nag_sparse_herm_basic_setup (f11grc) (PDF version)
f11 Chapter Contents
f11 Chapter Introduction
NAG Library Manual

NAG Library Function Document

nag_sparse_herm_basic_setup (f11grc)

+ Contents

    1  Purpose
    7  Accuracy

1  Purpose

nag_sparse_herm_basic_setup (f11grc) is a setup function, the first in a suite of three functions for the iterative solution of a complex Hermitian system of simultaneous linear equations. nag_sparse_herm_basic_setup (f11grc) must be called before nag_sparse_herm_basic_solver (f11gsc), the iterative solver. The third function in the suite, nag_sparse_herm_basic_diagnostic (f11gtc), can be used to return additional information about the computation.
These three functions are suitable for the solution of large sparse complex Hermitian systems of equations.

2  Specification

#include <nag.h>
#include <nagf11.h>
void  nag_sparse_herm_basic_setup (Nag_SparseSym_Method method, Nag_SparseSym_PrecType precon, Nag_SparseSym_Bisection sigcmp, Nag_NormType norm, Nag_SparseSym_Weight weight, Integer iterm, Integer n, double tol, Integer maxitn, double anorm, double sigmax, double sigtol, Integer maxits, Integer monit, Integer *lwreq, Complex work[], Integer lwork, NagError *fail)

