[petsc-dev] -dmmg_form_function_ghost

Barry Smith bsmith at mcs.anl.gov
Wed Mar 24 17:04:35 CDT 2010


   Sorry, I was so stunned when I saw that option I just automatically  
nuked it.

   I have pushed 3.1 and dev with the option again.

   Does your code support ghosted finite differences? With this option?


    Barry

On Mar 24, 2010, at 4:56 PM, Jed Brown wrote:

> I know this was an absolutely horrendous API, but this patch  
> completely
> removed the option which makes ex48 fail (and certainly some user  
> code).
> Releaes-3.1 either needs to get a proper API for this, or the option
> needs to remain available.  DMMGFormFunctionGhost is currently
> unreachable.
>
> (I'll add some ex48 tests to get them run in the nightlies, I should
> have done this long ago.)
>
> Jed
>
>
> changeset:   15513:e8e19296a5dc
> parent:      15511:2cefbdbb44a8
> user:        Barry Smith bsmith at mcs.anl.gov
> date:        Wed Mar 24 16:39:07 2010 -0500
> files:       src/snes/utils/damg.c src/snes/utils/damgsnes.c
> description:
> fixed bug in DMMGCreate() that it ALWAYS used ghosted coloring even  
> when it would not work.
>
>
> diff --git a/src/snes/utils/damg.c b/src/snes/utils/damg.c
> --- a/src/snes/utils/damg.c
> +++ b/src/snes/utils/damg.c
> @@ -65,7 +65,7 @@
>     p[i]->user     = user;
>     p[i]->updatejacobianperiod = 1;
>     p[i]->updatejacobian       = PETSC_TRUE;
> -    p[i]->isctype  = IS_COLORING_GHOSTED;   /* default to faster  
> version, requires DMMGSetSNESLocal() */
> +    p[i]->isctype  = IS_COLORING_GLOBAL;
>     ierr           = PetscStrallocpy(MATAIJ,&p[i]- 
> >mtype);CHKERRQ(ierr);
>   }
>   *dmmg = p;
> diff --git a/src/snes/utils/damgsnes.c b/src/snes/utils/damgsnes.c
> --- a/src/snes/utils/damgsnes.c
> +++ b/src/snes/utils/damgsnes.c
> @@ -942,16 +942,9 @@
>   CHKMEMQ;
>   ierr = PetscObjectGetCookie((PetscObject) dmmg[0]- 
> >dm,&cookie);CHKERRQ(ierr);
>   if (cookie == DM_COOKIE) {
> -    PetscTruth flag = PETSC_FALSE;
> -    /* it makes no sense to use an option to decide on ghost, it  
> depends on whether the
> -       formfunctionlocal computes ghost values in F or not. */
> -    ierr = PetscOptionsGetTruth(PETSC_NULL, "- 
> dmmg_form_function_ghost", &flag,PETSC_NULL);CHKERRQ(ierr);
> -    if (flag) {
> -      ierr =  
> DMMGSetSNES(dmmg,DMMGFormFunctionGhost,computejacobian);CHKERRQ(ierr);
> -    } else {
> -      ierr =  
> DMMGSetSNES(dmmg,DMMGFormFunction,computejacobian);CHKERRQ(ierr);
> -    }
> +    ierr =  
> DMMGSetSNES(dmmg,DMMGFormFunction,computejacobian);CHKERRQ(ierr);
>     for (i=0; i<nlevels; i++) {
> +      dmmg[i]->isctype  = IS_COLORING_GHOSTED;   /* switch to  
> faster version since have local function evaluation */
>       ierr = DASetLocalFunction((DA)dmmg[i]- 
> >dm,function);CHKERRQ(ierr);
>       dmmg[i]->lfj = (PetscErrorCode (*)(void))function;
>       ierr = DASetLocalJacobian((DA)dmmg[i]- 
> >dm,jacobian);CHKERRQ(ierr);




More information about the petsc-dev mailing list