hide long namesshow long names
hide short namesshow short names
Integer type:  int32  int64  nag_int  show int32  show int32  show int64  show int64  show nag_int  show nag_int

PDF version (NAG web site, 64-bit version, 64-bit version)
Chapter Contents
Chapter Introduction
NAG Toolbox

NAG Toolbox: nag_sparseig_complex_iter (f12ap)

Purpose

nag_sparseig_complex_iter (f12ap) is an iterative solver in a suite of functions consisting of nag_sparseig_complex_init (f12an), nag_sparseig_complex_iter (f12ap), nag_sparseig_complex_proc (f12aq), nag_sparseig_complex_option (f12ar) and nag_sparseig_complex_monit (f12as). It is used to find some of the eigenvalues (and optionally the corresponding eigenvectors) of a standard or generalized eigenvalue problem defined by complex nonsymmetric matrices.

Syntax

[irevcm, resid, v, x, mx, nshift, comm, icomm, ifail] = f12ap(irevcm, resid, v, x, mx, comm, icomm)
[irevcm, resid, v, x, mx, nshift, comm, icomm, ifail] = nag_sparseig_complex_iter(irevcm, resid, v, x, mx, comm, icomm)

Description

The suite of functions is designed to calculate some of the eigenvalues, λ λ , (and optionally the corresponding eigenvectors, x x ) of a standard eigenvalue problem Ax = λx Ax = λx , or of a generalized eigenvalue problem Ax = λBx Ax = λBx  of order n n , where n n  is large and the coefficient matrices A A  and B B  are sparse, complex and nonsymmetric. The suite can also be used to find selected eigenvalues/eigenvectors of smaller scale dense, complex and nonsymmetric problems.
nag_sparseig_complex_iter (f12ap) is a reverse communication function, based on the ARPACK routine znaupd, using the Implicitly Restarted Arnoldi iteration method. The method is described in Lehoucq and Sorensen (1996) and Lehoucq (2001) while its use within the ARPACK software is described in great detail in Lehoucq et al. (1998). An evaluation of software for computing eigenvalues of sparse nonsymmetric matrices is provided in Lehoucq and Scott (1996). This suite of functions offers the same functionality as the ARPACK software for complex nonsymmetric problems, but the interface design is quite different in order to make the option setting clearer and to simplify the interface of nag_sparseig_complex_iter (f12ap).
The setup function nag_sparseig_complex_init (f12an) must be called before nag_sparseig_complex_iter (f12ap), the reverse communication iterative solver. Options may be set for nag_sparseig_complex_iter (f12ap) by prior calls to the option setting function nag_sparseig_complex_option (f12ar) and a post-processing function nag_sparseig_complex_proc (f12aq) must be called following a successful final exit from nag_sparseig_complex_iter (f12ap). nag_sparseig_complex_monit (f12as) may be called following certain flagged intermediate exits from nag_sparseig_complex_iter (f12ap) to provide additional monitoring information about the computation.
nag_sparseig_complex_iter (f12ap) uses reverse communication, i.e., it returns repeatedly to the calling program with the parameter irevcm (see Section [Parameters]) set to specified values which require the calling program to carry out one of the following tasks:
compute the matrix-vector product y = OPx y = OPx , where OP OP  is defined by the computational mode;
compute the matrix-vector product y = Bx y = Bx ;
notify the completion of the computation;
allow the calling program to monitor the solution.
The problem type to be solved (standard or generalized), the spectrum of eigenvalues of interest, the mode used (regular, regular inverse, shifted inverse, shifted real or shifted imaginary) and other options can all be set using the option setting function nag_sparseig_complex_option (f12ar) (see Section [Description of the Optional s] in (f12ar) for details on setting options and of the default settings).

References

Lehoucq R B (2001) Implicitly restarted Arnoldi methods and subspace iteration SIAM Journal on Matrix Analysis and Applications 23 551–562
Lehoucq R B and Scott J A (1996) An evaluation of software for computing eigenvalues of sparse nonsymmetric matrices Preprint MCS-P547-1195 Argonne National Laboratory
Lehoucq R B and Sorensen D C (1996) Deflation techniques for an implicitly restarted Arnoldi iteration SIAM Journal on Matrix Analysis and Applications 17 789–821
Lehoucq R B, Sorensen D C and Yang C (1998) ARPACK Users' Guide: Solution of Large-scale Eigenvalue Problems with Implicitly Restarted Arnoldi Methods SIAM, Philidelphia

Parameters

Note: this function uses reverse communication. Its use involves an initial entry, intermediate exits and re-entries, and a final exit, as indicated by the parameter irevcm. Between intermediate exits and re-entries, all parameters other than x, mx and comm must remain unchanged.

