Re: [deal.II] Re: Run into a cluster

['Joaquin M Valencia Bravo' via deal.II User Group]

Hi Jean-Paul,

I can already run my code in the cluster with the lapack and blas module.

Thank you for your time.
Joaquin


2017-03-29 2:52 GMT-04:00 Jean-Paul Pelteret :

> Hi Joaquin,
>
> Its been a really long time since I compiled Lapack manually and linked it
> against deal.II (at that time, before we used CMake). I've attached for you
> a file with my personal notes on how I used to compile Lapack then (circa
> 2012). I think that you've pretty much covered all of the steps listed
> there, so "point 7" is the one that I think is most useful on this
> occasion. Lapack is built as a static library, and I think (although I
> stand to be corrected) that unless you specifically point CMake/deal.II to
> a static library then, when building a shared library version of deal.II,
> it will expect that the libraries it is to link against are also shared
> libraries.
>
> So what you could do (i.e. what might be easiest right now) is convert the
> static library of Lapack into a shared one. That you built it with the
> -FPIC flag would have ensured that this is possible. If you navigate to the
> lapack library location and run a sensibly modified version of the
> following two lines
>
>> ar -x liblapack.a && gcc -shared *.o -o liblapack.so
>> -L$DIR_BASE/OpenBLAS/build/lib -lopenblas && rm *.o
>> ar -x libtmglib.a && gcc -shared *.o -o libtmglib.so
>> -L$DIR_BASE/OpenBLAS/build/lib -lopenblas && rm *.o
>
> then you should be left with both a static and shared version of these two
> libraries.
>
> You can verify that the libraries were correctly assembled by running
>
>>   ld $DIR_BASE/LAPACK/build/lib/liblapack.so
>>   ld $DIR_BASE/LAPACK/build/lib/libtmglib.so
>
> If there are streams of "undefined reference to", then something's gone
> wrong.
>
> I hope that this finally gets you going.
>
> Best,
> Jean-Paul
>
>
>> After *$ make -j8 install*, the next error I got:
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>> *[100%] Built target obj_fe.releaseLinking CXX shared library
>> ../lib/libdeal_II.so/usr/bin/ld: cannot find -llapackcollect2: error: ld
>> returned 1 exit statusmake[2]: *** [lib/libdeal_II.so.8.4.1] Error
>> 1make[1]: *** [source/CMakeFiles/deal_II.dir/all] Error 2make[1]: ***
>> Waiting for unfinished jobs/usr/bin/ld: cannot find -llapackcollect2:
>> error: ld returned 1 exit statusmake[2]: *** [lib/libdeal_II.g.so.8.4.1]
>> Error 1make[1]: *** [source/CMakeFiles/deal_II.g.dir/all] Error 2make: ***
>> [all] Error 2*
>>
>> Apparently something else I need to do in the installation.
>>
>> Please help.
>>
> --
> The deal.II project is located at http://www.dealii.org/
> For mailing list/forum options, see https://groups.google.com/d/
> forum/dealii?hl=en
> ---
> You received this message because you are subscribed to a topic in the
> Google Groups "deal.II User Group" group.
> To unsubscribe from this topic, visit https://groups.google.com/d/
> topic/dealii/O0GlSdFavp8/unsubscribe.
> To unsubscribe from this group and all its topics, send an email to
> dealii+unsubscr...@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.
>



-- 
Joaquin Valencia
Graduate Student

-- 
The deal.II project is located at http://www.dealii.org/
For mailing list/forum options, see 
https://groups.google.com/d/forum/dealii?hl=en
--- 
You received this message because you are subscribed to the Google Groups 
"deal.II User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to dealii+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Re: [deal.II] Re: Run into a cluster

['Joaquin M Valencia Bravo' via deal.II User Group]

Thanks for the help Jean-Paul.

I installed lapack by myself, following the next steps:








*$ wget http://www.netlib.org/lapack/lapack-3.3.0.tgz
 $ tar zxvf lapack-3.3.0.tgz
$ cp INSTALL/make.inc.gfortran make.inc $ make blaslib $ make lapacklib $
ls *.a librefblas.a  liblapack.a*
*$ mv libreblas.a libblas.a*

I changed into the make.inc file as follows:







*FORTRAN  = gfortran -m64 -fPICOPTS = -O2 -m64 -fPICDRVOPTS  =
$(OPTS)NOOPT= -O0 -m64 -fPICLOADER   = gfortran -m64 -fPIC LOADOPTS =#*

Then I invoked cmake into the build directory:


*$ cmake -DDEAL_II_WITH_UMFPACK=ON
-DLAPACK_DIR=/work/fjust/jvalencia/thesis/lapack-3.7.0
-DDEAL_II_WITH_LAPACK=ON
-DCMAKE_INSTALL_PREFIX=/work/fjust/jvalencia/clfem/dealii ../*

After *$ make -j8 install*, the next error I got:












*[100%] Built target obj_fe.releaseLinking CXX shared library
../lib/libdeal_II.so/usr/bin/ld: cannot find -llapackcollect2: error: ld
returned 1 exit statusmake[2]: *** [lib/libdeal_II.so.8.4.1] Error
1make[1]: *** [source/CMakeFiles/deal_II.dir/all] Error 2make[1]: ***
Waiting for unfinished jobs/usr/bin/ld: cannot find -llapackcollect2:
error: ld returned 1 exit statusmake[2]: *** [lib/libdeal_II.g.so.8.4.1]
Error 1make[1]: *** [source/CMakeFiles/deal_II.g.dir/all] Error 2make: ***
[all] Error 2*

Apparently something else I need to do in the installation.

Please help.

