From: boost (boost_at_[hidden])
Date: 2002-02-02 04:37:45
On Friday 01 February 2002 21:22, jhrwalter wrote:
> --- In boost_at_y..., Peter Schmitteckert (boost) <boost_at_s...> wrote:
> > -> What's the "canonical" ublas way to specify the start of the
> memory block ?
> > Is &E fine ?
> This could even work. In the long term I would prefer to add a
> (somewhat dangerous) accessor to the internal data pointers.
At least it does work. I tried it for zhpevd and for svd routines.
> > -> How do I get the leading dimension for matrix_ranges ?
> If I understand your question correctly, you also need access to the
> internal matrix reference of a matrix_range.
It would be sufficient to get the dimension of the underlying matrix.
> P.S.: Do you currently try to interface CLAPACK? If yes, you could
> get problems with 0 vs. 1 based indexing. Otherwise, you could get
> problems with the matrix storage layout ;-)
No, my aim is to interface to standard LaPack or to vendor specific
LA routines, which normally have similar interfaces than LaPack.
P.S. Dear Joerg,
did you reveive my email I send to your t-online account in Dezember ?
I'm curious because it seems that several mails to t-online adresses got
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk