[petsc-users] Query regarding old PETSC version

[Satish Balay]

balay at es^/sandbox/balay/tmp/petsc-2.3.2-p10 $ grep KSPTrueMonitor include/petscksp.h
EXTERN PetscErrorCode PETSCKSP_DLLEXPORT KSPTrueMonitor(KSP,PetscInt,PetscReal,void *);

https://www.mcs.anl.gov/petsc/documentation/changes/233.html
 The following function names are changed to be consistant:
KSPTrueMonitor -> KSPMonitorTrueResidualNorm


src/ksp/ksp/interface/ftn-custom/zitfuncf.c has:

void  ksptruemonitor_(KSP *ksp,PetscInt *it,PetscReal *norm,void *ctx,PetscErrorCode *ierr)
{
  *ierr = KSPTrueMonitor(*ksp,*it,*norm,ctx);
}

For PETSC_NULL_OBJECT to work - you might have to change it to:

void  ksptruemonitor_(KSP *ksp,PetscInt *it,PetscReal *norm,void *ctx,PetscErrorCode *ierr)
{
  CHKFORTRANNULLOBJECT(ctx);
  *ierr = KSPTrueMonitor(*ksp,*it,*norm,ctx);
}

Satish

On Wed, 16 Aug 2017, Satish Balay wrote:

> neither of these appear to be in petsc-2.3.2
> 
> balay at es^/sandbox/balay/tmp/petsc-2.3.2-p10 $ grep -ri true_residual .
> balay at es^/sandbox/balay/tmp/petsc-2.3.2-p10 $ grep -ri KSPTrueResididual .
> balay at es^/sandbox/balay/tmp/petsc-2.3.2-p10 $ 
> 
> Satish
> 
> 
> On Wed, 16 Aug 2017, Matthew Knepley wrote:
> 
> > On Wed, Aug 16, 2017 at 10:01 AM, Rachit Prasad <rachitp at vt.edu> wrote:
> > 
> > > Hi,
> > >
> > > Due to legacy issues, I am working with tversion 2.3.2 of PETSc in
> > > FORTRAN. I want to monitor the true residual and hence I tried to use the
> > > function KSPTrueResididual(). However, I am unable to understand what kind
> > > of pointer is the last input parameter, "*dummy" supposed to be. I could
> > > not find an example on the website since this is an older implementation.
> > > Any insight on how to proceed further will be greatly helpful.
> > >
> > 
> > That is a very long time ago, but I think you can use PETSC_NULL_OBJECT (if
> > it exists). Also
> > -ksp_monitor_true_residual may work.
> > 
> >   Thanks,
> > 
> >      Matt
> > 
> > 
> > > Regards,
> > > Rachit
> > >
> > >
> > 
> > 
> > 
> 
>