ID but not currently used by MI6A/AD. NTL(1) and NTL() are the convergence tolerances (see Section 4). NTL(1) has default value u, where u is the relative machine precision (that is, the smallest machine number such that 1 + u > 1), while NTL() has default value zero. If INTL(4) is nonzero, NTL(1) and NTL() are not accessed by MI6A/AD. NTL(3) is the breakdown tolerance (see Section 4). It has default u, where u is the relative machine precision. NTRL(4) and NTRL(5) are set to zero by MI6I/ID but not currently used by MI6A/AD..3 Error diagnostics If MI6A/AD returns with a negative value of INFO(1), an error has occurred; if MI6A/AD returns with a positive value of INFO(1), a warning has been issued. Error messages are output on unit INTL(1) and warnings on unit INTL(). Possible non-zero values of INFO(1) are given below. 1 Value of N out of range. N < 1. Immediate return with input parameters unchanged. HSL 007 MI6 Version Documentation date: 1st August 007 3

4 MI6 HSL 007 Value of LDW out of range. LDW < N. Immediate return with input parameters unchanged. 3 Algorithm has broken down (see Section 4). 4 The maximum number of iterations determined by the control parameter INTL(6) has been exceeded. +1 The user-supplied convergence tolerance NTL(1) lies outside the interval (u,1.0), where u is the machine precision. NTL(1) is reset to the default convergence tolerance u..4 Underflows The nature of the calculations performed in this subroutine means that underflows are likely to occur. It is quite safe to set numbers that underflow to zero, and action by the user may be required to ensure that this is done efficiently by the computing system in use. 3 GENERAL INFORMATION Use of common: None. Other routines called directly: MI6A/AD calls HSL routine FD15A/AD, and the BLAS kernels SNRM/DNRM, SOPY/DOPY, SAXPY/DAXPY, SSAL/DSAL, SDOT/DDOT. Input/output: Error messages are printed on unit INTL(1) and warnings on unit INTL(); see Section.3. Restrictions: N 1, LDW N. 4 METHOD The Bionjugate Gradient Stabilized method is due to van der Vorst (199). The algorithm used by MI6A/AD proceeds as follows: heck the input data for errors. Set INFO(1) and return with IAT = 1 if a fatal error is encountered. if INTL(5) is nonzero let x be the initial guess supplied by the user Return with IAT = for the user to compute Ax. T set x = (0, 0,..., 0) ompute the initial residual r = b Ax if INTL(6) is greater then zero ITMAX = INTL(6) ITMAX = N do i = 1,,..., ITMAX T (i 1) ρ i 1 = (r ) r if ρ i 1 < NTL(3)*N (i 1) if ρ i 1 < NTL(3)* r r the method has broken down. Set INFO(1) and return with IAT = 1. if i = 1 (1) (1) p = u 4 MI6 Version Documentation date: 1st August 007 HSL 007

5 HSL 007 MI6 end do Reference β i 1 = (ρ i 1/ ρ i )(α i 1/ ω i 1) (i 1) i 1 i 1 p = r + β i 1(p ω i 1v ) if INTL(3) is nonzero Return with IAT = 3 for the user to compute p = PLPRp. p = p Return with IAT = for the user to compute v = Ap. T α i = ρ i 1/ (r ) v (i 1) s = r α i v (i 1) x = x + α i p Set r = s if INTL(4) is zero, (k) if r max( r NTL(1),NTL()) convergence has been achieved. Return with IAT = 1. Return with IAT = 1 to allow the user to check for convergence. if INTL(3) is nonzero Return with IAT = 3 for the user to compute ŝ = PLPRs. ŝ = s Return with IAT = for the user to compute t = Aŝ. T T ω i = t s/ t t if ω < NTL(3)*N i if ω i < NTL(3)* s / t the method has broken down. Set INFO(1) and return with IAT = 1. x = x + ω i ŝ r = s ω i t if INTL(4) is zero, (k) if r max( r NTL(1),NTL()) convergence has been achieved. Return with IAT = 1. Return with IAT = 1 to allow the user to check for convergence. H. van der Vorst (199). Bi-GSTAB: a fast and smoothly converging variant of Bi-G for the solution of nonsymmetric linear systems. SIAM J. Sci. Stat. omput., 13, HSL 007 MI6 Version Documentation date: 1st August 007 5

6 MI6 HSL EXAMPLE OF USE The following program illustrates the calling sequence for MI6. Solve the linear system A x = b, where A is tridiagonal with superdiagonal values 1, subdiagonals -1 and diagonals, and where the inverse of the diagonal of A is used as a preconditioner.. Parameters.. INTEGER N,LDW PARAMETER (N=10,LDW=N) DOUBLE PREISION TWO,ONE,THREE PARAMETER (TWO=.0D+0,ONE=1.0D+0,THREE=3.0D+0).... Local Scalars.. DOUBLE PREISION RESID INTEGER I,IAT,LOY,LOZ.... Local Arrays.. DOUBLE PREISION NTL(5),W(LDW,8) INTEGER INTL(8),INFO(4) INTEGER ISAVE(14) DOUBLE PREISION RSAVE(9).... External Subroutines.. EXTERNAL MI6AD,MI6ID.. ALL MI6ID(INTL,NTL,ISAVE,RSAVE) Preconditioning is required INTL(3) = 1 Set right hand side, b W(1,1) = THREE DO 10 I =,N - 1 W(I,1) = TWO 10 ONTINUE W(N,1) = ONE Perform an iteration of the Bionjugate Gradient Stabilized method IAT = 0 0 ONTINUE ALL MI6AD(IAT,N,W,LDW,LOY,LOZ,RESID,INTL,NTL,INFO,ISAVE, + RSAVE) IF (IAT.LT.0) THEN WRITE (6,FMT=900) INFO(1) GO TO 50 END IF IF (IAT.EQ.) THEN Perform the matrix-vector product W(1,LOY) = TWO*W(1,LOZ) + W(,LOZ) DO 40 I =,N - 1 W(I,LOY) = -W(I-1,LOZ) + TWO*W(I,LOZ) + W(I+1,LOZ) 40 ONTINUE W(N,LOY) = -W(N-1,LOZ) + TWO*W(N,LOZ) GO TO 0 END IF 6 MI6 Version Documentation date: 1st August 007 HSL 007

