[petsc-dev] Adding VecErrorNormWRMS()

Abhyankar, Shrirang G. abhyshr at anl.gov
Fri May 1 10:46:23 CDT 2015

I¹ll take the IS stuff out and update the TSSetTolerances() manpage to
state how one can set the error controller to work only with differential
variables. Is that ok?


-----Original Message-----
From: Emil Constantinescu <emconsta at mcs.anl.gov>
Date: Friday, May 1, 2015 at 10:42 AM
To: Jed Brown <jed at jedbrown.org>, Shri <abhyshr at anl.gov>
Cc: petsc-dev mailing list <petsc-dev at mcs.anl.gov>
Subject: Re: [petsc-dev] Adding VecErrorNormWRMS()

>On 5/1/15 10:14 AM, Jed Brown wrote:
>> "Abhyankar, Shrirang G." <abhyshr at anl.gov> writes:
>>> Thanks Jed, I didn't think about that. It certainly seems cleaner to
>>> just set the entries of vatol to infinity rather than creating index
>>> sets.
>> So can we delete the IS stuff before it goes into the upcoming release
>> and someone starts depending on it?
>I think we can remove it but we need to let the users know they need to
>set up ts_?tol accordingly - perhaps in TSSetTolerances(). I liked IS
>because it was explicit for the users, but I agree it's probably an
>unnecessary overhead. The other reason I liked it -- that probably has
>an obvious workaround which eludes me at the moment -- was that one can
>control the diff var tolerance from the command line (specify the scalar
>tol) because one cannot control individual components from the command
>line, right?

More information about the petsc-dev mailing list