Compulsory Input Parameters

1:     irevcm – int64int32nag_int scalar
On initial entry: irevcm = 0irevcm=0, otherwise an error condition will be raised.
On intermediate re-entry: must be unchanged from its previous exit value. Changing irevcm to any other value between calls will result in an error.
Constraint: on initial entry, irevcm = 0irevcm=0; on re-entry irevcm must remain unchanged.
2:     resid( : :) – complex array
Note: the dimension of the array resid must be at least nn (see nag_sparseig_complex_init (f12an)).
On initial entry: need not be set unless the option Initial Residual has been set in a prior call to nag_sparseig_complex_option (f12ar) in which case resid should contain an initial residual vector, possibly from a previous run.
On intermediate re-entry: must be unchanged from its previous exit. Changing resid to any other value between calls may result in an error exit.
3:     v(ldv, : :) – complex array
The first dimension of the array v must be at least nn
The second dimension of the array must be at least max (1,ncv)max(1,ncv)
On initial entry: need not be set.
On intermediate re-entry: must be unchanged from its previous exit.
4:     x( : :) – complex array
Note: the dimension of the array x must be at least n n  (see nag_sparseig_complex_init (f12an)).
On initial entry: need not be set, it is used as a convenient mechanism for accessing elements of comm.
On intermediate re-entry: if Pointers = YESPointers=YES, x need not be set.
If Pointers = NOPointers=NO, x must contain the result of y = OPxy=OPx when irevcm returns the value 1-1 or + 1+1. It must return the computed shifts when irevcm returns the value 33.
5:     mx( : :) – complex array
Note: the dimension of the array mx must be at least n n  (see nag_sparseig_complex_init (f12an)).
On initial entry: need not be set, it is used as a convenient mechanism for accessing elements of comm.
On intermediate re-entry: if Pointers = YESPointers=YES, mx need not be set.
If Pointers = NOPointers=NO, mx must contain the result of y = Bxy=Bx when irevcm returns the value 22.
6:     comm( : :) – complex array
Note: the dimension of the array comm must be at least max (1,lcomm)max(1,lcomm) (see nag_sparseig_complex_init (f12an)).
On initial entry: must remain unchanged following a call to the setup function nag_sparseig_complex_init (f12an).
7:     icomm( : :) – int64int32nag_int array
Note: the dimension of the array icomm must be at least max (1,licomm)max(1,licomm) (see nag_sparseig_complex_init (f12an)).
On initial entry: must remain unchanged following a call to the setup function nag_sparseig_complex_init (f12an).

Optional Input Parameters

None.

