Problems about step-34 of the new version 8.2.1

[叶常青]

Hello, I am a newer of Linux and deal.II.

Now I use Centos7. I have some questions.

Question1:

First question is about the step-35, which can be got from all the examples that 

need other software packages.

Deal.II is installed  on by default as the install introduction said. path=/usr/local

But I don't know why I got this problem. And even UMFPACK is installed, I 

got this problem. As the documentation said, the UMFPACK interface will be enabled by default.

[francis@localhost examples]$ cd step-35

[francis@localhost step-35]$ cmake .

CMake Error at CMakeLists.txt:41 (MESSAGE):

  

  Error! The deal.II library found at 

/home/francis/bin/deal.II was not

  configured with

      DEAL_II_WITH_UMFPACK = 

ON

  One or all of these are OFF in your installation but are required for this

  tutorial 

step.

-- Configuring incomplete, errors occurred!

Question2: The second is about the step-34.

The old version 8.1 hasn't got this problem, but the version 8.2.1.

[100%] Run step-34 with Debug configuration

DEAL::

DEAL::Parsing parameter file 

parameters.prm

DEAL::for a 2 dimensional simulation. 

DEAL:GMRES::Starting value 

2.21576

DEAL:GMRES::Convergence step 1 value 2.37635e-13

DEAL::Cycle 0:

DEAL::   

Number of active cells:       20

DEAL::   Number of degrees of freedom: 20

--------------------------------------------------------

An error occurred in line <745> 

of file </home/francis/Downloads/deal.II/source/numerics/data_out_dof_data.cc> 

in function

    void dealii::DataOut_DoFData<DH, patch_dim, 

patch_space_dim>::add_data_vector(const VECTOR&, const 

std::vector<std::basic_string<char> >&, dealii::DataOut_DoFData<DH, patch_dim, 

patch_space_dim>::DataVectorType, const 

std::vector<dealii::DataComponentInterpretation::DataComponentInterpretation>&

) [with VECTOR = dealii::Vector<double>; DH = dealii::DoFHandler<1, 2>; int 

patch_dim = 1; int patch_space_dim = 2]

The violated condition was: 

    (dofs == 0) || 

(triangulation->n_active_cells() != dofs->n_dofs())

The name and call sequence of 

the exception was:

    ExcMessage("Unable to determine the type of vector 

automatically because the number of DoFs " "is equal to the number of cells. Please 

specify DataVectorType.")

Additional Information: 

Unable to determine the type of 

vector automatically because the number of DoFs is equal to the number of cells. 

Please specify DataVectorType.

Stacktrace:

-----------

#0  

/home/francis/bin/deal.II/lib/libdeal_II.g.so.8.2.1: void 

dealii::DataOut_DoFData<dealii::DoFHandler<1, 2>, 1, 

2>::add_data_vector<dealii::Vector<double> >(dealii::Vector<double> const&, 

std::vector<std::string, std::allocator<std::string> > const&, 

dealii::DataOut_DoFData<dealii::DoFHandler<1, 2>, 1, 2>::DataVectorType, 

std::vector<dealii::DataComponentInterpretation::DataComponentInterpretation, 

std::allocator<dealii::DataComponentInterpretation::DataComponentInterpretation

> > const&)

#1  /home/francis/bin/deal.II/lib/libdeal_II.g.so.8.2.1: void 

dealii::DataOut_DoFData<dealii::DoFHandler<1, 2>, 1, 

2>::add_data_vector<dealii::Vector<double> >(dealii::Vector<double> const&, 

std::string const&, dealii::DataOut_DoFData<dealii::DoFHandler<1, 2>, 1, 

2>::DataVectorType, 

std::vector<dealii::DataComponentInterpretation::DataComponentInterpretation, 

std::allocator<dealii::DataComponentInterpretation::DataComponentInterpretation

> > const&)

#2  ./step-34: Step34::BEMProblem<2>::output_results(unsigned int)

#3  

./step-34: Step34::BEMProblem<2>::run()

#4  ./step-34: main

--------------------------------------------------------

make[3]: *** [CMakeFiles/run] 

Aborted (core dumped)

make[2]: *** [CMakeFiles/run.dir/all] Error 2

make[1]: *** 

[CMakeFiles/run.dir/rule] Error 2

make: *** [run] Error 2

[叶常青]

Question3 is about the installed path and Linux users group. As I installed deal.II in /usr/local with root privilage.

If I want to programming with IDE( such as Eclipse) and I log in Linux with non-root user, I can not use the examples of deal.II in IDE.

Does it mean as a non-root user, if I want to use the examples, I should install deal.II in /home/user path?

Or if I install deal.II in /usr/local, I have to log in Linux as root user to use the example with IDE?

