     Next: Setting up the problem Up: An Example for a Previous: The Reverse Communication Interface

## Post Processing for Eigenvalues and Eigenvectors

If dsaupd indicates that convergence has taken place, then various steps may be taken to recover the results in a useful form. This is done through the subroutine dseupd and is illustrated in Figure 2.3. In the simple mode described in this chapter, the computed   eigenvectors returned by dseupd are normalized to have unit length (in the 2-norm).

 ```c c %-------------------------------------------% c | No fatal errors occurred. | c | Postprocess using DSEUPD. | c | | c | Computed eigenvalues may be extracted. | c | | c | Eigenvectors may be also computed now if | c | desired. (indicated by rvec = .true.) | c | | c | The routine DSEUPD is called to do this | c | postprocessing (Other modes may require | c | more complicated postprocessing than | c | mode1.) | c | | c %-------------------------------------------% c rvec = .true. c call dseupd ( rvec, 'All', select, d, v, ldv, sigma, & bmat, n, which, nev, tol, resid, ncv, v, ldv, & iparam, ipntr, workd, workl, lworkl, ierr ) c c %----------------------------------------------% c | Eigenvalues are returned in the first column | c | of the two-dimensional array D and the | c | corresponding eigenvectors are returned in | c | the first NCONV (=IPARAM(5)) columns of the | c | two-dimensional array V if requested. | c | Otherwise, an orthogonal basis for the | c | invariant subspace corresponding to the | c | eigenvalues in D is returned in V. | c %----------------------------------------------% c ```

Chao Yang
11/7/1997