7 HSL 007 MI6 IF (IAT.EQ.3) THEN Perform the preconditioning operation DO 30 I = 1,N W(I,LOY) = W(I,LOZ)/TWO 30 ONTINUE GO TO 0 END IF Solution found WRITE (6,FMT=9000) INFO(), (W(I,),I=1,N) IF (INFO(1).GT.0) WRITE (6,FMT=9010) INFO(1) 50 ONTINUE STOP 9000 FORMAT (I6,' iterations required by MI6 ',//' Solution = ', + / (1P,5D10.)) 9010 FORMAT (' Warning: INFO( 1 ) = ',I,' on exit ') 900 FORMAT (' Error return: INFO( 1 ) = ',I,' on exit ') END This produces the following output: 10 iterations required by MI6 Solution = 1.00D D D D D D D D D D+00 HSL 007 MI6 Version Documentation date: 1st August 007 7

All use is subject to licence, see For any commercial application, a separate licence must be signed.

All use is subject to licence, see   For any commercial application, a separate licence must be signed. HS PAKAGE SPEIFIATION HS 2007 1 SUMMARY This routine uses the Generalized Minimal Residual method with restarts every m iterations, GMRES(m), to solve the n n unsymmetric linear system Ax = b, optionally

More information

PACKAGE SPECIFICATION HSL 2013

PACKAGE SPECIFICATION HSL 2013 HSL MC73 PACKAGE SPECIFICATION HSL 2013 1 SUMMARY Let A be an n n matrix with a symmetric sparsity pattern. HSL MC73 has entries to compute the (approximate) Fiedler vector of the unweighted or weighted

More information

PACKAGE SPECIFICATION HSL To solve a symmetric, sparse and positive definite set of linear equations Ax = b i.e. j=1

PACKAGE SPECIFICATION HSL To solve a symmetric, sparse and positive definite set of linear equations Ax = b i.e. j=1 MA61 PACKAGE SPECIFICATION HSL 2013 1 SUMMARY To solve a symmetric, sparse and positive definite set of linear equations Ax = b i.e. n a ij x j = b i j=1 i=1,2,...,n The solution is found by a preconditioned

More information

CALL MC36A(NMAX,NZMAX,TITLE,ELT,NROW,NCOL,NNZERO, * COLPTR,ROWIND,VALUES,MAXRHS,NRHS,NZRHS,RHSPTR, * RHSIND,RHSVAL,GUESS,SOLN,IW,ICNTL,INFO)

CALL MC36A(NMAX,NZMAX,TITLE,ELT,NROW,NCOL,NNZERO, * COLPTR,ROWIND,VALUES,MAXRHS,NRHS,NZRHS,RHSPTR, * RHSIND,RHSVAL,GUESS,SOLN,IW,ICNTL,INFO) PAKAGE SPEIFIATION 1 SUMMARY To read a sparse matrix, coded in a Harwell-Boeing format with possible right-hand sides. The subroutine reads assembled as well as unassembled matrices, and returns them in

More information

MA27 1 SUMMARY 2 HOW TO USE THE PACKAGE

MA27 1 SUMMARY 2 HOW TO USE THE PACKAGE PAKAGE SPEIFIATION HSL ARHIVE 1 SUMMARY To solve a sparse symmetric system of linear equations. Given a sparse symmetric matrix A = {a ij} n-vector b, this subroutine solves the system Ax=b. The matrix

More information

MC64 1 SUMMARY 2 HOW TO USE THE PACKAGE PACKAGE SPECIFICATION HSL 2013

MC64 1 SUMMARY 2 HOW TO USE THE PACKAGE PACKAGE SPECIFICATION HSL 2013 MC64 PACKAGE SPECIFICATION HSL 2013 1 SUMMARY Given a sparse matrix A = {a ij} n n, this subroutine attempts to find a column permutation vector that makes the permuted matrix have n entries on its diagonal.

More information

ME22. All use is subject to licence. ME22 v Documentation date: 19th March SUMMARY 2 HOW TO USE THE PACKAGE

ME22. All use is subject to licence. ME22 v Documentation date: 19th March SUMMARY 2 HOW TO USE THE PACKAGE ME22 PAKAGE SPEIFIATION HSL 2013 1 SUMMARY Given a sparse matrix A = {a ij} n n and a row permutation matrix P and a column permutation matrix Q, this subroutine performs the permutation à = PAQ. The non-zero

More information

PACKAGE SPECIFICATION HSL 2013

PACKAGE SPECIFICATION HSL 2013 PACKAGE SPECIFICATION HSL 2013 1 SUMMARY This subroutine uses the Lanczos algorithm to compute the part of the spectrum of a large symmetric matrix A that lies in a specified interval, that is, it computes

More information

All use is subject to licence. See For any commercial application, a separate license must be signed.