2017-03-26 2:19 GMT-04:00 Jean-Paul Pelteret :

> Dear Joaquin,
>
> This would appear to be an issue with the way that blas is compiled: See these
> search results
> 
> for some more info. If this were installed by a system administrator then
> you should ask them to recompile this package (and maybe lapack as
> well) with the "-fPIC" flag as has been suggested. If you installed this
> yourself then this is how it might be done
> 
> for 64-bit lapack, and I presume that similar would apply to a stock blas
>  installation.
>
> FORTRAN  = gfortran -m64 -fPIC
>> OPTS = -O2 -m64 -fPIC
>> DRVOPTS  = $(OPTS)
>> NOOPT= -O0 -m64 -fPIC
>> LOADER   = gfortran -m64 -fPIC
>>
>>
> Best,
> Jean-Paul
>
>
> On Sunday, March 26, 2017 at 1:53:22 AM UTC+1, Joaquin wrote:
>>
>> Dear Timo,
>> Thanks for your quick response. Sorry for the late answer. Another error
>> I got when I typed "make -j8 install, this is shown below:
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>> *Linking CXX shared library ../lib/libdeal_II.so/usr/bin/ld:
>> /cm/shared/apps/blas/open64/3.5.0/lib64/../lib64/libblas.a(cgbmv.o):
>> relocation R_X86_64_32S against `.rodata' can not be used when making a
>> shared object; recompile with
>> -fPIC/cm/shared/apps/blas/open64/3.5.0/lib64/../lib64/libblas.a: could not
>> read symbols: Bad valuecollect2: error: ld returned 1 exit statusmake[2]:
>> *** [lib/libdeal_II.so.8.4.1] Error 1make[1]: ***
>> [source/CMakeFiles/deal_II.dir/all] Error 2make[1]: *** Waiting for
>> unfinished jobs/usr/bin/ld:
>> /cm/shared/apps/blas/open64/3.5.0/lib64/../lib64/libblas.a(cgbmv.o):
>> relocation R_X86_64_32S against `.rodata' can not be used when making a
>> shared object; recompile with
>> -fPIC/cm/shared/apps/blas/open64/3.5.0/lib64/../lib64/libblas.a: could not
>> read symbols: Bad valuecollect2: error: ld returned 1 exit statusmake[2]:
>> *** [lib/libdeal_II.g.so.8.4.1] Error 1make[1]: ***
>> [source/CMakeFiles/deal_II.g.dir/all] Error 2make: *** [all] Error 2*Can
>> anybody help me please?.
>>
>> Thanks,
>> Joaquin
>>
>>
>> On Sunday, February 12, 2017 at 7:52:42 PM UTC-4, Joaquin wrote:
>>>
>>> Dear all:
>>>
>>> I am trying to run the tutorial programs into a cluster. For example in
>>> the case of step-3, after typing make, it sent the following error:
>>>
>>> [fjust@boqueron step-3]$ make
>>> Scanning dependencies of target step-3
>>> [100%] Building CXX object CMakeFiles/step-3.dir/step-3.cc.o
>>> cc1plus: error: invalid option argument ‘-Og’
>>> cc1plus: error: unrecognized command line option "-std=c++11"
>>> cc1plus: warning: unrecognized command line option
>>> "-Wno-unused-local-typedefs"
>>> cc1plus: warning: unrecognized command line option "-Wno-literal-suffix"
>>> make[2]: *** [CMakeFiles/step-3.dir/step-3.cc.o] Error 1
>>> make[1]: *** [CMakeFiles/step-3.dir/all] Error 2
>>> make: *** [all] Error 2
>>>
>>> Before these errors appeared, I typed the cmake command:
>>>
>>> [fjust@boqueron step-3]$ cmake 
>>> -DDEAL_II_DIR=/work/fjust/fjust/jvalencia/cfem/dealii
>>> ./
>>> -- The C compiler identification is GNU 4.4.7
>>> -- The CXX compiler identification is GNU 4.4.7
>>> -- Check for working C compiler: /usr/bin/cc
>>> -- Check for working C compiler: /usr/bin/cc -- works
>>> -- Detecting C compiler ABI info
>>> -- Detecting C compiler ABI info - done
>>> -- Check for working CXX compiler: /usr

Re: [deal.II] Re: ifem

['Joaquin M Valencia Bravo' via deal.II User Group]

Deal Luca and Jean-Paul,

Thanks for the help. I've already been able to run the code. Although it
runs with default values, this surely is related with the last lines
written by Luca in his message.

 jomivalen@Nalia ~/ifem/ans-ifem-master $ ./ifem parameters.prm
 ParameterHandler::read_input: could not open file
 for reading.
 Trying to make file  with default values for
you.

Thanks,
Joaquin

2016-12-17 11:22 GMT-04:00 Jean-Paul Pelteret :

> Dear Joaquin,
>
> As a supplement to Luca's message, with every release of deal.II there is
> the possibility that we introduce some incompatibilities that break
> existing code. This is what must have happened here - I assume that you've
> upgraded from version 8.2.1 or earlier to something newer, and the older
> (?) version of iFEM that you're using does not follow suite. When upgrading
> deal.II you should always keep this in mind, although we do our best to
> provide a warning period to change your code by marking certain functions
> and classes as deprecated. You can see the list of incompatibilities
> between 8.2.1 and the latest release here
> 
>  and
> here
> ,
> with a more detailed explanation on the use of the Point and Tensor class
> given here .
>
> Regards,
> Jean-Paul
>
> --
> The deal.II project is located at http://www.dealii.org/
> For mailing list/forum options, see https://groups.google.com/d/
> forum/dealii?hl=en
> ---
> You received this message because you are subscribed to a topic in the
> Google Groups "deal.II User Group" group.
> To unsubscribe from this topic, visit https://groups.google.com/d/
> topic/dealii/IL4Ls7CxCmc/unsubscribe.
> To unsubscribe from this group and all its topics, send an email to
> dealii+unsubscr...@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.
>



-- 
Joaquin Valencia
Graduate Student

-- 
The deal.II project is located at http://www.dealii.org/
For mailing list/forum options, see 
https://groups.google.com/d/forum/dealii?hl=en
--- 
You received this message because you are subscribed to the Google Groups 
"deal.II User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to dealii+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Re: [deal.II] Re: ifem

['Joaquin M Valencia Bravo' via deal.II User Group]

Dear Jean-Paul,

Thanks so much for your explanation, I used the second option you give me
according to the original line code. I also removed other warnings similar
to it.