3  Description

The suite consisting of the functions: is designed to solve the complex Hermitian system of simultaneous linear equations Ax=b of order n, where n is large and the matrix of the coefficients A is sparse.
nag_sparse_herm_basic_setup (f11grc) is a setup function which must be called before the iterative solver nag_sparse_herm_basic_solver (f11gsc). nag_sparse_herm_basic_diagnostic (f11gtc), the third function in the suite, can be used to return additional information about the computation. Either of two methods can be used:
1. Conjugate Gradient Method (CG)
For this method (see Hestenes and Stiefel (1952)Golub and Van Loan (1996)Barrett et al. (1994) and Dias da Cunha and Hopkins (1994)), the matrix A should ideally be positive definite. The application of the Conjugate Gradient method to indefinite matrices may lead to failure or to lack of convergence.
2. Lanczos Method (SYMMLQ)
This method, based upon the algorithm SYMMLQ (see Paige and Saunders (1975) and Barrett et al. (1994)), is suitable for both positive definite and indefinite matrices. It is more robust than the Conjugate Gradient method but less efficient when A is positive definite.
Both CG and SYMMLQ methods start from the residual r0=b-Ax0, where x0 is an initial estimate for the solution (often x0=0), and generate an orthogonal basis for the Krylov subspace spanAkr0, for k=0,1,, by means of three-term recurrence relations (see Golub and Van Loan (1996)). A sequence of real symmetric tridiagonal matrices Tk is also generated. Here and in the following, the index k denotes the iteration count. The resulting real symmetric tridiagonal systems of equations are usually more easily solved than the original problem. A sequence of solution iterates xk is thus generated such that the sequence of the norms of the residuals rk converges to a required tolerance. Note that, in general, the convergence is not monotonic.
In exact arithmetic, after n iterations, this process is equivalent to an orthogonal reduction of A to real symmetric tridiagonal form, Tn=QHAQ; the solution xn would thus achieve exact convergence. In finite-precision arithmetic, cancellation and round-off errors accumulate causing loss of orthogonality. These methods must therefore be viewed as genuinely iterative methods, able to converge to a solution within a prescribed tolerance.
The orthogonal basis is not formed explicitly in either method. The basic difference between the two methods lies in the method of solution of the resulting real symmetric tridiagonal systems of equations: the conjugate gradient method is equivalent to carrying out an LDLH (Cholesky) factorization whereas the Lanczos method (SYMMLQ) uses an LQ factorization.
Faster convergence can be achieved using a preconditioner (see Golub and Van Loan (1996) and Barrett et al. (1994)). A preconditioner maps the original system of equations onto a different system, say
A-x-=b-, (1)
with, hopefully, better characteristics with respect to its speed of convergence: for example, the condition number of the matrix of the coefficients can be improved or eigenvalues in its spectrum can be made to coalesce. An orthogonal basis for the Krylov subspace spanA-kr-0, for k=0,1,, is generated and the solution proceeds as outlined above. The algorithms used are such that the solution and residual iterates of the original system are produced, not their preconditioned counterparts. Note that an unsuitable preconditioner or no preconditioning at all may result in a very slow rate, or lack, of convergence. However, preconditioning involves a trade-off between the reduction in the number of iterations required for convergence and the additional computational costs per iteration. Also, setting up a preconditioner may involve non-negligible overheads.
A preconditioner must be Hermitian and positive definite, i.e., representable by M=EEH, where M is nonsingular, and such that A-=E-1AE-HIn in (1), where In is the identity matrix of order n. Also, we can define r-=E-1r and x-=EHx. These are formal definitions, used only in the design of the algorithms; in practice, only the means to compute the matrix-vector products v=Au and to solve the preconditioning equations Mv=u are required, that is, explicit information about M, E or their inverses is not required at any stage.
The first termination criterion
rkp τ bp + Ap × xkp (2)
is available for both conjugate gradient and Lanczos (SYMMLQ) methods. In (2), p=1,​ or ​2 and τ denotes a user-specified tolerance subject to max10,nετ<1, where ε is the machine precision. Facilities are provided for the estimation of the norm of the matrix of the coefficients A1=A, when this is not known in advance, used in (2), by applying Higham's method (see Higham (1988)). Note that A2 cannot be estimated internally. This criterion uses an error bound derived from backward error analysis to ensure that the computed solution is the exact solution of a problem as close to the original as the termination tolerance requires. Termination criteria employing bounds derived from forward error analysis could be used, but any such criteria would require information about the condition number κA which is not easily obtainable.
The second termination criterion
r-k2 τ max1.0, b2 / r02 r-02 + σ1 A- × Δx-k2 (3)
is available only for the Lanczos method (SYMMLQ). In (3), σ1A-=A-2 is the largest singular value of the (preconditioned) iteration matrix A-. This termination criterion monitors the progress of the solution of the preconditioned system of equations and is less expensive to apply than criterion (2). When σ1A- is not supplied, facilities are provided for its estimation by σ1A-maxkσ1Tk. The interlacing property σ1Tk-1σ1Tk and Gerschgorin's theorem provide lower and upper bounds from which σ1Tk can be easily computed by bisection. Alternatively, the less expensive estimate σ1A-maxkTk1 can be used, where σ1A-Tk1 by Gerschgorin's theorem. Note that only order of magnitude estimates are required by the termination criterion.
Termination criterion (2) is the recommended choice, despite its (small) additional costs per iteration when using the Lanczos method (SYMMLQ). Also, if the norm of the initial estimate is much larger than the norm of the solution, that is, if x0x, a dramatic loss of significant digits could result in complete lack of convergence. The use of criterion (2) will enable the detection of such a situation, and the iteration will be restarted at a suitable point. No such restart facilities are provided for criterion (3).
Optionally, a vector w of user-specified weights can be used in the computation of the vector norms in termination criterion (2), i.e., vpw=v w p, where v w i=wi vi, for i=1,2,,n. Note that the use of weights increases the computational costs.
The sequence of calls to the functions comprising the suite is enforced: first, the setup function nag_sparse_herm_basic_setup (f11grc) must be called, followed by the solver nag_sparse_herm_basic_solver (f11gsc). nag_sparse_herm_basic_diagnostic (f11gtc) can be called either when nag_sparse_herm_basic_solver (f11gsc) is carrying out a monitoring step or after nag_sparse_herm_basic_solver (f11gsc) has completed its tasks. Incorrect sequencing will raise an error condition.