All use is subject to licence. See   For any commercial application, a separate license must be signed. GALAHAD LRT USER DOCUMENTATION GALAHAD Optimization Library version 30 SUMMARY Given a real m by n matrix A, a real m vector b and scalars σ > 0, µ 0 and p, this package finds an approximate minimizer

More information

All use is subject to licence. See For any commercial application, a separate license must be signed.

All use is subject to licence. See  For any commercial application, a separate license must be signed. HSL HSL MI20 PACKAGE SPECIFICATION HSL 2007 1 SUMMARY Given an n n sparse matrix A and an n vector z, HSL MI20 computes the vector x = Mz, where M is an algebraic multigrid (AMG) v-cycle preconditioner

More information

NAG Fortran Library Routine Document F11DSF.1

NAG Fortran Library Routine Document F11DSF.1 NAG Fortran Library Routine Document Note: before using this routine, please read the Users Note for your implementation to check the interpretation of bold italicised terms and other implementation-dependent

More information

PACKAGE SPECIFICATION HSL 2013

PACKAGE SPECIFICATION HSL 2013 MC55 PACKAGE SPECIFICATION HSL 2013 1 SUMMARY To write a supplementary file in Rutherford-Boeing format. The HSL routine MC54 can be used to write matrices in Rutherford-Boeing format and the HSL routine

More information

PACKAGE SPECIFICATION HSL 2013

PACKAGE SPECIFICATION HSL 2013 MA41 PACKAGE SPECIFICATION HSL 2013 1 SUMMARY To solve a sparse unsymmetric system of linear equations. Given a square unsymmetric sparse matrix A of order n T and an n-vector b, this subroutine solves

More information

PACKAGE SPECIFICATION HSL 2013

PACKAGE SPECIFICATION HSL 2013 MC56 PACKAGE SPECIFICATION HSL 2013 1 SUMMARY To read a file held in Rutherford-Boeing format. This file can contain either a sparse matrix or supplementary data. The HSL routines MC54 and MC55 can be

More information

NAG Library Function Document nag_sparse_nsym_sol (f11dec)

NAG Library Function Document nag_sparse_nsym_sol (f11dec) f11 Large Scale Linear Systems NAG Library Function Document nag_sparse_nsym_sol () 1 Purpose nag_sparse_nsym_sol () solves a real sparse nonsymmetric system of linear equations, represented in coordinate

More information

HSL MA48 1 SUMMARY 2 HOW TO USE THE PACKAGE. All use is subject to licence. 1 C INTERFACE HSL 2013

HSL MA48 1 SUMMARY 2 HOW TO USE THE PACKAGE. All use is subject to licence.   1 C INTERFACE HSL 2013 HSL MA48 C INTERFACE HSL 2013 1 SUMMARY To solve a sparse unsymmetric system of linear equations. Given a sparse matrix A = {a i j } m n and a vector b, this subroutine solves the system Ax = b or the

More information

This subroutine finds a local minimizer of the differentiable function F(x), of the n variables x,..., x, subject to. 1 n general linear constraints

This subroutine finds a local minimizer of the differentiable function F(x), of the n variables x,..., x, subject to. 1 n general linear constraints PAKAGE SPEIFIATION HSL ARHIVE 1 SUMMARY This subroutine finds a local minimizer of the differentiable function F(x), of the n variables x,..., x, subect to 1 n general linear constraints (1) and simple

More information

nag sparse nsym sol (f11dec)

nag sparse nsym sol (f11dec) f11 Sparse Linear Algebra f11dec nag sparse nsym sol (f11dec) 1. Purpose nag sparse nsym sol (f11dec) solves a real sparse nonsymmetric system of linear equations, represented in coordinate storage format,

More information

C05 Roots of One or More Transcendental Equations. C05PCF NAG Fortran Library Routine Document

C05 Roots of One or More Transcendental Equations. C05PCF NAG Fortran Library Routine Document NAG Fortran Library Routine Document Note. Before using this routine, please read the Users Note for your implementation to check the interpretation of bold italicised terms and other implementation-dependent

More information

MA41 SUBROUTINE LIBRARY SPECIFICATION 1st Aug. 2000

MA41 SUBROUTINE LIBRARY SPECIFICATION 1st Aug. 2000 MA41 SUBROUTINE LIBRARY SPECIFICATION 1st Aug. 2000 1 SUMMARY To solve a sparse unsymmetric system of linear equations. Given a square unsymmetric sparse matrix T A of order n and a matrix B of nrhs right

More information

PACKAGE SPECIFICATION HSL 2013

PACKAGE SPECIFICATION HSL 2013 PACKAGE SPECIFICATION HSL 2013 1 SUMMARY Given a rank-one or rank-two allocatable array, reallocates the array to have a different size, and can copy all or part of the original array into the new array.

More information

F01BSF NAG Fortran Library Routine Document

F01BSF NAG Fortran Library Routine Document F01 Matrix Factorizations F01BSF NAG Fortran Library Routine Document Note. Before using this routine, please read the Users Note for your implementation to check the interpretation of bold italicised

More information

Nonsymmetric Problems. Abstract. The eect of a threshold variant TPABLO of the permutation

Nonsymmetric Problems. Abstract. The eect of a threshold variant TPABLO of the permutation Threshold Ordering for Preconditioning Nonsymmetric Problems Michele Benzi 1, Hwajeong Choi 2, Daniel B. Szyld 2? 1 CERFACS, 42 Ave. G. Coriolis, 31057 Toulouse Cedex, France (benzi@cerfacs.fr) 2 Department