Please, help me with another problem presented in the next line code:

   *p = fe_v_s.quadrature_point(qs) - par.ring_center;*

The error it sends me is the following:

jomivalen@Nalia ~/ifem/ans-ifem $ make
[ 25%] Building CXX object CMakeFiles/ifem.dir/source/immersed_fem.cc.o
/home/jomivalen/ifem/ans-ifem/source/immersed_fem.cc: In instantiation of
‘void ImmersedFEM::get_Pe_F_and_DPeFT_dxi_values(const
dealii::FEValues&, const std::vector&, const
dealii::Vector&, bool, std::vector >&,
std::vector >&,
std::vector > >&) [with int dim = 2]’:
/home/jomivalen/ifem/ans-ifem/source/immersed_fem.cc:2046:16:   required
from here
/home/jomivalen/ifem/ans-ifem/source/immersed_fem.cc:1742:5: error: no
match for ‘operator=’ (operand types are ‘dealii::Point<2, double>’ and
‘dealii::Tensor<1, 2, double>’)
   p = fe_v_s.quadrature_point(qs) - par.ring_center; // p =
fe_v_s.quadrature_point(qs) - par.ring_center;
 ^
/home/jomivalen/ifem/ans-ifem/source/immersed_fem.cc:1742:5: note:
candidates are:
In file included from
/home/jomivalen/cfem/dealii/include/deal.II/grid/tria.h:21:0,
 from
/home/jomivalen/ifem/ans-ifem/include/immersed_fem.h:19,
 from
/home/jomivalen/ifem/ans-ifem/source/immersed_fem.cc:16:
/home/jomivalen/cfem/dealii/include/deal.II/base/point.h:89:7: note:
dealii::Point<2, double>& dealii::Point<2, double>::operator=(const
dealii::Point<2, double>&)
 class Point : public Tensor<1,dim,Number>
   ^
/home/jomivalen/cfem/dealii/include/deal.II/base/point.h:89:7: note:   no
known conversion for argument 1 from ‘dealii::Tensor<1, 2, double>’ to
‘const dealii::Point<2, double>&’
/home/jomivalen/cfem/dealii/include/deal.II/base/point.h:89:7: note:
dealii::Point<2, double>& dealii::Point<2,
double>::operator=(dealii::Point<2, double>&&)
/home/jomivalen/cfem/dealii/include/deal.II/base/point.h:89:7: note:   no
known conversion for argument 1 from ‘dealii::Tensor<1, 2, double>’ to
‘dealii::Point<2, double>&&’


Thanks for your time,
Joaquín

2016-12-15 6:22 GMT-04:00 Jean-Paul Pelteret :

> Dear Joaquín,
>
> From your previous posts I assume that all of PeFT, Pe[qs] and F[qs]
> represent rank 2 tensors. If this is correct then no, the contraction
> indices that you've specified are not correct. They're out of the range
> [0,1] that are valid for rank-2 tensors, and this would presumably trigger
> an assertion in debug mode (at the very least).
>
> For simplicity, lets call these three rank-2 tensors "A,B,C". Then we
> could compute the two possible single index contraction results, expressed
> in index notation, as follows:
> (1a)  A_{ij} = B_{ik} C_{kj} --> A = contract<1,0>(B,C); // The second
> index of the first tensor contracts with the first index of the second
> tensor
> or, more simply,
> (1b) A = B*C;
> (2)A_{ij} = B_{ik} C_{kj} --> A = contract<1,1>(B,C); // The second index
> of the first tensor contracts with the second index of the second tensor
>
> Does this explain a bit better how the function works?
>
> Regards,
> J-P
>
> --
> The deal.II project is located at http://www.dealii.org/
> For mailing list/forum options, see https://groups.google.com/d/
> forum/dealii?hl=en
> ---
> You received this message because you are subscribed to a topic in the
> Google Groups "deal.II User Group" group.
> To unsubscribe from this topic, visit https://groups.google.com/d/
> topic/dealii/IL4Ls7CxCmc/unsubscribe.
> To unsubscribe from this group and all its topics, send an email to
> dealii+unsubscr...@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.
>



-- 
Joaquin Valencia
Graduate Student

-- 
The deal.II project is located at http://www.dealii.org/
For mailing list/forum options, see 
https://groups.google.com/d/forum/dealii?hl=en
--- 
You received this message because you are subscribed to the Google Groups 
"deal.II User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to dealii+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Re: [deal.II] Re: ifem

['Joaquin M Valencia Bravo' via deal.II User Group]

Dear Jean-Paul,

Thank you for your reply.  I tried with the second alternative you give me.
According to the example, contract<0,2> (t1, t2) , I modify the code as
follows:

 PeFT = contract<2,2> (Pe[qs], F[qs]);

I interprete that PeFT is the result, Pe[qs] would be the tensor t1, and
F[qs] would be the tensor t2.  Then index_1 == 2 belonging to Pe[qs], and
index_2 == 2 belonging to F[qs].

Please could you tell me if the modification I did is ok or not?.


