[petsc-dev] composite vs. shell

[Smith, Barry F.]

  Pierre,

https://bitbucket.org/petsc/petsc/pull-requests/809/only-propagate-operators-into-inner-pcs-in/diff

Sorry for the long delay in responding.

  Barry


> On Nov 10, 2017, at 1:36 PM, Jed Brown <jed at jedbrown.org> wrote:
> 
> "Smith, Barry F." <bsmith at mcs.anl.gov> writes:
> 
>>> On Nov 9, 2017, at 11:13 PM, Jed Brown <jed at jedbrown.org> wrote:
>>> 
>>> "Smith, Barry F." <bsmith at mcs.anl.gov> writes:
>>> 
>>>> Jed,
>>>> 
>>>>   Please articulate in a bit more detail. From what I can interpolate you are saying 
>>>> 
>>>> 1) that if we only propagate the outer matrices to inner matrices that the user has not set we will get a better more intuitive interface for users
>>>> 
>>>> but 
>>>> 
>>>> 2) the whole idea of propagating in is probably flawed.
>>> 
>>> The idea of having only two matrices, Amat and Pmat, is flawed.
>> 
>>  Yes
>> 
>>> A
>>> composite preconditioner, for example, may need more.  Having users
>>> unwrap solvers to manually set matrices is ugly, but in lieu of a better
>>> way (like named auxiliary matrices that can be requested by nested
>>> preconditioners), we should honor the user's manual choices.
>> 
>>   Yes. 
>> 
>>    But you are not directly answering my question. Should we change
>>    the code to not propagate if already set?
> 
> Yes, I think so.  That is what I meant by "we should honor the user's
> manual choices" above.