For Programmers: Free Programming Magazines  


Home > Archive > Fortran > March 2004 > IBM and HP specific : Sign of eigenvector in DSYGV









You are viewing an archived Text-only version of the thread. To view this thread in it's original format and/or if you want to reply to this thread please [click here]

 

Author IBM and HP specific : Sign of eigenvector in DSYGV
Arindam Chakraborty

2004-03-28, 12:00 am

I am solving a generalized eigenvalue equation :
A x = lambda B x
where A is symmetric and B is positive defined and symmetric, x is the
eigenvector with eigenvalue lambda.

I am using DSYGV routine from ESSL (version 3.3) on IBM SP.
On HP(HP Server RX2600 nodes) I use the same routine (DSYGV)
from LAPACK.

To test the routines on the different machines, the test matrix from
ESSL
doc pages was used :
http://hpcf.nersc.gov/vendor_docs/i....html#HDRHSSYGV
EXAMPLE : 2
----------------------------------------------------------------
Test matrices stored in lower packed mode
----------------------------------------------------------------
* *
| -1.0 . . |
A = | 1.0 1.0 . |
| -1.0 -1.0 1.0 |
* *

* *
| 2.0 . . |
B = | 1.0 2.0 . |
| 0.0 1.0 2.0 |
* *
-------------------------------------------------------------------
One of the eigenvectors from the calculation of HP
had an sign difference from the ESSL results.

(The results from the ESSL calculation can be found here :
http://hpcf.nersc.gov/vendor_docs/i....html#HDRHSSYGV )

Since eigenvectors are arbitrary to with a sign difference....I was
wondering
if LAPACK (or LAPACK in HP ?) and ESSL follow different sign
conventions for
eigenvectors.
Herman D. Knoble

2004-03-29, 9:46 am

You may wish to try ACM (T.O.M.S.) Algorithm 343:
Certification of algorithm 343 [F1]:
eigenvalues and eigenvectors of a real general matrix:
http://portal.acm.org/citation.cfm?...FTOKEN=87455273

(I can send you both the original algorithm and certificaton algorithm (Fortran 77).

Skip Knoble, Penn State


On 27 Mar 2004 00:33:11 -0800, arimail77@yahoo.com (Arindam Chakraborty) wrote:

-|I am solving a generalized eigenvalue equation :
-| A x = lambda B x
-|where A is symmetric and B is positive defined and symmetric, x is the
-|eigenvector with eigenvalue lambda.
-|
-|I am using DSYGV routine from ESSL (version 3.3) on IBM SP.
-|On HP(HP Server RX2600 nodes) I use the same routine (DSYGV)
-|from LAPACK.
-|
-|To test the routines on the different machines, the test matrix from
-|ESSL
-|doc pages was used :
-| http://hpcf.nersc.gov/vendor_docs/i....html#HDRHSSYGV
-| EXAMPLE : 2
-|----------------------------------------------------------------
-|Test matrices stored in lower packed mode
-|----------------------------------------------------------------
-| * *
-| | -1.0 . . |
-|A = | 1.0 1.0 . |
-| | -1.0 -1.0 1.0 |
-| * *
-|
-| * *
-| | 2.0 . . |
-|B = | 1.0 2.0 . |
-| | 0.0 1.0 2.0 |
-| * *
-|-------------------------------------------------------------------
-|One of the eigenvectors from the calculation of HP
-|had an sign difference from the ESSL results.
-|
-|(The results from the ESSL calculation can be found here :
-| http://hpcf.nersc.gov/vendor_docs/i....html#HDRHSSYGV )
-|
-|Since eigenvectors are arbitrary to with a sign difference....I was
-|wondering
-|if LAPACK (or LAPACK in HP ?) and ESSL follow different sign
-|conventions for
-|eigenvectors.


Herman D. (Skip) Knoble, Research Associate
(a computing professional for 38 years)
Email: SkipKnobleLESS at SPAMpsu dot edu
Web: http://www.personal.psu.edu/hdk
Penn State Information Technology Services
Academic Services and Emerging Technologies
Graduate Education and Research Services
Penn State University
214C Computer Building
University Park, PA 16802-21013
Phone:+1 814 865-0818 Fax:+1 814 863-7049
Sponsored Links







Also available: Server administration forum archive | Web Design forum archive | Software forum archive | Hardware reviews archive

Copyright 2008 codecomments.com