home *** CD-ROM | disk | FTP | other *** search
/ NetNews Usenet Archive 1992 #18 / NN_1992_18.iso / spool / sci / math / numanal / 2482 < prev    next >
Encoding:
Internet Message Format  |  1992-08-18  |  1.2 KB
  1. Path: sparky!uunet!haven.umd.edu!purdue!mentor.cc.purdue.edu!noose.ecn.purdue.edu!gn.ecn.purdue.edu!zika
  2. From: zika@gn.ecn.purdue.edu (Michael Zika)
  3. Newsgroups: sci.math.num-analysis
  4. Subject: Power Method on a Generalized Eigenvalue System
  5. Message-ID: <1992Aug18.173918.29872@gn.ecn.purdue.edu>
  6. Date: 18 Aug 92 17:39:18 GMT
  7. Organization: Purdue University Engineering Computer Network
  8. Lines: 23
  9.  
  10.  
  11.   I'm trying to solve an eigenvalue system that can easily
  12. be cast as a generalized eigenvalue system:
  13.  
  14.                 Ax = wBx    (dimension 1332)
  15.  
  16. EISPACK successfully found the eigenvalue specturm, now I would
  17. like to compare the EISPACK results to a power method.  I would
  18. like to avoid the numerical error introduced by inverting B and
  19. applying it to the LHS.
  20.  
  21.   Is there a power method for generalized systems?  Or perhaps a
  22. method of refining B-inverse to reduce numerical error?  I've
  23. check Golub & Van Loan and didn't get and helpful directions.
  24.  
  25.   Any suggestions would be appreciated.
  26.  
  27.  
  28. -- 
  29.                                      |
  30. Michael Zika (zika@ecn.purdue.edu)   |    Hey don't ask me "why?",
  31. Purdue University                    |    I'm still working on "how?" ! 
  32. School of Nuclear Engineering        |
  33.