[叶常青]

Another question, if I want to install all the other needed packages, such as UMFPACK. which kind of path are appropriate for root or non-root users? If I want to programming with in IDE?

Thank you very much.  

在 2015年1月25日星期日 UTC+8上午11:16:58，叶常青写道：

[Bruno Turcksin]

Hi,

you should have a file called detailed.log can you send it ?

Unless you have a good reason to install deal.II as root, I wouldn't do it. You can use -DCMAKE_INSTALL_PREFIX to choose the directory where deal.II will be installed (see https://dealii.org/8.2.1/index.html).

Best,

Bruno

[Francis Green]

Thank you very much.

I got your mean. But I still donot know why the document of Deal.II said all the packages are set ON by default. But I read from the detailed.log they are OFF when installing deal.II.  

And I kown PETSC is installed as root, but why I got some results said the package is OFF. Sorry, I am newcomer of DEAL.II and LINUX system.

So do you mean that I have to install all the other packages in /home/mypath/ as a non-root user? Including the Deal.II?

And the detailed.log is below here.

###

#

#  deal.II configuration:

#        CMAKE_BUILD_TYPE:       DebugRelease

#        BUILD_SHARED_LIBS:      ON

#        CMAKE_INSTALL_PREFIX:   /home/francis/bin/deal.II

#        CMAKE_SOURCE_DIR:       /home/francis/Downloads/deal.II (Version 8.2.1)

#        CMAKE_BINARY_DIR:       /home/francis/Downloads/build

#        CMAKE_CXX_COMPILER:     GNU 4.8.2 on platform Linux x86_64

#                                /usr/bin/c++

#        CMAKE_C_COMPILER:       /usr/bin/cc

#        CMAKE_Fortran_COMPILER: /usr/bin/gfortran

#        CMAKE_GENERATOR:        Unix Makefiles

#

#  Base configuration (prior to feature configuration):

#        DEAL_II_CXX_FLAGS:            -pedantic -fpic -Wall -Wpointer-arith -Wwrite-strings -Wsynth -Wsign-compare -Wswitch -Wno-unused-local-typedefs -Wno-long-long -Wno-deprecated -Wno-deprecated-declarations -std=c++11

#        DEAL_II_CXX_FLAGS_RELEASE:    -O2 -funroll-loops -funroll-all-loops -fstrict-aliasing -Wno-unused

#        DEAL_II_CXX_FLAGS_DEBUG:      -Og -ggdb -Wa,--compress-debug-sections

#        DEAL_II_LINKER_FLAGS:         -Wl,--as-needed -rdynamic

#        DEAL_II_LINKER_FLAGS_RELEASE: 

#        DEAL_II_LINKER_FLAGS_DEBUG:   -ggdb

#        DEAL_II_DEFINITIONS:          

#        DEAL_II_DEFINITIONS_RELEASE:  

#        DEAL_II_DEFINITIONS_DEBUG:    DEBUG

#        DEAL_II_USER_DEFINITIONS:     

#        DEAL_II_USER_DEFINITIONS_REL: 

#        DEAL_II_USER_DEFINITIONS_DEB: DEBUG

#        DEAL_II_INCLUDE_DIRS          

#        DEAL_II_USER_INCLUDE_DIRS:    

#        DEAL_II_BUNDLED_INCLUDE_DIRS: 

#        DEAL_II_LIBRARIES:            m

#        DEAL_II_LIBRARIES_RELEASE:    

#        DEAL_II_LIBRARIES_DEBUG:      

#

#  Configured Features (DEAL_II_ALLOW_BUNDLED = ON, DEAL_II_ALLOW_AUTODETECTION = ON):

#      ( DEAL_II_WITH_64BIT_INDICES = OFF )

#      ( DEAL_II_WITH_ARPACK = OFF )

#        DEAL_II_WITH_BOOST set up with external dependencies

#            BOOST_VERSION = 1.53.0

#            BOOST_INCLUDE_DIRS = /usr/include

#            BOOST_USER_INCLUDE_DIRS = /usr/include

#            BOOST_LIBRARIES = /usr/lib64/libboost_iostreams-mt.so;/usr/lib64/libboost_serialization-mt.so;/usr/lib64/libboost_system-mt.so;/usr/lib64/libboost_thread-mt.so