4  References

Barrett R, Berry M, Chan T F, Demmel J, Donato J, Dongarra J, Eijkhout V, Pozo R, Romine C and Van der Vorst H (1994) Templates for the Solution of Linear Systems: Building Blocks for Iterative Methods SIAM, Philadelphia
Dias da Cunha R and Hopkins T (1994) PIM 1.1 — the parallel iterative method package for systems of linear equations user's guide — Fortran 77 version Technical Report Computing Laboratory, University of Kent at Canterbury, Kent, UK
Golub G H and Van Loan C F (1996) Matrix Computations (3rd Edition) Johns Hopkins University Press, Baltimore
Hestenes M and Stiefel E (1952) Methods of conjugate gradients for solving linear systems J. Res. Nat. Bur. Stand. 49 409–436
Higham N J (1988) FORTRAN codes for estimating the one-norm of a real or complex matrix, with applications to condition estimation ACM Trans. Math. Software 14 381–396
Paige C C and Saunders M A (1975) Solution of sparse indefinite systems of linear equations SIAM J. Numer. Anal. 12 617–629

5  Arguments

1:     methodNag_SparseSym_MethodInput
On entry: the iterative method to be used.
method=Nag_SparseSym_CG
Conjugate gradient method.
method=Nag_SparseSym_SYMMLQ
Lanczos method (SYMMLQ).
Constraint: method=Nag_SparseSym_CG or Nag_SparseSym_SYMMLQ.
2:     preconNag_SparseSym_PrecTypeInput
On entry: determines whether preconditioning is used.
precon=Nag_SparseSym_NoPrec
No preconditioning.
precon=Nag_SparseSym_Prec
Preconditioning.
Constraint: precon=Nag_SparseSym_NoPrec or Nag_SparseSym_Prec.
3:     sigcmpNag_SparseSym_BisectionInput
On entry: determines whether an estimate of σ1A-=E-1AE-H2, the largest singular value of the preconditioned matrix of the coefficients, is to be computed using the bisection method on the sequence of tridiagonal matrices Tk generated during the iteration. Note that A-=A when a preconditioner is not used.
If sigmax>0.0 (see below), i.e., when σ1A- is supplied, the value of sigcmp is ignored.
sigcmp=Nag_SparseSym_Bisect
σ1A- is to be computed using the bisection method.
sigcmp=Nag_SparseSym_NoBisect
The bisection method is not used.
If the termination criterion (3) is used, requiring σ1A-, an inexpensive estimate is computed and used (see Section 3).
Suggested value: sigcmp=Nag_SparseSym_NoBisect.
Constraint: sigcmp=Nag_SparseSym_Bisect or Nag_SparseSym_NoBisect.
4:     normNag_NormTypeInput
On entry: defines the matrix and vector norm to be used in the termination criteria.
norm=Nag_OneNorm
Use the l1 norm.
norm=Nag_InfNorm
Use the l norm.
norm=Nag_TwoNorm
Use the l2 norm.
Suggested values:
  • if iterm=1, norm=Nag_InfNorm;
  • if iterm=2, norm=Nag_TwoNorm.
Constraints:
  • if iterm=1, norm=Nag_OneNorm, Nag_InfNorm or Nag_TwoNorm;
  • if iterm=2, norm=Nag_TwoNorm.
5:     weightNag_SparseSym_WeightInput
On entry: specifies whether a vector w of user-supplied weights is to be used in the vector norms used in the computation of termination criterion (2) (iterm=1): vpw=v w p, where vi w =wi vi, for i=1,2,,n. The suffix p=1,2, denotes the vector norm used, as specified by the argument norm. Note that weights cannot be used when iterm=2, i.e., when criterion (3) is used.
weight=Nag_SparseSym_Weighted
User-supplied weights are to be used and must be supplied on initial entry to nag_sparse_herm_basic_solver (f11gsc).
weight=Nag_SparseSym_UnWeighted
All weights are implicitly set equal to one. Weights do not need to be supplied on initial entry to nag_sparse_herm_basic_solver (f11gsc).
Suggested value: weight=Nag_SparseSym_UnWeighted.
Constraints:
  • if iterm=1, weight=Nag_SparseSym_Weighted or Nag_SparseSym_UnWeighted;
  • if iterm=2, weight=Nag_SparseSym_UnWeighted.
6:     itermIntegerInput
On entry: defines the termination criterion to be used.
iterm=1
Use the termination criterion defined in (2) (both conjugate gradient and Lanczos (SYMMLQ) methods).
