[petsc-dev] Segmentation faults in MatMatMult & MatTransposeMatMult

[Pierre Jolivet]

> On 16 Jan 2019, at 3:22 PM, Jed Brown <jed at jedbrown.org> wrote:
> 
> Matthew Knepley <knepley at gmail.com> writes:
> 
>> On Wed, Jan 16, 2019 at 9:01 AM Jed Brown via petsc-dev <
>> petsc-dev at mcs.anl.gov> wrote:
>> 
>>> Pierre Jolivet via petsc-dev <petsc-dev at mcs.anl.gov> writes:
>>> 
>>>> OK, I was wrong about MATAIJ, as Jed already pointed out.
>>>> What about BAIJ or Dense matrices?
>>> 
>>> BAIJ (and SBAIJ) is handled by MatXAIJSetPreallocation.
>>> 
>> 
>> Dense matrices don't need preallocation.
> 
> True, MatMPIDenseSetPreallocation is more like a *PlaceArray or
> Create*WithArray.  The caller needs to be aware of parallelism to create
> and use arrays; I guess Pierre does that but doesn't want the noise at
> the particular call site (though at least the if statements aren't
> needed).

That is exactly what I’m doing, and what I would want. But I can live with a couple of if—else statements (should this never get fixed).