Input Parameters Omitted from the MATLAB Interface

ldv

Output Parameters

1:     irevcm – int64int32nag_int scalar
On intermediate exit: has the following meanings.
irevcm = -1irevcm=-1
The calling program must compute the matrix-vector product y = OPxy=OPx, where xx is stored in x (by default) or in the array comm (starting from the location given by the first element of icomm) when the option Pointers = YESPointers=YES is set in a prior call to nag_sparseig_complex_option (f12ar). The result yy is returned in x (by default) or in the array comm (starting from the location given by the second element of icomm) when the option Pointers = YESPointers=YES is set.
irevcm = 1irevcm=1
The calling program must compute the matrix-vector product y = OPxy=OPx. This is similar to the case irevcm = -1irevcm=-1 except that the result of the matrix-vector product BxBx (as required in some computational modes) has already been computed and is available in mx (by default) or in the array comm (starting from the location given by the third element of icomm) when the option Pointers = YESPointers=YES is set.
irevcm = 2irevcm=2
The calling program must compute the matrix-vector product y = Bxy=Bx, where xx is stored in x and yy is returned in mx (by default) or in the array comm (starting from the location given by the second element of icomm) when the option Pointers = YESPointers=YES is set.
irevcm = 3irevcm=3
Compute the nshift complex shifts. This value of irevcm will only arise if the optional parameter Supplied Shifts is set in a prior call to nag_sparseig_complex_option (f12ar) which is intended for experienced users only; the default and recommended option is to use exact shifts (see Lehoucq et al. (1998) for details).
irevcm = 4irevcm=4
Monitoring step: a call to nag_sparseig_complex_monit (f12as) can now be made to return the number of Arnoldi iterations, the number of converged Ritz values, the array of converged values, and the corresponding Ritz estimates.
On final exit: irevcm = 5irevcm=5: nag_sparseig_complex_iter (f12ap) has completed its tasks. The value of ifail determines whether the iteration has been successfully completed, or whether errors have been detected. On successful completion nag_sparseig_complex_proc (f12aq) must be called to return the requested eigenvalues and eigenvectors (and/or Schur vectors).
2:     resid( : :) – complex array
Note: the dimension of the array resid must be at least nn (see nag_sparseig_complex_init (f12an)).
On intermediate exit: contains the current residual vector.
On final exit: contains the final residual vector.
3:     v(ldv, : :) – complex array
The first dimension of the array v will be nn
The second dimension of the array will be max (1,ncv)max(1,ncv)
ldv = icomm(comm(1) . re + 1)ldv=icommcomm1.re+1.
On intermediate exit: contains the current set of Arnoldi basis vectors.
ldv = icomm(comm(1) . re + 1)ldv=icommcomm1.re+1.
On final exit: contains the final set of Arnoldi basis vectors.
4:     x( : :) – complex array
Note: the dimension of the array x must be at least n n  (see nag_sparseig_complex_init (f12an)).
On intermediate exit: if Pointers = YESPointers=YES, x is not referenced.
If Pointers = NOPointers=NO, x contains the vector xx when irevcm returns the value 1-1 or + 1+1.
On final exit: does not contain useful data.
5:     mx( : :) – complex array
Note: the dimension of the array mx must be at least n n  (see nag_sparseig_complex_init (f12an)).
On intermediate exit: if Pointers = YESPointers=YES, mx is not referenced.
If Pointers = NOPointers=NO, mx contains the vector BxBx when irevcm returns the value + 1+1.
On final exit: does not contain any useful data.
6:     nshift – int64int32nag_int scalar
On intermediate exit: if the option Supplied Shifts is set and irevcm returns a value of 33, nshift returns the number of complex shifts required.
7:     comm( : :) – complex array
Note: the dimension of the array comm must be at least max (1,lcomm)max(1,lcomm) (see nag_sparseig_complex_init (f12an)).
Contains data defining the current state of the iterative process.
8:     icomm( : :) – int64int32nag_int array
Note: the dimension of the array icomm must be at least max (1,licomm)max(1,licomm) (see nag_sparseig_complex_init (f12an)).
Contains data defining the current state of the iterative process.
9:     ifail – int64int32nag_int scalar
ifail = 0ifail=0 unless the function detects an error (see [Error Indicators and Warnings]).

Error Indicators and Warnings

Errors or warnings detected by the function:

Cases prefixed with W are classified as warnings and do not generate an error of type NAG:error_n. See nag_issue_warnings.

  ifail = 1ifail=1
On initial entry, the maximum number of iterations 0 0 , the option Iteration Limit has been set to a non-positive value.
  ifail = 2ifail=2
The options Generalized and Regular are incompatible.
  ifail = 3ifail=3
The option Initial Residual was selected but the starting vector held in resid is zero.
W ifail = 4ifail=4
The maximum number of iterations has been reached. Some Ritz values may have converged; a subsequent call to nag_sparseig_complex_proc (f12aq) will return the number of converged values and the converged values.
  ifail = 5ifail=5
No shifts could be applied during a cycle of the implicitly restarted Arnoldi iteration. One possibility is to increase the size of ncv relative to nev (see Section [Parameters] in (f12an) for details of these parameters).
  ifail = 6ifail=6
Could not build an Arnoldi factorization. Consider changing ncv or nev in the initialization function (see Section [Parameters] in (f12an) for details of these parameters).
  ifail = 7ifail=7
Unexpected error in internal call to compute eigenvalues and corresponding error bounds of the current upper Hessenberg matrix. Please contact NAG.
  ifail = 8ifail=8
Either the initialization function nag_sparseig_complex_init (f12an) has not been called prior to the first call of this function or a communication array has become corrupted.
  ifail = 9ifail=9
An unexpected error has occurred. Please contact NAG.

Accuracy

The relative accuracy of a Ritz value, λ λ , is considered acceptable if its Ritz estimate Tolerance × |λ| Tolerance × |λ| . The default Tolerance used is the machine precision given by nag_machine_precision (x02aj).

Further Comments

None.

