On Tue, 20 Nov 2012, Jens Lohne Eftang wrote:

> http://sourceforge.net/mailarchive/message.php?msg_id=24029850
>
> in the devel list there were plans on changing the API for
> cholesky_solve and lu_solve such that the interface with two arguments
> did not change the arguments, and another interface with only one
> argument would.
>
> There seems to be only one implementation if cholesky_solve and lu_solve
> in the current libmesh version, which takes two arguments. Does it
> change the rhs input vector? (It seems to me that it doesn't but I just
> wanted to make sure).

I changed the cholesky_solve implementation recently (r6285, just a
few weeks ago) so that it shouldn't modify the rhs.  John Peterson
seems to have changed the lu_solve implementation in a similar way,
three years ago shortly after that thread.

Nobody seems to have wanted to bother to add single-argument versions
of the methods.

> Also it seems that cholesky_solve() will change my matrix (due to the
> factorization I assume).

Yes; the factorization is still in-place.

> Is it correct that subsequent solves will give the correct answer,
> but that if I need the matrix for matrix multiplication I will have
> to keep a copy?

This should be correct.
---
Roy

------------------------------------------------------------------------------
Monitor your physical, virtual and cloud infrastructure from a single
web console. Get in-depth insight into apps, servers, databases, vmware,
SAP, cloud infrastructure, etc. Download 30-day Free Trial.
Pricing starts from $795 for 25 servers or applications!
http://p.sf.net/sfu/zoho_dev2dev_nov
_______________________________________________
Libmesh-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/libmesh-users

Reply via email to