On Jun 24, 2015, at 10:03 AM, Matthew Knepley <[email protected]> wrote:

On Wed, Jun 24, 2015 at 9:48 AM, John O'Sullivan <[email protected]> wrote: Hi Matt,

I's just wondering if you'd had a chance to do the Fortran interface for SNESSetUpdate?

I don't know how to do this one since we overhauled our Fortran callbacks. Jed or Satish?

Also what would you guys recommend for interrupting SNESSolve? For example when the solution vector is outside the range for the RHS function that can be evaluated.

I assume you mean the "current solution" vector is outside of the domain of the RHS?

I think the new scheme is to return a NaN. Is that right Barry?

Yes, just put a Inf into the function vector (at any location). Then then SNES will end with a SNES_DIVERGED_FNORM_NAN. Barry

Matt

Cheers John

-- Dr John O'Sullivan Lecturer Department of Engineering Science University of Auckland, New Zealand email: jp.osullivan at auckland.ac.nz

tel: +64 (0)9 923 85353

From: [email protected] [[email protected]] on behalf of Matthew Knepley [[email protected]] Sent: Wednesday, 24 June 2015 10:30 p.m. To: Marco Zocca Cc: PETSc Subject: Re: [petsc-dev] "pure" subset of operators

On Wed, Jun 24, 2015 at 2:48 AM, Marco Zocca <[email protected]> wrote: Hello, is there an index of the PETSc operations (e.g. mathematical on Vec's and Mat's) that do NOT overwrite the operands? I understand in-place operations are more efficient, but they make it harder to reason about the program's operation.

We do not make a separate list of these.

Thanks,

Matt

Thank you in advance

-- What most experimenters take for granted before they begin their experiments is infinitely more interesting than any results to which their experiments lead. -- Norbert Wiener

-- What most experimenters take for granted before they begin their experiments is infinitely more interesting than any results to which their experiments lead. -- Norbert Wiener