Example

function nag_sparseig_complex_iter_example
n = int64(100);
nx = int64(10);
nev = int64(4);
ncv = int64(20);

irevcm = int64(0);
resid = complex(zeros(100,1));
v = complex(zeros(100,20));
x = complex(zeros(100,1));
mx = complex(zeros(100,1));

sigma = complex(0);


% Initialisation Step
[icomm, comm, ifail] = nag_sparseig_complex_init(n, nev, ncv);

% Solve
while (irevcm ~= 5)
  [irevcm, resid, v, x, mx, nshift, comm, icomm, ifail] = ...
    nag_sparseig_complex_iter(irevcm, resid, v, x, mx, comm, icomm);
  if (irevcm == 1 || irevcm == -1)
    x = f12_av(nx, x);
  elseif (irevcm == 4)
    [niter, nconv, ritz, rzest] = nag_sparseig_complex_monit(icomm, comm);
    fprintf('Iteration %d, No. converged = %d, norm of estimates = %16.8g\n',  ...
            niter, nconv, norm(rzest(1:double(nev)),2));
  end
end

% Post-process to compute eigenvalues/vectors
[nconv, d, z, v, comm, icomm, ifail] = ...
    nag_sparseig_complex_proc(sigma, resid, v, comm, icomm);
% sort to avoid difference in order showing as an error



function [w] = f12_av(nx, v)

  inx = double(nx); % nx is int64

  w = complex(zeros(inx*inx,1));

  h2 = double(-(inx+1)^2);

  w(1:inx) = tv(inx, v(1:inx));
  w(1:inx) = h2*v(inx+1:2*inx)+w(1:inx);

  for j=2:double(inx-1)
    lo = (j-1)*inx +1;
    hi = j*inx;

    w(lo:hi) = tv(inx, v(lo:hi));
    w(lo:hi) = h2*v(lo-inx:lo-1)+w(lo:hi);
    w(lo:hi) = h2*v(hi+1:hi+inx)+w(lo:hi);
  end


  lo = (inx-1)*inx +1;
  hi = inx*inx;
  w(lo:hi) = tv(inx, v(lo:hi));
  w(lo:hi) = h2*v(lo-inx:lo-1)+w(lo:hi);

function [y] = tv(inx,x)

  y = zeros(inx,1);

  dd = double(4*(inx+1)^2);
  dl = double(-(inx+1)^2 - 0.5*100*(inx+1));
  du = double(-(inx+1)^2 + 0.5*100*(inx+1));

  y(1) = dd*x(1) + du*x(2);
  for j=2:double(inx-1)
    y(j) = dl*x(j-1) + dd*x(j) + du*x(j+1);
  end
  y(inx) = dl*x(inx-1) + dd*x(inx);
 
Iteration 1, No. converged = 0, norm of estimates =        133.43269
Iteration 2, No. converged = 0, norm of estimates =        99.725711
Iteration 3, No. converged = 0, norm of estimates =         42.48146
Iteration 4, No. converged = 0, norm of estimates =        8.1805323
Iteration 5, No. converged = 0, norm of estimates =        1.7794793
Iteration 6, No. converged = 0, norm of estimates =       0.49794897
Iteration 7, No. converged = 0, norm of estimates =       0.12604717
Iteration 8, No. converged = 0, norm of estimates =      0.026883649
Iteration 9, No. converged = 0, norm of estimates =     0.0070182429
Iteration 10, No. converged = 0, norm of estimates =     0.0014438635
Iteration 11, No. converged = 0, norm of estimates =    0.00040007427
Iteration 12, No. converged = 0, norm of estimates =     0.0001059392
Iteration 13, No. converged = 0, norm of estimates =     2.803229e-05
Iteration 14, No. converged = 0, norm of estimates =    7.7343157e-06
Iteration 15, No. converged = 0, norm of estimates =    1.9579037e-06
Iteration 16, No. converged = 0, norm of estimates =      6.15616e-07
Iteration 17, No. converged = 0, norm of estimates =    1.2591405e-07
Iteration 18, No. converged = 0, norm of estimates =    3.7790713e-08
Iteration 19, No. converged = 2, norm of estimates =    8.3787181e-09
Iteration 20, No. converged = 2, norm of estimates =    1.8827432e-09
Iteration 21, No. converged = 2, norm of estimates =    1.6920778e-10
Iteration 22, No. converged = 2, norm of estimates =     2.299125e-11
Iteration 23, No. converged = 2, norm of estimates =     3.391976e-12
Iteration 24, No. converged = 2, norm of estimates =    5.5954988e-13

function f12ap_example
n = int64(100);
nx = int64(10);
nev = int64(4);
ncv = int64(20);

irevcm = int64(0);
resid = complex(zeros(100,1));
v = complex(zeros(100,20));
x = complex(zeros(100,1));
mx = complex(zeros(100,1));

sigma = complex(0);


