Question about dgecon routine

Open discussion regarding features, bugs, issues, vendors, etc.

Question about dgecon routine

Postby Mr. President » Thu Apr 21, 2005 4:14 am

As the subject name suggets I have some problems running the dgecon routine to calculate the reciprocal condition number.
I make a LU-decomposition with the dgetrf-routine.
No problems there.

But in dgecon, the 5 parameter (standard ANORM) is an input estimate of the norm of the original matrix A. Dgecon then proceeds to calculate the condition number by calculating the norm of the inverse of A and multiplying this with your input.

However I can't find any routine to calculate the norm of A.
Does anybody have any suggestions.?
It would help me out a lot.
Mr. President
 
Posts: 2
Joined: Thu Apr 21, 2005 4:05 am
Location: Eindhoven

Postby ravi » Thu Apr 21, 2005 8:59 am

u can try with the follwing routine

ANORM = PDLANGE( 'I', N, N, A, 1, 1, DESCA, WORK )

ravi
ravi
 
Posts: 9
Joined: Wed Apr 13, 2005 7:08 am

Postby Julien Langou » Thu Apr 21, 2005 2:24 pm

Hello,
just one comment regarding the first post:
DGECON does not compute the reciprocal condition number but does provide an estimate for it (in the same range of idea, during the computation, it does not compute the norm of the inverse of A, it only computes an estimation of it).
To get the 1-norm (Inf, or Fro), you can use the LAPACK routine DLANGE.
Julien
Julien Langou
 
Posts: 734
Joined: Thu Dec 09, 2004 12:32 pm
Location: Denver, CO, USA

Postby Mr. President » Fri Apr 22, 2005 2:52 am

I just need to know if the matrix is well conditioned or not.
So an estimate will do just fine.

Thanks for the help
Mr. President
 
Posts: 2
Joined: Thu Apr 21, 2005 4:05 am
Location: Eindhoven


Return to User Discussion

Who is online

Users browsing this forum: Bing [Bot], Google [Bot] and 1 guest