Thanks for the help,
Joaquín


2016-12-15 2:16 GMT-04:00 Jean-Paul Pelteret :

> Dear Joaquin,
>
> The warning appears because you are calling a function
> 
> that is marked as deprecated, so although it remains in the library for now
> it will be removed in the near future.
>
> There are two alternate functions you can call if you would like to remove
> the warning right now. They are the normal "* operator"
> 
>  (if
> contracting over the inner two indices of the tensors) and the general
> contract
> 
> function that takes in two tensors and returns the result. The indices over
> which the contract function operates are specified as template parameters.
> You can see an example of how to use it in the function documentation.
>
> I hope that this helps you.
> Jean-Paul
>
>
> On Thursday, December 15, 2016 at 12:06:16 AM UTC+1, Joaquin wrote:
>>
>> Dear all,
>>
>> I am trying to run the IFEM code on FSI problems, developed by Luca
>> Heltai. I am using deal.II 8.4.1. I edited some parts of the code due to
>> warnings because it is considered deprecated for the version I am using. I
>> stopped when warning appears for line 1038:
>>
>> 1037  if((!par.semi_implicit) || (!par.use_spread))
>> *1038  contract (PeFT, Pe[qs], 2, F[qs], 2);*
>>
>> The warning I see is as follows:
>>
>> jomivalen@Nalia ~/ifem/ans-ifem $ make
>> Scanning dependencies of target ifem
>> [ 25%] Building CXX object CMakeFiles/ifem.dir/source/immersed_fem.cc.o
>> /home/jomivalen/ifem/ans-ifem/source/immersed_fem.cc: In instantiation
>> of ‘void 
>> ImmersedFEM::residual_and_or_Jacobian(dealii::BlockVector&,
>> dealii::BlockSparseMatrix&, const dealii::BlockVector&,
>> const dealii::BlockVector&, double, double) [with int dim = 2]’:
>> /home/jomivalen/ifem/ans-ifem/source/immersed_fem.cc:2046:16:   required
>> from here
>> /home/jomivalen/ifem/ans-ifem/source/immersed_fem.cc:1038:44: warning:
>> ‘void dealii::contract(dealii::Tensor<2, dim, Number>&, const
>> dealii::Tensor<2, dim, Number>&, unsigned int, const dealii::Tensor<2, dim,
>> Number>&, unsigned int) [with int dim = 2; Number = double]’ is deprecated
>> (declared at 
>> /home/jomivalen/cfem/dealii/include/deal.II/base/tensor_deprecated.h:262)
>> [-Wdeprecated-declarations]
>>  contract (PeFT, Pe[qs], 2, F[qs], 2);   // contract (PeFT,
>> Pe[qs], 2, F[qs], 2);
>> ^
>>
>> I don't know how to actualize the contract function despite having seen
>> the indications in the file "tensor.h".
>>
>> Thanks for any help,
>> Joaquín
>>
> --
> The deal.II project is located at http://www.dealii.org/
> For mailing list/forum options, see https://groups.google.com/d/
> forum/dealii?hl=en
> ---
> You received this message because you are subscribed to a topic in the
> Google Groups "deal.II User Group" group.
> To unsubscribe from this topic, visit https://groups.google.com/d/
> topic/dealii/IL4Ls7CxCmc/unsubscribe.
> To unsubscribe from this group and all its topics, send an email to
> dealii+unsubscr...@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.
>



-- 
Joaquin Valencia
Graduate Student

-- 
The deal.II project is located at http://www.dealii.org/
For mailing list/forum options, see 
https://groups.google.com/d/forum/dealii?hl=en
--- 
You received this message because you are subscribed to the Google Groups 
"deal.II User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to dealii+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Re: [deal.II] Re: Interfacing dealii-8.4.1 with trilinos-12.6.2 - make test failed

['Joaquin M Valencia Bravo' via deal.II User Group]

Thanks to all for your suggestions and explanations.

I already installed and configured dealii + trilinos and passed the 5 steps
after doing make test. The way I did that is:

1. I delete the intel compiler directory.
2. I follow the same steps as I sent in my message above.

My question now is if it will be there any negative consequences after I
removed the intel directory. Everytime I open the terminal appears the next
message:

bash:
/home/jomivalen/intel/compilers_and_libraries/linux/bin/compilervars.sh: No
such file or directory

Now I'm going to configure and install petsc and p4est.

Best,
Joaquín


2016-11-10 12:06 GMT-04:00 Wolfgang Bangerth :

> On 11/09/2016 05:35 PM, 'Joaquin M Valencia Bravo' via deal.II User Group
> wrote:
>
>>
>> jomivalen@Nalia ~/cfem/trilinos/lib $ ldd libepetra.so.12.6.2 | grep mpi
>> libmpi_cxx.so.1 => /usr/lib/libmpi_cxx.so.1 (0x7fb4ac8f7000)
>> libmpi.so.1 => /usr/lib/libmpi.so.1 (0x7fb4ac576000)
>>
>> jomivalen@Nalia ~/thesis/dealii-8.4.1/build/lib $ ldd
>> libdeal_II.g.so.8.4.1 | grep mpi
>> libmpicxx.so.12 =>
>> /home/jomivalen/intel/compilers_and_libraries_2016.1.150/lin
>> ux/mpi/intel64/lib/libmpicxx.so.12
>> (0x7f75b64c3000)
>> libmpi.so.12 =>
>> /home/jomivalen/intel/compilers_and_libraries_2016.1.150/lin
>> ux/mpi/intel64/lib/libmpi.so.12
>> (0x7f75b3855000)
>> libmpi_cxx.so.1 => /usr/lib/libmpi_cxx.so.1 (0x7f75ae037000)
>> libmpi.so.1 => /usr/lib/libmpi.so.1 (0x7f75adcb5000)
>>
>> What does these results mean?. What should i do?
>>
>
> Since the second of your questions was already answered, let me answer the
> first one (what does it mean?).
>
> You happen to have two different MPI installations on your system. One is
> apparently the one provided by Intel as part of the Intel compilers that
> you have installed on your system. The other one is just a regular MPI
> implementation in the standard /usr/lib directories. Apparently, when you
> built Trilinos, the Trilinos cmake configurationi scripts chose to use the
> regular MPI implementation. But for some reason, when you built deal.II,
> the deal.II cmake configuration scripts chose the Intel MPI.
>
> I don't know why that happens. It is possible, for example, that your
> $PATH includes both /usr/bin and /home/jomivalen/intel/compiler
> s_and_libraries_2016.1.150/linux/mpi/intel64/bin, and that in one
> directory you have an executable mpic++ and in the other mpicxx. Usually,
> they are interchangeable, and we just search for the first one we find. I
> assume that Trilinos does the same. If they search in a different order,
> then Trilinos finds one and deal.II finds the other one. If that's the
> case, then it is probably a bad idea to have both of these MPI
> implementations accessible by default.
>
> It is also possible that you loaded one module when you built Trilinos,
> but another when you built deal.II.
>
>
> Best
>  W.
>
> --
> 
> Wolfgang Bangerth  email: bange...@colostate.edu
>www: http://www.math.colostate.edu/~bangerth/
>
> --
> The deal.II project is located at http://www.dealii.org/
> For mailing list/forum options, see https://groups.google.com/d/fo
> rum/dealii?hl=en
> --- You received this message because you are subscribed to a topic in the
> Google Groups "deal.II User Group" group.
> To unsubscribe from this topic, visit https://groups.google.com/d/to
> pic/dealii/r4B-B-s6fO4/unsubscribe.
> To unsubscribe from this group and all its topics, send an email to
> dealii+unsubscr...@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.
>



