On Wed, 21 Oct 2009, Derek Gaston wrote:
Just to be clear since I haven't chimed in here... all these changes
sound good to me. I think you're really the only one using the new
interface anyway...
Yeah, so far, but you're paying for half of it. ;-)
I've barely even started with the code that *only* PECOS is likely to
use. Adjoint-based parameter sensitivity and error indicators are
both surely in INL's best interests, yes? Where we really go down the
rabbit hole is for the inverse UQ acceleration; forward sensitivity
solves are just step one towards an option for full sensitivity
Hessians and/or Hessian-vector product.
That latter stuff shouldn't require many further API changes, though.
One method for full Hessians would require the user to provide
contractions with a qoi_second_derivative operator, but everything
else (including a slightly-less-efficient full Hessian method) should
be doable via the existing interface.
so I'm good with whatever you think is best!
Good to know.
---
Roy
------------------------------------------------------------------------------
Come build with us! The BlackBerry(R) Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay
ahead of the curve. Join us from November 9 - 12, 2009. Register now!
http://p.sf.net/sfu/devconference
_______________________________________________
Libmesh-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/libmesh-devel
