Hi,
I am trying to port my recently developed CFD code to a IBM machine running
AIX but I am having some trouble building petsc-dev although I was able to
build regular petsc on the same system. I guess it has to do with C++ and
SIEVE. The error message I got was:
=================================================================================
Configuring PETSc to compile on your system
=================================================================================
TESTING: alternateConfigureLibrary from PETSc.packages.blopex
(config/PETSc/package*********************************************************************************
UNABLE to CONFIGURE with GIVEN OPTIONS (see configure.log for
details):
---------------------------------------------------------------------------------------
Sieve requires boost, and configure could not locate it. Suggest using
--download-boost=1
*********************************************************************************
The way I used to configure is:
PETSC_DIR=`pwd` config/configure.py --with-fortran=0
--with-clanguage=C++ --with-sieve=1 --download-boost=1
I have attached the configure.log file for your review. I noticed it says
something about empty header file docsonly/include/petscfix.h, which I do
find to be empty even on my Linux clusters. But the Linux systems have a
different non-empty copy at linux-gnu-cxx-opt/include after the successful
build.
Please help. Thank you very much.
Shi
--
Sincerely,
Shi Jin, Ph.D.
http://www.ualberta.ca/~sjin1/
Hi,
I am having a small problem with my CFD code: the vorticity I computed from
the code using sieve data structure has the wrong sign. I am wondering
whether this is something to be expected or some error made by myself.
Basically, I suspect that the elements may be ordered in a left handed
manner instead of the common right handed. I noticed this by getting a
detJ=-1 for a standard tetrahedral element. I also noticed in the
computeGeometry() code there is a added "-1" to the inverse Jacobian matrix
to make the first element normal direction pointing outwards. If this is the
case, then I am happy to live with the left handed system by simply flipping
the signs of the vorticity. Please advice. Thanks a lot.
--
Sincerely,
Shi Jin, Ph.D.
http://www.ualberta.ca/~sjin1/
Hi there,
Suppose I am given some point in the sieve data structure, such as a face,
edge or vertex. Is there a function that can return a list of owning
elements, faces or edges? I know there is a nCone() function that can return
a list of vertexes, edges or faces for a given element. But I wonder whether
the reverse is possible.
Thank you very much.
--
Sincerely,
Shi Jin, Ph.D.
http://www.ualberta.ca/~sjin1/