-- 
Joaquin Valencia
Graduate Student

-- 
The deal.II project is located at http://www.dealii.org/
For mailing list/forum options, see 
https://groups.google.com/d/forum/dealii?hl=en
--- 
You received this message because you are subscribed to the Google Groups 
"deal.II User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to dealii+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Re: [deal.II] Re: Interfacing dealii-8.4.1 with trilinos-12.6.2 - make test failed

['Joaquin M Valencia Bravo' via deal.II User Group]

Sorry for the late answer. I had to re-install dealii + trilinos.

jomivalen@Nalia ~/cfem/trilinos/lib $ ldd libepetra.so.12.6.2 | grep mpi
libmpi_cxx.so.1 => /usr/lib/libmpi_cxx.so.1 (0x7fb4ac8f7000)
libmpi.so.1 => /usr/lib/libmpi.so.1 (0x7fb4ac576000)

jomivalen@Nalia ~/thesis/dealii-8.4.1/build/lib $ ldd libdeal_II.g.so.8.4.1
| grep mpi
libmpicxx.so.12 =>
/home/jomivalen/intel/compilers_and_libraries_2016.1.150/linux/mpi/intel64/lib/libmpicxx.so.12
(0x7f75b64c3000)
libmpi.so.12 =>
/home/jomivalen/intel/compilers_and_libraries_2016.1.150/linux/mpi/intel64/lib/libmpi.so.12
(0x7f75b3855000)
libmpi_cxx.so.1 => /usr/lib/libmpi_cxx.so.1 (0x7f75ae037000)
libmpi.so.1 => /usr/lib/libmpi.so.1 (0x7f75adcb5000)

What does these results mean?. What should i do?

Thanks,
Joaquin

2016-11-09 18:56 GMT-04:00 Wolfgang Bangerth :

> On 11/09/2016 03:52 PM, 'Joaquin M Valencia Bravo' via deal.II User Group
> wrote:
>
>>
>> In case of trilinos, I followed the next steps:
>> $  mkdir build
>> $  cd build
>> $
>>
>> cmake \
>> -DTrilinos_ENABLE_Sacado=ON \
>> -DTrilinos_ENABLE_MueLu:BOOL=ON \
>> -DTrilinos_ENABLE_Stratimikos=ON \
>> -DCMAKE_BUILD_TYPE=RELEASE \
>> -DCMAKE_CXX_FLAGS="-g -O3" \
>> -DCMAKE_C_FLAGS="-g -O3" \
>> -DCMAKE_FORTRAN_FLAGS="-g -O5" \
>> -DTrilinos_EXTRA_LINK_FLAGS="-lgfortran" \
>> -DCMAKE_VERBOSE_MAKEFILE=FALSE \
>> -DTrilinos_VERBOSE_CONFIGURE=FALSE \
>> -DTPL_ENABLE_MPI=ON \
>> -DBUILD_SHARED_LIBS=ON \
>> -DCMAKE_INSTALL_PREFIX:PATH=$HOME/share/trilinos \
>>
>> $ make -j8 install
>>
>> In case of dealii, I followed the next steps:
>>
>> $ mkdir build
>>
>> $ cd build
>>
>> $ cmake -DCMAKE_INSTALL_PREFIX=/home/jomivalen/cfem/deal.II \
>> -DTRILINOS_DIR=/home/jomivalen/thesis/trilinos-12.6.2 \
>>
>> -DDEAL_II_WITH_TRILINOS=ON -DDEAL_II_WITH_MPI=ON ../
>>
>> $ make -j8 install
>>
>> $ make test (failed)
>>
>
> That's interesting then. Can you try to do
>   ldd $HOME/share/trilinos/lib/libepetra.so.12.6.2
>   ldd /home/jomivalen/cfem/deal.II/lib/libdeal_II.g.so.8.4.1
> and in both cases see which MPI libraries they link to? For example, for
> me this yields the following:
>
> rilinos-12.8.1/lib> ldd libepetra.so.12.8.1 | grep mpi
> libmpi_usempi.so.1 => /usr/lib64/openmpi/lib/libmpi_usempi.so.1
> (0x7f1107a5f000)
> libmpi_mpifh.so.12 => /usr/lib64/openmpi/lib/libmpi_mpifh.so.12
> (0x7f110780b000)
> libmpi_cxx.so.1 => /usr/lib64/openmpi/lib/libmpi_cxx.so.1
> (0x7f1107091000)
> libmpi.so.12 => /usr/lib64/openmpi/lib/libmpi.so.12
> (0x7f1106dad000)
> libopen-rte.so.12 => /usr/lib64/openmpi/lib/libopen-rte.so.12
> (0x7f1105d31000)
> libopen-pal.so.13 => /usr/lib64/openmpi/lib/libopen-pal.so.13
> (0x7f1105a8e000)
> rilinos-12.8.1/lib> cd
> f/bangerth> cd p/deal.II/1/build/lib/
> build/lib> ldd libdeal_II.so.8.5.0-pre | grep mpi
> libmpi_cxx.so.1 => /usr/lib64/openmpi/lib/libmpi_cxx.so.1
> (0x7f7e22f28000)
> libmpi.so.12 => /usr/lib64/openmpi/lib/libmpi.so.12
> (0x7f7e22181000)
> libmpi_usempi.so.1 => /usr/lib64/openmpi/lib/libmpi_usempi.so.1
> (0x7f7e152ff000)
> libmpi_mpifh.so.12 => /usr/lib64/openmpi/lib/libmpi_mpifh.so.12
> (0x7f7e150ab000)
> libopen-rte.so.12 => /usr/lib64/openmpi/lib/libopen-rte.so.12
> (0x7f7e148cf000)
> libopen-pal.so.13 => /usr/lib64/openmpi/lib/libopen-pal.so.13
> (0x7f7e1462c000
>
> So they are the same.
>
>
> Best
>  W.
>
> --
> 
> Wolfgang Bangerth  email: bange...@colostate.edu
>www: http://www.math.colostate.edu/~bangerth/
>
> --
> The deal.II project is located at http://www.dealii.org/
> For mailing list/forum options, see https://groups.google.com/d/fo
> rum/dealii?hl=en
> --- You received this message because you are subscribed to a topic in the
> Google Groups "deal.II User Group" group.
> To unsubscribe from this topic, visit https://groups.google.com/d/to
> pic/dealii/r4B-B-s6fO4/unsubscribe.
> To unsubscribe from this group and all its topics, send an email to
> dealii+unsubscr...@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.
>