% Initialisation Step
[icomm, comm, ifail] = f12an(n, nev, ncv);

% Solve
while (irevcm ~= 5)
  [irevcm, resid, v, x, mx, nshift, comm, icomm, ifail] = ...
    f12ap(irevcm, resid, v, x, mx, comm, icomm);
  if (irevcm == 1 || irevcm == -1)
    x = f12_av(nx, x);
  elseif (irevcm == 4)
    [niter, nconv, ritz, rzest] = f12as(icomm, comm);
    fprintf('Iteration %d, No. converged = %d, norm of estimates = %16.8g\n', ...
            niter, nconv, norm(rzest(1:double(nev)),2));
  end
end

% Post-process to compute eigenvalues/vectors
[nconv, d, z, v, comm, icomm, ifail] = f12aq(sigma, resid, v, comm, icomm);
% sort to avoid difference in order showing as an error



function [w] = f12_av(nx, v)

  inx = double(nx); % nx is int64

  w = complex(zeros(inx*inx,1));

  h2 = double(-(inx+1)^2);

  w(1:inx) = tv(inx, v(1:inx));
  w(1:inx) = h2*v(inx+1:2*inx)+w(1:inx);

  for j=2:double(inx-1)
    lo = (j-1)*inx +1;
    hi = j*inx;

    w(lo:hi) = tv(inx, v(lo:hi));
    w(lo:hi) = h2*v(lo-inx:lo-1)+w(lo:hi);
    w(lo:hi) = h2*v(hi+1:hi+inx)+w(lo:hi);
  end


  lo = (inx-1)*inx +1;
  hi = inx*inx;
  w(lo:hi) = tv(inx, v(lo:hi));
  w(lo:hi) = h2*v(lo-inx:lo-1)+w(lo:hi);

function [y] = tv(inx,x)

  y = zeros(inx,1);

  dd = double(4*(inx+1)^2);
  dl = double(-(inx+1)^2 - 0.5*100*(inx+1));
  du = double(-(inx+1)^2 + 0.5*100*(inx+1));

  y(1) = dd*x(1) + du*x(2);
  for j=2:double(inx-1)
    y(j) = dl*x(j-1) + dd*x(j) + du*x(j+1);
  end
  y(inx) = dl*x(inx-1) + dd*x(inx);
 
Iteration 1, No. converged = 0, norm of estimates =        133.43269
Iteration 2, No. converged = 0, norm of estimates =        99.725711
Iteration 3, No. converged = 0, norm of estimates =         42.48146
Iteration 4, No. converged = 0, norm of estimates =        8.1805323
Iteration 5, No. converged = 0, norm of estimates =        1.7794793
Iteration 6, No. converged = 0, norm of estimates =       0.49794897
Iteration 7, No. converged = 0, norm of estimates =       0.12604717
Iteration 8, No. converged = 0, norm of estimates =      0.026883649
Iteration 9, No. converged = 0, norm of estimates =     0.0070182429
Iteration 10, No. converged = 0, norm of estimates =     0.0014438635
Iteration 11, No. converged = 0, norm of estimates =    0.00040007427
Iteration 12, No. converged = 0, norm of estimates =     0.0001059392
Iteration 13, No. converged = 0, norm of estimates =     2.803229e-05
Iteration 14, No. converged = 0, norm of estimates =    7.7343157e-06
Iteration 15, No. converged = 0, norm of estimates =    1.9579037e-06
Iteration 16, No. converged = 0, norm of estimates =      6.15616e-07
Iteration 17, No. converged = 0, norm of estimates =    1.2591405e-07
Iteration 18, No. converged = 0, norm of estimates =    3.7790713e-08
Iteration 19, No. converged = 2, norm of estimates =    8.3787181e-09
Iteration 20, No. converged = 2, norm of estimates =    1.8827432e-09
Iteration 21, No. converged = 2, norm of estimates =    1.6920778e-10
Iteration 22, No. converged = 2, norm of estimates =     2.299125e-11
Iteration 23, No. converged = 2, norm of estimates =     3.391976e-12
Iteration 24, No. converged = 2, norm of estimates =    5.5954988e-13


PDF version (NAG web site, 64-bit version, 64-bit version)
Chapter Contents
Chapter Introduction
NAG Toolbox

© The Numerical Algorithms Group Ltd, Oxford, UK. 2009–2013