November 2014 - Nektar-users

Error when compiling Nektar with Blas and Lapack
by Kamil ÖZDEN 03 Dec '14

03 Dec '14

Dear All, I'm trying to install Nektar++ 4.0 to a cluster with the options *NEKTAR_USE_BLAS_LAPACK ON ** ** NEKTAR_USE_SYSTEM_BLAS_LAPACK ON** * with the path of the libraries as follows: * NATIVE_BLAS /truba/sw/centos6.4/lib/blas/netlib-gcc/blas_LINUX.a ** ** NATIVE_LAPACK /truba/sw/centos6.4/lib/lapack/netlib-3.5.0-gcc/liblapack.a * However, I'm getting the following error: *Linking CXX shared library libLibUtilities.so** **/usr/bin/ld: /truba/sw/centos6.4/lib/lapack/netlib-3.5.0-gcc/liblapack.a(dgbtrf.o): relocation R_X86_64_32 against `.rodata' can not be used when making a shared object; recompile with -fPIC** **/truba/sw/centos6.4/lib/lapack/netlib-3.5.0-gcc/liblapack.a: could not read symbols: Bad value** **collect2: ld returned 1 exit status** **make[2]: *** [library/LibUtilities/libLibUtilities.so.4.0.0] Error 1** **make[1]: *** [library/LibUtilities/CMakeFiles/LibUtilities.dir/all] Error 2** **make: *** [all] Error 2* What may be the reason for this problem and how can I solve it? Regards, Kamil

5 22

Time Integration Scheme Choice
by Kamil ÖZDEN 21 Nov '14

21 Nov '14

Dear All, Which one of the time integration schemes for Incompressible Navier Stokes Solver (pg. 62 in user guide of Nektar++) should be preferred to solve a problem related with blood flow in a stenosed vessel? Regards, Kamil

2 1

FldToTecplot problem
by Kamil ÖZDEN 10 Nov '14

10 Nov '14

Dear All, When I exectute FldToVtk to get a .vtu file from a .fld/ directory it works without problem and gives me a .vtu file. But when I try to get a .dat file with FldToTecplot by typing /*./FldToTecplot Re_1000_v2_2.xml Re_1000_v2_2.fld/*/ it doesn't give a .dat file and also doesn't give any error message. What may be the reason for this problem? Regards, Kamil

2 1

Restarting from a directory
by Kamil Ozden 10 Nov '14

10 Nov '14

Dear All, I want to use the .fld directory (includes 12 .fld files as P0000000.fld etc. and one Info.xml file) of an analysis (run with 12 processors) as initial condition to another analysis. I've changed the name of the file from /*Re_1000_v2.fld*/ to /*Re_1000_v2.rst*/ and generated an initlal condition part in my .xml file as follows: /*<FUNCTION NAME="InitialConditions">*//* *//* <F VAR="u" FILE="Re_1000_v2.rst/" />*//* *//* <F VAR="v" FILE="Re_1000_v2.rst/" />*//* *//* <F VAR="w" FILE="Re_1000_v2.rst/" />*//* *//* <F VAR="p" FILE="Re_1000_v2.rst/" />*//* *//* </FUNCTION>*/ but I'm getting segmentation fault as follows: /*mpirun noticed that process rank 1 with PID 98467 on node levrek1.yonetim exited on signal 11 (Segmentation fault)*/ Am I doing something wrong about restarting by using a .fld directory? What is the way of doing this? Regards, Kamil

2 3

Combining .fld and .chk files
by Kamil ÖZDEN 08 Nov '14

08 Nov '14

Dear All, When I run an analysis with multiprocessor I got .fld and .chk files in directories as much as number of processors like /*P0000000.fld,P0000001.fld*/... etc. How can I combine these files into a single .fld and .chk file? There are some utilites in /*~/build/utilities/PostProcessing */directory like /*CombinePartitions, FldAddFld etc.*/ but as far as I see there are no information in user guide about how to use these. Regards, Kamil

2 1

History Point problem
by Kamil ÖZDEN 08 Nov '14

08 Nov '14

Dear All, When I try to run an analysis by requesting a History File, Nektar++ 4.0 tells me that: /*History point 0,0,0 cannot be found in the mesh.*/ but in the .xml file 0, 0, 0 point exists as vertex as you see below: /* <V ID="9761">0.00000000e+00 0.00000000e+00 0.00000000e+00</V>*/ When I remove history points part from the .xml file, it runs without problem. What may be the reason for this problem? NOTE: The mesh is composed of both pyrisms and hexahedras. So .xml file includes two composites. Regards, Kamil.

2 2

Iterative Solvers for Incompressible Navier Stokes Equation
by Kamil ÖZDEN 05 Nov '14

05 Nov '14

Dear All, I'm solving a pipe flow problem with Incompressible Navier Stokes solver in a cluster with mpirun command. I saw that only IterativeFull solver reduces the run time of analysis when number of CPU is increased. Is this expected in Nektar? Regards, Kamil

2 1