-- 
Joaquin Valencia
Graduate Student

-- 
The deal.II project is located at http://www.dealii.org/
For mailing list/forum options, see 
https://groups.google.com/d/forum/dealii?hl=en
--- 
You received this message because you are subscribed to the Google Groups 
"deal.II User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to dealii+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Re: [deal.II] Re: Interfacing dealii-8.4.1 with trilinos-12.6.2 - make test failed

['Joaquin M Valencia Bravo' via deal.II User Group]

Apparently I used the MPI by default for both of them.

In case of trilinos, I followed the next steps:
$  mkdir build
$  cd build
$

cmake \
-DTrilinos_ENABLE_Sacado=ON \
-DTrilinos_ENABLE_MueLu:BOOL=ON \
-DTrilinos_ENABLE_Stratimikos=ON \
-DCMAKE_BUILD_TYPE=RELEASE \
-DCMAKE_CXX_FLAGS="-g -O3" \
-DCMAKE_C_FLAGS="-g -O3" \
-DCMAKE_FORTRAN_FLAGS="-g -O5" \
-DTrilinos_EXTRA_LINK_FLAGS="-lgfortran" \
-DCMAKE_VERBOSE_MAKEFILE=FALSE \
-DTrilinos_VERBOSE_CONFIGURE=FALSE \
-DTPL_ENABLE_MPI=ON \
-DBUILD_SHARED_LIBS=ON \
-DCMAKE_INSTALL_PREFIX:PATH=$HOME/share/trilinos \

$ make -j8 install

In case of dealii, I followed the next steps:

$ mkdir build

$ cd build

$ cmake -DCMAKE_INSTALL_PREFIX=/home/jomivalen/cfem/deal.II \
-DTRILINOS_DIR=/home/jomivalen/thesis/trilinos-12.6.2 \

-DDEAL_II_WITH_TRILINOS=ON -DDEAL_II_WITH_MPI=ON ../

$ make -j8 install

$ make test (failed)

Thank you for your patience,

Joaquin


2016-11-09 18:47 GMT-04:00 Wolfgang Bangerth :

> On 11/09/2016 03:44 PM, 'Joaquin M Valencia Bravo' via deal.II User Group
> wrote:
>
>>
>> cmake \
>> -D Trilinos_ENABLE_Sacado=ON \
>> -D Trilinos_ENABLE_MueLu:BOOL=ON \
>> -D Trilinos_ENABLE_Stratimikos=ON \
>> -D CMAKE_BUILD_TYPE=RELEASE \
>> -D CMAKE_CXX_FLAGS="-g -O3" \
>> -D CMAKE_C_FLAGS="-g -O3" \
>> -D CMAKE_FORTRAN_FLAGS="-g -O5" \
>> -D Trilinos_EXTRA_LINK_FLAGS="-lgfortran" \
>> -D CMAKE_VERBOSE_MAKEFILE=FALSE \
>> -D Trilinos_VERBOSE_CONFIGURE=FALSE \
>> -D TPL_ENABLE_MPI=ON \
>> -D BUILD_SHARED_LIBS=ON \
>> -D CMAKE_INSTALL_PREFIX:PATH=$HOME/share/trilinos \
>>
>>
>> In case of deal.II, I followed the next steps:
>>  $   mkdir build
>>  $   cd build
>>  $   cmake -DCMAKE_INSTALL_PREFIX=/home/jomivalen/cfem/deal.II ../
>>
>
> Does this build deal.II with MPI? If you enter the last command, it will
> provide you with a summary at the end that states whether MPI was used or
> not.
>
>
> Best
>  W.
>
> --
> 
> Wolfgang Bangerth  email: bange...@colostate.edu
>www: http://www.math.colostate.edu/~bangerth/
>
> --
> The deal.II project is located at http://www.dealii.org/
> For mailing list/forum options, see https://groups.google.com/d/fo
> rum/dealii?hl=en
> --- You received this message because you are subscribed to a topic in the
> Google Groups "deal.II User Group" group.
> To unsubscribe from this topic, visit https://groups.google.com/d/to
> pic/dealii/r4B-B-s6fO4/unsubscribe.
> To unsubscribe from this group and all its topics, send an email to
> dealii+unsubscr...@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.
>