More information

NAG Fortran Library Routine Document C05PBF.1

NAG Fortran Library Routine Document C05PBF.1 C05 Roots of One or More Transcendental Equations NAG Fortran Library Routine Document Note: before using this routine, please read the Users Note for your implementation to check the interpretation of

More information

E04JAF NAG Fortran Library Routine Document

E04JAF NAG Fortran Library Routine Document E04JAF NAG Fortran Library Routine Document Note. Before using this routine, please read the Users Note for your implementation to check the interpretation of bold italicised terms and other implementation-dependent

More information

Accelerating a Simulation of Type I X ray Bursts from Accreting Neutron Stars Mark Mackey Professor Alexander Heger

Accelerating a Simulation of Type I X ray Bursts from Accreting Neutron Stars Mark Mackey Professor Alexander Heger Accelerating a Simulation of Type I X ray Bursts from Accreting Neutron Stars Mark Mackey Professor Alexander Heger The goal of my project was to develop an optimized linear system solver to shorten the

More information

NAG Fortran Library Routine Document F08FQF (CHEEVD=ZHEEVD).1

NAG Fortran Library Routine Document F08FQF (CHEEVD=ZHEEVD).1 NAG Fortran Library Routine Document Note: before using this routine, please read the Users Note for your implementation to check the interpretation of bold italicised terms and other implementation-dependent

More information

E04JYF NAG Fortran Library Routine Document

E04JYF NAG Fortran Library Routine Document E04JYF NAG Fortran Library Routine Document Note. Before using this routine, please read the Users Note for your implementation to check the interpretation of bold italicised terms and other implementation-dependent

More information

PACKAGE SPECIFICATION 21 I S 22 P = I

PACKAGE SPECIFICATION 21 I S 22 P = I PACKAGE SPECFCATON 1 SUMMARY For a full matrix that is real symmetric, complex Hermitian, or complex symmetric, this module performs partial or full factorizations and performs solutions of corresponding

More information

HARWELL SUBROUTINE LIBRARY SPECIFICATION Release 13 (1998)

HARWELL SUBROUTINE LIBRARY SPECIFICATION Release 13 (1998) HSL HARWELL SUBROUINE LIBRARY SPECIFICAION Release 13 (1998) 1 SUMMARY o compute an orthogonal factorization of a sparse overdetermined matrix A and optionally to solve the least squares problem min b

More information

E04DGF NAG Fortran Library Routine Document

E04DGF NAG Fortran Library Routine Document E04 Minimizing or Maximizing a Function E04DGF NAG Fortran Library Routine Document Note. Before using this routine, please read the Users Note for your implementation to check the interpretation of bold

More information

NAG Library Routine Document C05RBF.1

NAG Library Routine Document C05RBF.1 C05 Roots of One or More Transcendental Equations NAG Library Routine Document Note: before using this routine, please read the Users Note for your implementation to check the interpretation of bold italicised

More information

NAG Library Routine Document C05QBF.1

NAG Library Routine Document C05QBF.1 NAG Library Routine Document Note: before using this routine, please read the Users Note for your implementation to check the interpretation of bold italicised terms and other implementation-dependent

More information

NAG Library Routine Document C05PBF/C05PBA

NAG Library Routine Document C05PBF/C05PBA NAG Library Routine Document /C05PBA Note: before using this routine, please read the Users Note for your implementation to check the interpretation of bold italicised terms and other implementation-dependent

More information

HSL MC64 1 SUMMARY 2 HOW TO USE THE PACKAGE. All use is subject to licence. 1 C INTERFACE HSL 2013

HSL MC64 1 SUMMARY 2 HOW TO USE THE PACKAGE. All use is subject to licence.   1 C INTERFACE HSL 2013 HSL MC64 C INTERFACE HSL 2013 1 SUMMARY Given a sparse matrix A = {a i j m n, m n, HSL MC64 attempts to find row and column permutation vectors that make the permuted matrix have n entries on its diagonal.

More information

NAG Library Function Document nag_sparse_sym_sol (f11jec)

NAG Library Function Document nag_sparse_sym_sol (f11jec) f11 Large Scale Linear Systems f11jec NAG Library Function Document nag_sparse_sym_sol (f11jec) 1 Purpose nag_sparse_sym_sol (f11jec) solves a real sparse symmetric system of linear equations, represented

More information

NAG Fortran Library Routine Document F08KAF (DGELSS).1

NAG Fortran Library Routine Document F08KAF (DGELSS).1 NAG Fortran Library Routine Document Note: before using this routine, please read the Users Note for your implementation to check the interpretation of bold italicised terms and other implementation-dependent

More information

HSL MA79 1 SUMMARY 2 HOW TO USE THE PACKAGE. All use is subject to licence. 1 PACKAGE SPECIFICATION HSL 2013

HSL MA79 1 SUMMARY 2 HOW TO USE THE PACKAGE. All use is subject to licence.  1 PACKAGE SPECIFICATION HSL 2013 HSL MA79 PACKAGE SPECIFICATION HSL 2013 1 SUMMARY HSL MA79 is a mixed precision sparse symmetric solver for solving one or more linear systems AX = B. A factorization of A using single precision (that

More information

Porting the NAS-NPB Conjugate Gradient Benchmark to CUDA. NVIDIA Corporation

Porting the NAS-NPB Conjugate Gradient Benchmark to CUDA. NVIDIA Corporation Porting the NAS-NPB Conjugate Gradient Benchmark to CUDA NVIDIA Corporation Outline! Overview of CG benchmark! Overview of CUDA Libraries! CUSPARSE! CUBLAS! Porting Sequence! Algorithm Analysis! Data/Code

More information

ME38. All use is subject to licence. ME38 v Documentation date: 15th April SUMMARY 2 HOW TO USE THE PACKAGE

ME38. All use is subject to licence. ME38 v Documentation date: 15th April SUMMARY 2 HOW TO USE THE PACKAGE ME38 PACKAGE SPECIFICAION HSL 2013 1 SUMMARY his package solves a sparse complex unsymmetric system of n linear equations in n unknowns using an unsymmetric multifrontal variant of Gaussian elimination.