iterm=2
Use the termination criterion defined in (3) (Lanczos method (SYMMLQ) only).
Suggested value: iterm=1.
Constraints:
  • if method=Nag_SparseSym_CG, iterm=1;
  • if method=Nag_SparseSym_SYMMLQ, iterm=1 or 2.
7:     nIntegerInput
On entry: n, the order of the matrix A.
Constraint: n>0.
8:     toldoubleInput
On entry: the tolerance τ for the termination criterion.
If tol0.0, τ=maxε,nε is used, where ε is the machine precision.
Otherwise τ=maxtol,10ε,nε is used.
Constraint: tol<1.0.
9:     maxitnIntegerInput
On entry: the maximum number of iterations.
Constraint: maxitn>0.
10:   anormdoubleInput
On entry: if anorm>0.0, the value of Ap to be used in the termination criterion (2) (iterm=1).
If anorm0.0, iterm=1 and norm=Nag_OneNorm or Nag_InfNorm, then A1=A is estimated internally by nag_sparse_herm_basic_solver (f11gsc).
If iterm=2, then anorm is not referenced.
Constraint: if iterm=1 and norm=Nag_TwoNorm, anorm>0.0.
11:   sigmaxdoubleInput
On entry: if sigmax>0.0, the value of σ1A-=E-1AE-H2.
If sigmax0.0, σ1A- is estimated by nag_sparse_herm_basic_solver (f11gsc) when either sigcmp=Nag_SparseSym_Bisect or termination criterion (3) (iterm=2) is employed, though it will be used only in the latter case.
Otherwise, sigmax is not referenced.
12:   sigtoldoubleInput
On entry: the tolerance used in assessing the convergence of the estimate of σ1A-=A-2 when the bisection method is used.
If sigtol0.0, the default value sigtol=0.01 is used. The actual value used is maxsigtol,ε.
If sigcmp=Nag_SparseSym_NoBisect or sigmax>0.0, then sigtol is not referenced.
Suggested value: sigtol=0.01 should be sufficient in most cases.
Constraint: if sigcmp=Nag_SparseSym_Bisect and sigmax0.0, sigtol<1.0.
13:   maxitsIntegerInput
On entry: the maximum iteration number k=maxits for which σ1Tk is computed by bisection (see also Section 3). If sigcmp=Nag_SparseSym_NoBisect or sigmax>0.0, then maxits is not referenced.
Suggested value: maxits=min10,n when sigtol is of the order of its default value 0.01.
Constraint: if sigcmp=Nag_SparseSym_Bisect and sigmax0.0, 1maxitsmaxitn.
14:   monitIntegerInput
On entry: if monit>0, the frequency at which a monitoring step is executed by nag_sparse_herm_basic_solver (f11gsc): the current solution and residual iterates will be returned by nag_sparse_herm_basic_solver (f11gsc) and a call to nag_sparse_herm_basic_diagnostic (f11gtc) made possible every monit iterations, starting from iteration number monit. Otherwise, no monitoring takes place. There are some additional computational costs involved in monitoring the solution and residual vectors when the Lanczos method (SYMMLQ) is used.
Constraint: monitmaxitn.
15:   lwreqInteger *Output
On exit: the minimum amount of workspace required by nag_sparse_herm_basic_solver (f11gsc). (See also Section 5 in nag_sparse_herm_basic_solver (f11gsc).)
16:   work[lwork]ComplexCommunication Array
On exit: the array work is initialized by nag_sparse_herm_basic_setup (f11grc). It must not be modified before calling the next function in the suite, namely nag_sparse_herm_basic_solver (f11gsc).
17:   lworkIntegerInput
On entry: the dimension of the array work.
Constraint: lwork120.
Note:  although the minimum value of lwork ensures the correct functioning of nag_sparse_herm_basic_setup (f11grc), a larger value is required by the other functions in the suite, namely nag_sparse_herm_basic_solver (f11gsc) and nag_sparse_herm_basic_diagnostic (f11gtc). The required value is as follows:
Method Requirements
CG lwork=120+5n+p
SYMMLQ lwork=120+6n+p
where
  • p=2×maxits+1, when an estimate of σ1A (sigmax) is computed;
  • p=0, otherwise.
18:   failNagError *Input/Output
The NAG error argument (see Section 3.6 in the Essential Introduction).

