[petsc-users] Input argument out of range with MatZeroRows

[Matthew Knepley]

On Fri, Apr 15, 2022 at 7:07 PM Jennifer Ellen Fromm <jefromm at eng.ucsd.edu>
wrote:

> Thank you for your reply, with petsc4py the only error message I get is:
>
> Traceback (most recent call last):
>   File "../../../exhume-fenics-prototype/demos/exhume_poisson.py", line
> 228, in <module>
>     solveKSP(dR_b,R_b,u_p, method=LINEAR_SOLVER, PC=PRECONDITIONER,
> bfr_tol=bfr_tol_set)
>   File
> "/home/jennifer/Work/exhume/exhume-fenics-prototype/EXHUME/common.py", line
> 958, in solveKSP
>     A,b = trimNodes(A,b=b, bfr_tol=bfr_tol)
>   File
> "/home/jennifer/Work/exhume/exhume-fenics-prototype/EXHUME/common.py", line
> 314, in trimNodes
>     A.zeroRows(nz_id)
>   File "PETSc/Mat.pyx", line 993, in petsc4py.PETSc.Mat.zeroRows
> petsc4py.PETSc.Error: error code 63
>
> When I print out the matrix size and the row IDs I want to zero (for each
> processor), I get
> ((36, 72), (36, 72))
> [36, 37, 43, 45, 62, 64, 70, 71]
> ((36, 72), (36, 72))
> [0, 1, 2, 8, 9, 11, 24, 26, 27, 33, 34, 35]
>
> Is there a way I can get petsc4py to pass me the entire PETSC error
> message?
>

Hmm, maybe something is messing with the signal handler. Can you try adding

petsc4py.PETSc.Sys.popErrorHandler()

after petsc4py.init()

  Thanks,

      Matt


> On Fri, Apr 15, 2022 at 3:42 PM Barry Smith <bsmith at petsc.dev> wrote:
>
>>
>>   Yes, this sounds like the intended usage. The only expected reason this
>> would error is if one of the indices is less than 0 or greater than or
>> equal to the number of rows in the entire matrix. In the error message, it
>> should print both the index and the number of rows in the matrix. Can you
>> send the full error message, cut and paste?
>>
>>   Barry
>>
>>
>> > On Apr 15, 2022, at 5:09 PM, Jennifer Ellen Fromm <jefromm at eng.ucsd.edu>
>> wrote:
>> >
>> > Hello,
>> > I am using petsc4py, and attempting to zero rows of an mpiaij matrix
>> (and set the diagonal to 1). I loop over the owned row indices to identify
>> the rows I want to zero, then attempt to call zeroRows() with the list of
>> owned indices. When I run in parallel (mpirun --np 2) I get petsc4py error
>> 63, 'input argument out of range'. I have tried to adjust several matrix
>> options but nothing has worked. Is this a valid method for trying to zero
>> rows?
>> > Thanks,
>> > Jennifer
>>
>>

-- 
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

https://www.cse.buffalo.edu/~knepley/ <http://www.cse.buffalo.edu/~knepley/>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20220417/0861a4f3/attachment.html>