problem with poisson solver on sphere

keweive — Thu, 23 Oct 2008 13:36:42 GMT

Today I read the mannul how to use the poisson solver on a sphere and test this example Fortran program . however I found something confused:

I test this Program /mkl/10.0.1.014/examples/pdepoissonf/source/d_sph_with_poles_f.f90

before I change anything, it tests OK, so I think it should be OK for poisson problem

but when I set q=0.d0 for the Poisson problem and I get this msg:

MKL POISSON LIBRARY WARNING:
The problem is degenerate up to rounding errors! The approximate solution
that provides the minimal Eucledean norm of the residual will be computed...
Double precision Helmholtz example on a whole sphere FAILED to compute the solu
tion...

Any ideas are much appreciated!

Re: problem with poisson solver on sphere

Sergey_G_Intel — Mon, 27 Oct 2008 06:32:36 GMT

Hi,

Thank you for using Intel MKL!

The best way to solve your problem is to check for errors onlyif q=0.0. To do this, change status checking from

if (stat.ne.0) goto 999

if (stat.le.0) goto 999

after Poisson Library routines. It should be enough to change this line only after commit and computational routines. With this setting, theexample will proceed with computations despite of the warning produced and will compute the solution with minimal Euclidean norm.

Please let me know if the proposed solution works fine for you. Thank you!

With kind regards,

Sergey

topic Re: problem with poisson solver on sphere in Intel® oneAPI Math Kernel Library

problem with poisson solver on sphere

Re: problem with poisson solver on sphere