6  Error Indicators and Warnings

NE_BAD_PARAM
On entry, argument value had an illegal value.
NE_CONSTRAINT
On entry, maxits=value, maxitn=value, sigcmp=value, and sigmax=value.
Constraint: if sigcmp=Nag_SparseSym_Bisect and sigmax0.0, 1maxitsmaxitn.
On entry, norm=value, iterm=value, and anorm=value.
Constraint: if iterm=1 and norm=Nag_TwoNorm or Nag_InfNorm, anorm>0.0.
NE_ENUM_INT
On entry, iterm=value and method=value.
Constraint: if method=Nag_SparseSym_CG, iterm=1. If method=Nag_SparseSym_SYMMLQ, iterm=1 or 2.
On entry, iterm=value and norm=value.
Constraint: if iterm=2, norm=Nag_TwoNorm or Nag_InfNorm.
On entry, iterm=value and weight=value.
Constraint: if iterm=2, weight=Nag_SparseSym_UnWeighted.
NE_ENUM_REAL_2
On entry, sigcmp=value, sigtol=value, and sigmax=value.
Constraint: if sigcmp=Nag_SparseSym_Bisect and sigmax0.0, sigtol<1.0.
NE_INT
On entry, lwork=value.
Constraint: lwork120.
On entry, maxitn=value.
Constraint: maxitn>0.
On entry, n=value.
Constraint: n>0.
NE_INT_2
On entry, monit=value and maxitn=value.
Constraint: monitmaxitn.
NE_INTERNAL_ERROR
An internal error has occurred in this function. Check the function call and any array sizes. If the call is correct then please contact NAG for assistance.
NE_OUT_OF_SEQUENCE
nag_sparse_herm_basic_setup (f11grc) has been called out of sequence: either nag_sparse_herm_basic_setup (f11grc) has been called twice or nag_sparse_herm_basic_solver (f11gsc) has not terminated its current task.
NE_REAL
On entry, tol=value.
Constraint: tol<1.0.

7  Accuracy

Not applicable.

8  Parallelism and Performance

Not applicable.

9  Further Comments

When σ1A- is not supplied (sigmax0.0) but it is required, it is estimated by nag_sparse_herm_basic_solver (f11gsc) using either of the two methods described in Section 3, as specified by the argument sigcmp. In particular, if sigcmp=Nag_SparseSym_Bisect, then the computation of σ1A- is deemed to have converged when the differences between three successive values of σ1Tk differ, in a relative sense, by less than the tolerance sigtol, i.e., when
max σ 1 k - σ 1 k - 1 σ 1 k , σ 1 k - σ 1 k - 2 σ 1 k sigtol .
The computation of σ1A- is also terminated when the iteration count exceeds the maximum value allowed, i.e., kmaxits.
Bisection is increasingly expensive with increasing iteration count. A reasonably large value of sigtol, of the order of the suggested value, is recommended and an excessive value of maxits should be avoided. Under these conditions, σ1A- usually converges within very few iterations.

10  Example

This example solves a complex Hermitian system of simultaneous linear equations using the conjugate gradient method, where the matrix of the coefficients A, has a random sparsity pattern. An incomplete Cholesky preconditioner is used (nag_sparse_sym_chol_fac (f11jac) and nag_sparse_sym_precon_ichol_solve (f11jbc)).

10.1  Program Text

Program Text (f11grce.c)

10.2  Program Data

Program Data (f11grce.d)

10.3  Program Results

Program Results (f11grce.r)


nag_sparse_herm_basic_setup (f11grc) (PDF version)
f11 Chapter Contents
f11 Chapter Introduction
NAG Library Manual

© The Numerical Algorithms Group Ltd, Oxford, UK. 2014