Hello,

It has been a while since I looked at ex3... However, my first piece of
advice is that I think you should first consult the reduced basis
literature to understand the mathematics of what you're trying to
implement. If I recall correctly, one way to do what you want is to train
the model with an "impulse" force (i.e. 1 at t=0, 0 otherwise), and then
use a convolution to obtain any time-varying forcing that you want. I don't
remember the details of what is involved though, so better to check the
literature first.

In general I think libMesh's reduced basis framework should be flexible
enough to handle what you need, though you may need to over-ride certain
parts of the code (e.g. with subclasses and virtual functions) to achieve
it.

I don't have much concrete advice at this stage, though, other than that
you should make sure you understand the details of what you want to
implement, and then you'll need to study the reduced basis code in some
detail to understand what modifications may be required. I can help more
once you get to that point.

David



On Thu, Feb 1, 2018 at 10:29 PM, Xie_Jinyi <ophelie_...@qq.com> wrote:

> These days I am trying to make an adjustment of the ex3 in reduced_basis
> category, that is to replace the constant source term whose value is 1 with
> a time-dependent term, e.g 2t.
>
>
> (my idea is to simply modify the loop of unsteady mode which I think
> already exists, because the program has successfully solved an unsteady
> example like ex3, where the equation has the term du/dt.)
>
>
>
> However I haven't find out where to make my idea happen, as I don't know
> which part of the program recalculate the du/dt term at every time step.
>
>
> I wonder whether I can acheive my goal by simply adding the time-dependent
> source term in the loop or not.
>
>
> Thanks for all kinds of help.
> ------------------------------------------------------------
> ------------------
> Check out the vibrant tech community on one of the world's most
> engaging tech sites, Slashdot.org! http://sdm.link/slashdot
> _______________________________________________
> Libmesh-users mailing list
> Libmesh-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/libmesh-users
>
------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
Libmesh-users mailing list
Libmesh-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/libmesh-users

Reply via email to