-- 
Joaquin Valencia
Graduate Student

-- 
The deal.II project is located at http://www.dealii.org/
For mailing list/forum options, see 
https://groups.google.com/d/forum/dealii?hl=en
--- 
You received this message because you are subscribed to the Google Groups 
"deal.II User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to dealii+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Re: [deal.II] Re: Interfacing dealii-8.4.1 with trilinos-12.6.2 - make test failed

['Joaquin M Valencia Bravo' via deal.II User Group]

Apparently I used the same MPI by default for both of them.

In case of trilinos, I followed the next steps:
$  mkdir build
$  cd build
$

cmake \
-DTrilinos_ENABLE_Sacado=ON \
-DTrilinos_ENABLE_MueLu:BOOL=ON \
-DTrilinos_ENABLE_Stratimikos=ON \
-DCMAKE_BUILD_TYPE=RELEASE \
-DCMAKE_CXX_FLAGS="-g -O3" \
-DCMAKE_C_FLAGS="-g -O3" \
-DCMAKE_FORTRAN_FLAGS="-g -O5" \
-DTrilinos_EXTRA_LINK_FLAGS="-lgfortran" \
-DCMAKE_VERBOSE_MAKEFILE=FALSE \
-DTrilinos_VERBOSE_CONFIGURE=FALSE \
-DTPL_ENABLE_MPI=ON \
-DBUILD_SHARED_LIBS=ON \
-DCMAKE_INSTALL_PREFIX:PATH=$HOME/share/trilinos \

$ make -j8 install

In case of dealii, I followed the next steps:

$ mkdir build

$ cd build

$ cmake -DCMAKE_INSTALL_PREFIX=/home/jomivalen/cfem/deal.II \
-DTRILINOS_DIR=/home/jomivalen/thesis/trilinos-12.6.2 \

-DDEAL_II_WITH_TRILINOS=ON -DDEAL_II_WITH_MPI=ON ../

$ make -j8 install

$ make test (failed)

Thank you for your patience,

Joaquin


2016-11-09 18:31 GMT-04:00 Wolfgang Bangerth :

> On 11/09/2016 03:30 PM, 'Joaquin M Valencia Bravo' via deal.II User Group
> wrote:
>
>>
>> I understand, but I don't know how to compile both of them with the same
>> version. Could you please tell me the steps to do that?
>>
>
> Well, how did you compile the two before? :-) Presumably you used some way
> to tell Trilinos which MPI to use, and some way to tell deal.II which MPI
> to use.
>
>
> Best
>  W.
>
> --
> 
> Wolfgang Bangerth  email: bange...@colostate.edu
>www: http://www.math.colostate.edu/~bangerth/
>
> --
> The deal.II project is located at http://www.dealii.org/
> For mailing list/forum options, see https://groups.google.com/d/fo
> rum/dealii?hl=en
> --- You received this message because you are subscribed to a topic in the
> Google Groups "deal.II User Group" group.
> To unsubscribe from this topic, visit https://groups.google.com/d/to
> pic/dealii/r4B-B-s6fO4/unsubscribe.
> To unsubscribe from this group and all its topics, send an email to
> dealii+unsubscr...@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.
>



-- 
Joaquin Valencia
Graduate Student

-- 
The deal.II project is located at http://www.dealii.org/
For mailing list/forum options, see 
https://groups.google.com/d/forum/dealii?hl=en
--- 
You received this message because you are subscribed to the Google Groups 
"deal.II User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to dealii+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Re: [deal.II] Re: Interfacing dealii-8.4.1 with trilinos-12.6.2 - make test failed

['Joaquin M Valencia Bravo' via deal.II User Group]

Apparently I used the same MPI by default for both of them.
In case of trilinos, I followed the next steps:
$   mkdir build
$   cd build
$

cmake \
-D Trilinos_ENABLE_Sacado=ON \
-D Trilinos_ENABLE_MueLu:BOOL=ON \
-D Trilinos_ENABLE_Stratimikos=ON \
-D CMAKE_BUILD_TYPE=RELEASE \
-D CMAKE_CXX_FLAGS="-g -O3" \
-D CMAKE_C_FLAGS="-g -O3" \
-D CMAKE_FORTRAN_FLAGS="-g -O5" \
-D Trilinos_EXTRA_LINK_FLAGS="-lgfortran" \
-D CMAKE_VERBOSE_MAKEFILE=FALSE \
-D Trilinos_VERBOSE_CONFIGURE=FALSE \
-D TPL_ENABLE_MPI=ON \
-D BUILD_SHARED_LIBS=ON \
-D CMAKE_INSTALL_PREFIX:PATH=$HOME/share/trilinos \


In case of deal.II, I followed the next steps:
 $   mkdir build
 $   cd build
 $   cmake -DCMAKE_INSTALL_PREFIX=/home/jomivalen/cfem/deal.II ../
 $   make -j8 install
 $   make test

2016-11-09 18:31 GMT-04:00 Wolfgang Bangerth :