More information

PARDISO Version Reference Sheet Fortran

PARDISO Version Reference Sheet Fortran PARDISO Version 5.0.0 1 Reference Sheet Fortran CALL PARDISO(PT, MAXFCT, MNUM, MTYPE, PHASE, N, A, IA, JA, 1 PERM, NRHS, IPARM, MSGLVL, B, X, ERROR, DPARM) 1 Please note that this version differs significantly

More information

HSL MA97 1 SUMMARY 2 HOW TO USE THE PACKAGE. All use is subject to licence. 1 C INTERFACE

HSL MA97 1 SUMMARY 2 HOW TO USE THE PACKAGE. All use is subject to licence.  1 C INTERFACE HSL MA97 C INTERFACE HSL 1 SUMMARY HSL MA97 solves one or more sets of n n sparse symmetric equations AX = B using a multifrontal method. The package covers the following cases: 1. A is positive definite.

More information

Libraries for Scientific Computing: an overview and introduction to HSL

Libraries for Scientific Computing: an overview and introduction to HSL Libraries for Scientific Computing: an overview and introduction to HSL Mario Arioli Jonathan Hogg STFC Rutherford Appleton Laboratory 2 / 41 Overview of talk Brief introduction to who we are An overview

More information

NAG Library Function Document nag_sparse_nherm_jacobi (f11dxc)

NAG Library Function Document nag_sparse_nherm_jacobi (f11dxc) f11 Large Scale Linear Systems NAG Library Function Document nag_sparse_nherm_jacobi () 1 Purpose nag_sparse_nherm_jacobi () computes the approximate solution of a complex, Hermitian or non- Hermitian,

More information

Lecture 15: More Iterative Ideas

Lecture 15: More Iterative Ideas Lecture 15: More Iterative Ideas David Bindel 15 Mar 2010 Logistics HW 2 due! Some notes on HW 2. Where we are / where we re going More iterative ideas. Intro to HW 3. More HW 2 notes See solution code!

More information

NAG Library Routine Document F08FDF (DSYEVR)

NAG Library Routine Document F08FDF (DSYEVR) NAG Library Routine Document (DSYEVR) Note: before using this routine, please read the Users' Note for your implementation to check the interpretation of bold italicised terms and other implementation-dependent

More information

Report of Linear Solver Implementation on GPU

Report of Linear Solver Implementation on GPU Report of Linear Solver Implementation on GPU XIANG LI Abstract As the development of technology and the linear equation solver is used in many aspects such as smart grid, aviation and chemical engineering,

More information

NAG Fortran Library Routine Document E01AEF.1

NAG Fortran Library Routine Document E01AEF.1 E01 Interpolation NAG Fortran Library Routine Document Note: before using this routine, please read the Users Note for your implementation to check the interpretation of bold italicised terms and other

More information

Sparse Direct Solvers for Extreme-Scale Computing

Sparse Direct Solvers for Extreme-Scale Computing Sparse Direct Solvers for Extreme-Scale Computing Iain Duff Joint work with Florent Lopez and Jonathan Hogg STFC Rutherford Appleton Laboratory SIAM Conference on Computational Science and Engineering

More information

NAG Library Function Document nag_real_symm_sparse_eigensystem_sol (f12fcc)

NAG Library Function Document nag_real_symm_sparse_eigensystem_sol (f12fcc) NAG Library Function Document nag_real_symm_sparse_eigensystem_sol () Note: this function uses optional arguments to define choices in the problem specification. If you wish to use default settings for

More information

NAG C Library Function Document. nag_real_sparse_eigensystem_iter (f12abc),

NAG C Library Function Document. nag_real_sparse_eigensystem_iter (f12abc), f12 Large Scale Eigenproblems f12aac 1 Purpose NAG C Library Function Document nag_real_sparse_eigensystem_init (f12aac) nag_real_sparse_eigensystem_init (f12aac) is a setup function in a suite of functions

More information

3.3 Function minimization

3.3 Function minimization 3.3. Function minimization 55 3.3 Function minimization Beneath the problem of root-finding, minimizing functions constitutes a major problem in computational economics. Let f () : X R a function that

More information

AN UNCONSTRAINED NONLINEAR OPTIMIZATION SOLVER: A USER S GUIDE

AN UNCONSTRAINED NONLINEAR OPTIMIZATION SOLVER: A USER S GUIDE AN UNONSTRAINED NONLINEAR OPTIMIZATION SOLVER: A USER S GUIDE 1. Introduction This modular software package has been designed to solve the unconstrained nonlinear optimization problem. This problem entails

More information

NAG Library Function Document nag_zero_nonlin_eqns_rcomm (c05qdc)

NAG Library Function Document nag_zero_nonlin_eqns_rcomm (c05qdc) c05 Roots of One or More Transcendental Equations c05qdc NAG Library Function Document nag_zero_nonlin_eqns_rcomm (c05qdc) 1 Purpose nag_zero_nonlin_eqns_rcomm (c05qdc) is a comprehensive reverse communication

