[petsc-users] Suggestions for code with dof >> 1 ?

[Christophe Ortiz]

Thanks. I got it.

For the moment I have some general questions.

- I see that in this example all terms--diffusion and reactions--are put
into RHSFunction. There is no IFunction. Is there a reason for that ?  In
which cases one should use IFunction or RHSFunction ? In ts/.../ex25.c,
diffusion terms are defined in IFunction and reaction terms in RHSFunction.
Why not here ?

- If only RHSFunction is used, how is it possible to define boundary
conditions ? For instance, a Dirichlet boundary condition, u(i=0)=3. In
ts/.../ex2.c, only RHSFunction is used and the boundary conditions are
defined through u_t,

- In this type of problem, which TS scheme is recommended ? TSARKIMEX ?

- The option TSARKIMEXSetFullyImplicit is used. Why ? Does it help ?

Christophe

CIEMAT
Laboratorio Nacional de Fusión por Confinamiento Magnético
Unidad de Materiales
Edificio 2 - Planta 0 - Despacho 28m
Avenida Complutense 40,
28040 Madrid, Spain
Tel: +34 91496 2582
Fax: +34 91346 6442

--
Q
Por favor, piense en el medio ambiente antes de imprimir este mensaje.
Please consider the environment before printing this email.


On Mon, Oct 7, 2013 at 3:23 PM, Barry Smith <bsmith at mcs.anl.gov> wrote:

>
>   I have put it into the next branch. So just download petsc from the
> bitbucket site with git and then do git checkout next
>
>    The code won't work with the current PETSc release.
>
> On Oct 7, 2013, at 8:16 AM, Christophe Ortiz <christophe.ortiz at ciemat.es>
> wrote:
>
> > Hi Barry,
> >
> > Thanks for the reply.
> >
> > I had a look at the example ex10.c. But I see it does not fully
> corresponds to what one can see in the branch. For instance, in the example
> that is found in the tutorials, there is no RHS defined, no Jacobian, just
> the IFunction.
> >
> > Could you please send me the latest version of ex10.c ? Many thanks in
> advance.
> >
> > Christophe
> >
> > CIEMAT
> > Laboratorio Nacional de Fusión por Confinamiento Magnético
> > Unidad de Materiales
> > Edificio 2 - Planta 0 - Despacho 28m
> > Avenida Complutense 40,
> > 28040 Madrid, Spain
> > Tel: +34 91496 2582
> > Fax: +34 91346 6442
> >
> > --
> > Q
> > Por favor, piense en el medio ambiente antes de imprimir este mensaje.
> > Please consider the environment before printing this email.
> >
> >
> > On Mon, Oct 7, 2013 at 2:00 PM, Barry Smith <bsmith at mcs.anl.gov> wrote:
> >
> >   Chris,
> >
> >     Take a look in the petsc branch barry/wirth-fusion-materials
> https://bitbucket.org/petsc/petsc/branch/barry%2Fwirth-fusion-materials
> > in the file
> src/ts/examples/tutorials/advection-diffusion-reaction/ex10.c   feel free
> to ask any questions.
> >
> >     Barry
> >
> > On Oct 7, 2013, at 3:03 AM, Christophe Ortiz <christophe.ortiz at ciemat.es>
> wrote:
> >
> > > Hi all
> > >
> > > I need some suggestions to design a code with PETSc.
> > >
> > > I want to solve a 1D problem composed of several diffusion equations
> and a lot of ODEs (thousands).
> > >
> > > - After discretization, the diffusion equations have terms in i-1, i
> and i+1.
> > >
> > > - The diffusion equations have additional terms due to couping with
> all the ODEs. These terms are non-linear.
> > >
> > > - The terms of the ODEs are local (only depend on node i) but are
> non-linear.
> > >
> > > Any advice to design the code ?
> > >
> > > How should I distribute the terms between IFunction and RHSFunction ?
> > >
> > > Any special attention to DMDA ? Should I declare just one DMDA with
> dof >>1 ?
> > >
> > > Many thanks in advance !
> > > Christophe
> > >
> >
> >
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20131007/a1b90380/attachment.html>