> On 11/09/2016 03:30 PM, 'Joaquin M Valencia Bravo' via deal.II User Group
> wrote:
>
>>
>> I understand, but I don't know how to compile both of them with the same
>> version. Could you please tell me the steps to do that?
>>
>
> Well, how did you compile the two before? :-) Presumably you used some way
> to tell Trilinos which MPI to use, and some way to tell deal.II which MPI
> to use.
>
>
> Best
>  W.
>
> --
> 
> Wolfgang Bangerth  email: bange...@colostate.edu
>www: http://www.math.colostate.edu/~bangerth/
>
> --
> The deal.II project is located at http://www.dealii.org/
> For mailing list/forum options, see https://groups.google.com/d/fo
> rum/dealii?hl=en
> --- You received this message because you are subscribed to a topic in the
> Google Groups "deal.II User Group" group.
> To unsubscribe from this topic, visit https://groups.google.com/d/to
> pic/dealii/r4B-B-s6fO4/unsubscribe.
> To unsubscribe from this group and all its topics, send an email to
> dealii+unsubscr...@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.
>



-- 
Joaquin Valencia
Graduate Student

-- 
The deal.II project is located at http://www.dealii.org/
For mailing list/forum options, see 
https://groups.google.com/d/forum/dealii?hl=en
--- 
You received this message because you are subscribed to the Google Groups 
"deal.II User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to dealii+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Re: [deal.II] Re: Interfacing dealii-8.4.1 with trilinos-12.6.2 - make test failed

['Joaquin M Valencia Bravo' via deal.II User Group]

I understand, but I don't know how to compile both of them with the same
version. Could you please tell me the steps to do that?

Thanks for your time,
Joaquin

2016-11-09 18:10 GMT-04:00 Wolfgang Bangerth :

> On 11/09/2016 02:46 PM, 'Joaquin M Valencia Bravo' via deal.II User Group
> wrote:
>
>>
>> Which version of trilinos may I use in my case?. I followed the
>> instructions on page:
>>
>>
>> https://www.dealii.org/8.4.1/external-libs/trilinos.html
>>
>> although, it was for trilinos-12.0.1. Should I use the same version?.
>> What suggestions could you give me?
>>
>
> Joaquin,
> you ask the wrong question. The issue is not whether you should use
> Trilinos 12.0 or 12.4 or 12.6. The problem was that you compiled Trilinos
> *with one version* of MPI, and deal.II *with another version* of MPI. This
> is what you need to fix. You can continue the same version of Trilinos, as
> long as you install it with the same MPI as you use for deal.II.
>
> Best
>  W.
>
>
> --
> 
> Wolfgang Bangerth  email: bange...@colostate.edu
>www: http://www.math.colostate.edu/~bangerth/
>
> --
> The deal.II project is located at http://www.dealii.org/
> For mailing list/forum options, see https://groups.google.com/d/fo
> rum/dealii?hl=en
> --- You received this message because you are subscribed to a topic in the
> Google Groups "deal.II User Group" group.
> To unsubscribe from this topic, visit https://groups.google.com/d/to
> pic/dealii/r4B-B-s6fO4/unsubscribe.
> To unsubscribe from this group and all its topics, send an email to
> dealii+unsubscr...@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.
>



-- 
Joaquin Valencia
Graduate Student

-- 
The deal.II project is located at http://www.dealii.org/
For mailing list/forum options, see 
https://groups.google.com/d/forum/dealii?hl=en
--- 
You received this message because you are subscribed to the Google Groups 
"deal.II User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to dealii+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Re: [deal.II] Re: Interfacing dealii-8.4.1 with trilinos-12.6.2 - make test failed

['Joaquin M Valencia Bravo' via deal.II User Group]

Thank you for your response:

Which version of trilinos may I use in my case?. I followed the
instructions on page:


https://www.dealii.org/8.4.1/external-libs/trilinos.html

although, it was for trilinos-12.0.1. Should I use the same version?. What
suggestions could you give me?

Sincerely,
Joaquin



2016-11-09 15:51 GMT-04:00 Wolfgang Bangerth :

> On 11/09/2016 12:04 PM, 'Joaquin' via deal.II User Group wrote:
>
>>
>> I hope someone help me. I want to run step-40, so I need trilinos
>> and petsc. First, I'm trying to interface trilinos-12.6.2 with
>> deal.II-8.4.1. My computer has a gnu 4.8.4 compiler. Everything is
>> ok until the "make test". The five step failed. I'm sending
>> detailed.log and quicktests.log files.
>>
>>
>>  Thanks in advance.
>>
>
> The error you show
>
> /usr/bin/ld: warning: libmpi.so.1, needed by /home/jomivalen/thesis/trilino
> s-12.6.2/lib/libmuelu-adapters.so, may conflict with libmpi.so.12
> ../../lib/libdeal_II.g.so.8.4.1: undefined reference to
> `Teuchos::details::safeCommFree(int*)'
> ../../lib/libdeal_II.g.so.8.4.1: undefined reference to
> `Epetra_MpiComm::Epetra_MpiComm(int)'
>
> suggests that you build Trilinos with a different MPI implementation than
> you used for deal.II. This is not going to work.
>
> Best
>  W>
>
> --
> 
> Wolfgang Bangerth  email: bange...@colostate.edu
>www: http://www.math.colostate.edu/~bangerth/
>
>
> --
> The deal.II project is located at http://www.dealii.org/
> For mailing list/forum options, see https://groups.google.com/d/fo
> rum/dealii?hl=en
> --- You received this message because you are subscribed to a topic in the
> Google Groups "deal.II User Group" group.
> To unsubscribe from this topic, visit https://groups.google.com/d/to
> pic/dealii/r4B-B-s6fO4/unsubscribe.
> To unsubscribe from this group and all its topics, send an email to
> dealii+unsubscr...@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.
>



-- 
Joaquin Valencia
Graduate Student

-- 
The deal.II project is located at http://www.dealii.org/
For mailing list/forum options, see 
https://groups.google.com/d/forum/dealii?hl=en
--- 
You received this message because you are subscribed to the Google Groups 
"deal.II User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to dealii+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.