More information

Analysis of the GCR method with mixed precision arithmetic using QuPAT

Analysis of the GCR method with mixed precision arithmetic using QuPAT Analysis of the GCR method with mixed precision arithmetic using QuPAT Tsubasa Saito a,, Emiko Ishiwata b, Hidehiko Hasegawa c a Graduate School of Science, Tokyo University of Science, 1-3 Kagurazaka,

More information

An Asynchronous Algorithm on NetSolve Global Computing System

An Asynchronous Algorithm on NetSolve Global Computing System An Asynchronous Algorithm on NetSolve Global Computing System Nahid Emad S. A. Shahzadeh Fazeli Jack Dongarra Abstract The Explicitly Restarted Arnoldi Method (ERAM) allows to find a few eigenpairs of

More information

NAG Fortran Library Routine Document F04BGF.1

NAG Fortran Library Routine Document F04BGF.1 NAG Fortran Library Routine Document Note: before using this routine, please read the Users Note for your implementation to check the interpretation of bold italicised terms and other implementation-dependent

More information

NAG Library Function Document nag_real_symm_sparse_eigensystem_init (f12fac)

NAG Library Function Document nag_real_symm_sparse_eigensystem_init (f12fac) 1 Purpose NAG Library Function Document nag_real_symm_sparse_eigensystem_init () nag_real_symm_sparse_eigensystem_init () is a setup function in a suite of functions consisting of nag_real_symm_sparse_eigensystem_init

More information

BDDCML. solver library based on Multi-Level Balancing Domain Decomposition by Constraints copyright (C) Jakub Šístek version 1.

BDDCML. solver library based on Multi-Level Balancing Domain Decomposition by Constraints copyright (C) Jakub Šístek version 1. BDDCML solver library based on Multi-Level Balancing Domain Decomposition by Constraints copyright (C) 2010-2012 Jakub Šístek version 1.3 Jakub Šístek i Table of Contents 1 Introduction.....................................

More information

ICCGLU. A Fortran IV subroutine to solve large sparse general systems of linear equations. J.J. Dongarra, G.K. Leaf and M. Minkoff.

ICCGLU. A Fortran IV subroutine to solve large sparse general systems of linear equations. J.J. Dongarra, G.K. Leaf and M. Minkoff. http://www.netlib.org/linalg/ig-do 1 of 8 12/7/2009 11:14 AM ICCGLU A Fortran IV subroutine to solve large sparse general systems of linear equations. J.J. Dongarra, G.K. Leaf and M. Minkoff July, 1982

More information

F02WUF NAG Fortran Library Routine Document

F02WUF NAG Fortran Library Routine Document F02 Eigenvalues and Eigenvectors F02WUF NAG Fortran Library Routine Document Note. Before using this routine, please read the Users Note for your implementation to check the interpretation of bold italicised

More information

NAG Library Routine Document F08FAF (DSYEV)

NAG Library Routine Document F08FAF (DSYEV) NAG Library Routine Document (DSYEV) Note: before using this routine, please read the Users' Note for your implementation to check the interpretation of bold italicised terms and other implementation-dependent

More information

All use is subject to licence. See For any commercial application, a separate license must be signed.

All use is subject to licence. See   For any commercial application, a separate license must be signed. GALAHAD SLS USER DOCUMENTATION GALAHAD Optimization Library version 3.0 1 SUMMARY This package solves dense or sparse symmetric systems of linear equations using variants of Gaussian elimination. Given

More information

An Out-of-Core Sparse Cholesky Solver

An Out-of-Core Sparse Cholesky Solver An Out-of-Core Sparse Cholesky Solver JOHN K. REID and JENNIFER A. SCOTT Rutherford Appleton Laboratory 9 Direct methods for solving large sparse linear systems of equations are popular because of their

More information

Technical Report TR

Technical Report TR Technical Report TR-2012-04 SPIKE::GPU - A GPU-based Banded Linear System Solver Ang Li, Andrew Seidl, Dan Negrut November 15, 2012 Abstract The SPIKE algorithm [1, 2] is an efficient generic divide-and-conquer

More information

Lab # 2 - ACS I Part I - DATA COMPRESSION in IMAGE PROCESSING using SVD

Lab # 2 - ACS I Part I - DATA COMPRESSION in IMAGE PROCESSING using SVD Lab # 2 - ACS I Part I - DATA COMPRESSION in IMAGE PROCESSING using SVD Goals. The goal of the first part of this lab is to demonstrate how the SVD can be used to remove redundancies in data; in this example

More information

NAG Library Routine Document F08LEF (DGBBRD)

NAG Library Routine Document F08LEF (DGBBRD) NAG Library Routine Document (DGBBRD) Note: before using this routine, please read the Users Note for your implementation to check the interpretation of bold italicised terms and other implementation-dependent

More information

NAG Library Routine Document E04GYF.1

NAG Library Routine Document E04GYF.1 NAG Library Routine Document Note: before using this routine, please read the Users Note for your implementation to check the interpretation of bold italicised terms and other implementation-dependent

More information

arxiv: v1 [math.oc] 30 Jan 2013

arxiv: v1 [math.oc] 30 Jan 2013 On the effects of scaling on the performance of Ipopt J. D. Hogg and J. A. Scott Scientific Computing Department, STFC Rutherford Appleton Laboratory, Harwell Oxford, Didcot OX11 QX, UK arxiv:131.7283v1

More information

NAG Library Function Document nag_zgelss (f08knc)

NAG Library Function Document nag_zgelss (f08knc) NAG Library Function Document nag_zgelss () 1 Purpose nag_zgelss () computes the minimum norm solution to a complex linear least squares problem minkb Axk 2 : 2 Specification x #include #include

More information

Iterative Algorithms I: Elementary Iterative Methods and the Conjugate Gradient Algorithms

