В сообщении от Четверг 10 июня 2010 13:34:57 вы написали:
> Although it would be a lot of work, have you considered implementing any
> of the eigenvalue routines, like those in dgeev?  Even if it were just for
> symmetric matrices it would be a big step forward.

You are right, there is a lot of work here. But this task is not impossible. 
xGEEx (xGEEV and almost identical to it xGEES) code consists of approx. 7400 
Fortran lines (including comments and excluding similar routines). Approx. 20% 
of xGEEx is already implemented [1]. I think xGEEx will be implemented some 
day, by me or somebody else.

As for me, I'm inclining to import the generalized EV routines xGGEx (xGGEV 
and similar xGGES) first (see TODO file in mt addon). xGGEx code consists of 
approx. 9200 Fortran lines (including comments and excluding duplicated code). 
But:
- approx. 40% of xGGEx is already implemented [1];
- xGGEV will allow me to advance in yet another addon (still in development).

---
[1] Callgraphs for LAPACK eigenvalue-related routines are referenced in
http://www.jsoftware.com/jwiki/Addons/math/mt/MATRIX

-- 
WBR
Igor
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to