...
URL:
http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20061204/7b3bb05d/attachment.htm
:
http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20061204/d470cc1e/attachment.htm
On 12/4/06, Ben Tay zonexo at gmail.com wrote:
Hi,
regarding qn 4 ie the requirement of
VecAssemblyBegin,VecAssemblyEnd, I used VecPlaceArray,
VecResetArray to change the values. I did not use Assembly after the 1st
time step and it works ok. So that is fine too, right?
Its alright, but
nope - PETSc uses MPI for parallelism internally - so using the
additional compiler directive is not required. It mignt work - but it
is not appropriate. [we never tested this mode]
If you use this option - you'll need to understand what exactly it
means [in this context of MPI etc.. ] - and how
G.D.
You can compile PETSc to use long double with the additional
configure/configure.py options --with-precision=longdouble
--download-c-blas-lapack
It compiles a special version of Blas/Lapack for long double.
I'm not sure this is really the best way to go.
I would suggest
I strongly suspect that long double reals will not help
you at all. It is very unlikely that your nonlinear system
fails to converge due to loss of precision. If the Jacobian
does become near singular near the solution, it often
indicates a deficiency in the formulation. Could you
indicate why you