Iterative Algorithms I: Elementary Iterative Methods and the Conjugate Gradient Algorithms Iterative Algorithms I: Elementary Iterative Methods and the Conjugate Gradient Algorithms By:- Nitin Kamra Indian Institute of Technology, Delhi Advisor:- Prof. Ulrich Reude 1. Introduction to Linear

More information

MA36. All use is subject to licence. MA36 v Documentation date: 8th February SUMMARY 2 HOW TO USE THE PACKAGE

MA36. All use is subject to licence. MA36 v Documentation date: 8th February SUMMARY 2 HOW TO USE THE PACKAGE PACKAGE SPECIFICAION 1 SUARY his subroutine solves a system of linear equations whose matrix is banded, symmetric and positive definite. It uses symmetric triangular decomposition without interchanges

More information

Intel Math Kernel Library (Intel MKL) Sparse Solvers. Alexander Kalinkin Intel MKL developer, Victor Kostin Intel MKL Dense Solvers team manager

Intel Math Kernel Library (Intel MKL) Sparse Solvers. Alexander Kalinkin Intel MKL developer, Victor Kostin Intel MKL Dense Solvers team manager Intel Math Kernel Library (Intel MKL) Sparse Solvers Alexander Kalinkin Intel MKL developer, Victor Kostin Intel MKL Dense Solvers team manager Copyright 3, Intel Corporation. All rights reserved. Sparse

More information

Lecture 17: More Fun With Sparse Matrices

Lecture 17: More Fun With Sparse Matrices Lecture 17: More Fun With Sparse Matrices David Bindel 26 Oct 2011 Logistics Thanks for info on final project ideas. HW 2 due Monday! Life lessons from HW 2? Where an error occurs may not be where you

More information

OpenFOAM on GPUs. 3rd Northern germany OpenFoam User meeting. Institute of Scientific Computing. September 24th 2015

OpenFOAM on GPUs. 3rd Northern germany OpenFoam User meeting. Institute of Scientific Computing. September 24th 2015 OpenFOAM on GPUs 3rd Northern germany OpenFoam User meeting September 24th 2015 Haus der Wissenschaften, Braunschweig Overview HPC on GPGPUs OpenFOAM on GPUs 2013 OpenFOAM on GPUs 2015 BiCGstab/IDR(s)

More information

NAG Library Function Document nag_dgelsd (f08kcc)

NAG Library Function Document nag_dgelsd (f08kcc) NAG Library Function Document nag_dgelsd () 1 Purpose nag_dgelsd () computes the minimum norm solution to a real linear least squares problem 2 Specification minkb Axk 2 : x #include #include

More information

NAG Library Routine Document D02HBF.1

NAG Library Routine Document D02HBF.1 D02 Ordinary Differential Equations NAG Library Routine Document Note: before using this routine, please read the Users Note for your implementation to check the interpretation of bold italicised terms

More information

NAG Fortran Library Routine Document F04JAF.1

NAG Fortran Library Routine Document F04JAF.1 F4 Simultaneous Linear Equations NAG Fortran Library Routine Document Note: before using this routine, please read the Users Note for your implementation to check the interpretation of bold italicised

More information

F11DCFP.1. NAG Parallel Library Routine Document

F11DCFP.1. NAG Parallel Library Routine Document F11 Sparse Linear Algebra F11DCFP NAG Parallel Library Routine Document Note: Before using this routine, please read the Users Note for your implementation to check for implementation-dependent details.

More information

Module 5.5: nag sym bnd lin sys Symmetric Banded Systems of Linear Equations. Contents

Module 5.5: nag sym bnd lin sys Symmetric Banded Systems of Linear Equations. Contents Module Contents Module 5.5: nag sym bnd lin sys Symmetric Banded Systems of nag sym bnd lin sys provides a procedure for solving real symmetric or complex Hermitian banded systems of linear equations with

More information

Numerical Linear Algebra

Numerical Linear Algebra Numerical Linear Algebra Probably the simplest kind of problem. Occurs in many contexts, often as part of larger problem. Symbolic manipulation packages can do linear algebra "analytically" (e.g. Mathematica,

More information

NAG Library Function Document nag_superlu_lu_factorize (f11mec)

NAG Library Function Document nag_superlu_lu_factorize (f11mec) NAG Library Function Document nag_superlu_lu_factorize () 1 Purpose nag_superlu_lu_factorize () computes the LU factorization of a real sparse matrix in compressed column (Harwell Boeing), column-permuted

More information

GPU Implementation of Elliptic Solvers in NWP. Numerical Weather- and Climate- Prediction

GPU Implementation of Elliptic Solvers in NWP. Numerical Weather- and Climate- Prediction 1/8 GPU Implementation of Elliptic Solvers in Numerical Weather- and Climate- Prediction Eike Hermann Müller, Robert Scheichl Department of Mathematical Sciences EHM, Xu Guo, Sinan Shi and RS: http://arxiv.org/abs/1302.7193

More information

Artificial Intelligence for Robotics: A Brief Summary

Artificial Intelligence for Robotics: A Brief Summary Artificial Intelligence for Robotics: A Brief Summary This document provides a summary of the course, Artificial Intelligence for Robotics, and highlights main concepts. Lesson 1: Localization (using Histogram

More information

NAG Library Routine Document G02DDF.1

NAG Library Routine Document G02DDF.1 G02 Correlation and Regression Analysis NAG Library Routine Document Note: before using this routine, please read the Users Note for your implementation to check the interpretation of bold italicised terms

More information

Solving Sparse Linear Systems. Forward and backward substitution for solving lower or upper triangular systems

Solving Sparse Linear Systems. Forward and backward substitution for solving lower or upper triangular systems AMSC 6 /CMSC 76 Advanced Linear Numerical Analysis Fall 7 Direct Solution of Sparse Linear Systems and Eigenproblems Dianne P. O Leary c 7 Solving Sparse Linear Systems Assumed background: Gauss elimination

More information

Efficient Multi-GPU CUDA Linear Solvers for OpenFOAM

Efficient Multi-GPU CUDA Linear Solvers for OpenFOAM Efficient Multi-GPU CUDA Linear Solvers for OpenFOAM Alexander Monakov, amonakov@ispras.ru Institute for System Programming of Russian Academy of Sciences March 20, 2013 1 / 17 Problem Statement In OpenFOAM,

More information

LAPACK. Linear Algebra PACKage. Janice Giudice David Knezevic 1

LAPACK. Linear Algebra PACKage. Janice Giudice David Knezevic 1 LAPACK Linear Algebra PACKage 1 Janice Giudice David Knezevic 1 Motivating Question Recalling from last week... Level 1 BLAS: vectors ops Level 2 BLAS: matrix-vectors ops 2 2 O( n ) flops on O( n ) data

More information

NAG Fortran Library Routine Document F04JGF.1

NAG Fortran Library Routine Document F04JGF.1 F4 Simultaneous Linear Equations NAG Fortran Library Routine Document Note: before using this routine, please read the Users Note for your implementation to check the interpretation of bold italicised

More information

Algorithm 837: AMD, An Approximate Minimum Degree Ordering Algorithm

Algorithm 837: AMD, An Approximate Minimum Degree Ordering Algorithm Algorithm 837: AMD, An Approximate Minimum Degree Ordering Algorithm PATRICK R. AMESTOY, ENSEEIHT-IRIT, and TIMOTHY A. DAVIS University of Florida and IAIN S. DUFF CERFACS and Rutherford Appleton Laboratory

More information

Figure 6.1: Truss topology optimization diagram.

Figure 6.1: Truss topology optimization diagram. 6 Implementation 6.1 Outline This chapter shows the implementation details to optimize the truss, obtained in the ground structure approach, according to the formulation presented in previous chapters.

More information

D02KAF NAG Fortran Library Routine Document

D02KAF NAG Fortran Library Routine Document D02KAF NAG Fortran Library Routine Document Note. Before using this routine, please read the Users Note for your implementation to check the interpretation of bold italicised terms and other implementation-dependent

More information

NAG Library Function Document nag_sparse_nsym_precon_ssor_solve (f11ddc)

NAG Library Function Document nag_sparse_nsym_precon_ssor_solve (f11ddc) NAG Library Function Document nag_sparse_nsym_precon_ssor_solve () 1 Purpose nag_sparse_nsym_precon_ssor_solve () solves a system of linear equations involving the preconditioning matrix corresponding

More information

NAG Fortran Library Routine Document D02KAF.1

NAG Fortran Library Routine Document D02KAF.1 NAG Fortran Library Routine Document Note: before using this routine, please read the Users Note for your implementation to check the interpretation of bold italicised terms and other implementation-dependent

More information

NAG Library Function Document nag_zero_nonlin_eqns_deriv_1 (c05ubc)

NAG Library Function Document nag_zero_nonlin_eqns_deriv_1 (c05ubc) c05 Roots of One or More Transcendental Equations c05ubc 1 Purpose NAG Library Function Document nag_zero_nonlin_eqns_deriv_1 (c05ubc) nag_zero_nonlin_eqns_deriv_1 (c05ubc) finds a solution of a system

More information

Tomonori Kouya Shizuoka Institute of Science and Technology Toyosawa, Fukuroi, Shizuoka Japan. October 5, 2018

Tomonori Kouya Shizuoka Institute of Science and Technology Toyosawa, Fukuroi, Shizuoka Japan. October 5, 2018 arxiv:1411.2377v1 [math.na] 10 Nov 2014 A Highly Efficient Implementation of Multiple Precision Sparse Matrix-Vector Multiplication and Its Application to Product-type Krylov Subspace Methods Tomonori

More information

Intel MKL Sparse Solvers. Software Solutions Group - Developer Products Division

Intel MKL Sparse Solvers. Software Solutions Group - Developer Products Division Intel MKL Sparse Solvers - Agenda Overview Direct Solvers Introduction PARDISO: main features PARDISO: advanced functionality DSS Performance data Iterative Solvers Performance Data Reference Copyright

More information

What is Multigrid? They have been extended to solve a wide variety of other problems, linear and nonlinear.

What is Multigrid? They have been extended to solve a wide variety of other problems, linear and nonlinear. AMSC 600/CMSC 760 Fall 2007 Solution of Sparse Linear Systems Multigrid, Part 1 Dianne P. O Leary c 2006, 2007 What is Multigrid? Originally, multigrid algorithms were proposed as an iterative method to

More information

Parallel Numerical Algorithms

Parallel Numerical Algorithms Parallel Numerical Algorithms Chapter 4 Sparse Linear Systems Section 4.3 Iterative Methods Michael T. Heath and Edgar Solomonik Department of Computer Science University of Illinois at Urbana-Champaign

More information

CSDP User s Guide. Brian Borchers. August 15, 2006

CSDP User s Guide. Brian Borchers. August 15, 2006 CSDP User s Guide Brian Borchers August 5, 6 Introduction CSDP is a software package for solving semidefinite programming problems. The algorithm is a predictor corrector version of the primal dual barrier

More information