#        DEAL_II_WITH_BZIP2 set up with external dependencies

#            BZIP2_VERSION = 1.0.6

#            BZIP2_INCLUDE_DIRS = /usr/include

#            BZIP2_LIBRARIES = /usr/lib64/libbz2.so

#        DEAL_II_WITH_CXX11 = ON

#      ( DEAL_II_WITH_HDF5 = OFF )

#      ( DEAL_II_WITH_LAPACK = OFF )

#      ( DEAL_II_WITH_METIS = OFF )

#      ( DEAL_II_WITH_MPI = OFF )

#      ( DEAL_II_WITH_MUMPS = OFF )

#        DEAL_II_WITH_MUPARSER set up with bundled packages

#            MUPARSER_BUNDLED_INCLUDE_DIRS = /home/francis/Downloads/deal.II/bundled/muparser_v2_2_3//include

#      ( DEAL_II_WITH_NETCDF = OFF )

#      ( DEAL_II_WITH_OPENCASCADE = OFF )

#      ( DEAL_II_WITH_P4EST = OFF )

#      ( DEAL_II_WITH_PETSC = OFF )

#      ( DEAL_II_WITH_SLEPC = OFF )

#        DEAL_II_WITH_THREADS set up with external dependencies

#            TBB_VERSION = 4.1

#            THREADS_LINKER_FLAGS = -pthread

#            THREADS_INCLUDE_DIRS = /usr/include

#            THREADS_USER_INCLUDE_DIRS = /usr/include

#            THREADS_LIBRARIES = /usr/lib64/libtbb.so

#      ( DEAL_II_WITH_TRILINOS = OFF )

#      ( DEAL_II_WITH_UMFPACK = OFF )

#        DEAL_II_WITH_ZLIB set up with external dependencies

#            ZLIB_VERSION = 1.2.7

#            ZLIB_INCLUDE_DIRS = /usr/include

#            ZLIB_LIBRARIES = /usr/lib64/libz.so

#

#  Component configuration:

#        DEAL_II_COMPONENT_COMPAT_FILES

#      ( DEAL_II_COMPONENT_DOCUMENTATION = OFF )

#        DEAL_II_COMPONENT_EXAMPLES

#        DEAL_II_COMPONENT_MESH_CONVERTER

#      ( DEAL_II_COMPONENT_PACKAGE = OFF )

#      ( DEAL_II_COMPONENT_PARAMETER_GUI = OFF )

#

###

在 2015年1月27日星期二 UTC+8上午2:59:30，Bruno Turcksin写道：

[Francis Green]

And should I install all the other packages (petsc, trilion etc) first before the deal.II. 

And I shoud configure all the paths of the other packages when deal.II is installing?

Thanks very much!

在 2015年1月25日星期日 UTC+8上午11:16:58，Francis Green写道：

[Uwe Köcher]

Hej Francis,

first of all, I've tested the current version v8.2.1 of the deal.II library. It did compile with

the support of umfpack on a CentOS 7 and step-35 is running fine.

For this, make sure that you have installed the suitesparse and the suitesparse-devel

system package, if you want to use umfpack from the system or, use the switch

to force deal.II to use the bundled version: see here

https://www.dealii.org/8.0.0/development/cmake.html

Generelly, you have to compile every additional package (such as Trilinos, Petsc)

before you compile deal.II

For this, I use an automated installation tool, you can find it on GitHub:

  git clone https://github.com/koecher/candi

see also the former discussions here on this group, such as

https://groups.google.com/forum/#!searchin/dealii/candi/dealii/bC312KH11wc/dBCceS-FOb0J

Best

  Uwe

[Francis Green]

Thank you very much. 

在 2015年1月27日星期二 UTC+8下午8:48:56，Uwe Köcher写道：

[Uwe Köcher]

Hej Francis,

you are right, in step-34 is a problem with the data out handling.

When you change the lines around 1017 to

    dataout.add_data_vector(dh, phi, "phi");
    dataout.add_data_vector(dh, alpha, "alpha");

in words, put the DoFHandler "dh" in front of the vectors, the program will run.

I'll set up a pull request for this issue soon.

[Francis Green]

Sorry, I can find any automated installation tool in you web in github.

would please tell me how to find and install this tool?

在 2015年1月27日星期二 UTC+8下午8:48:56，Uwe Köcher写道：

Hej Francis,

[Uwe Köcher]

Simply clone it with (as I have given it in the former post)

  git clone https://github.com/koecher/candi

[Martin Steigemann]

Dear Francis,

 

when configuring deal.II the CMAKE scripts look in standard places for the needed components, you see in your log file that boost was found and tbb and so on, if some component was not found, it is set to OFF, if you have something in a non-standard place, for example PETSc in your home or whereever, you have to tell this to CMAKE, for example

 

cmake –DCMAKE_INSTALL_PREFIX=/some/dir/in/your/home \

  -DDEAL_II_WITH_PETSC=ON \

  -DPETSC_DIR=/path/to/petsc \

  -DPETSC_ARCH=architecture \

 

  ... maybe some more options ...

 

../dealii

 

have a look here for options

 

http://www.dealii.org/8.2.1/readme.html

 

Maybe this helps,

 

Best,

Martin

 

[Francis Green]

Deal Martin

Thank for your exact answer and I have some other qustions about compile with Eclispe! Eclipse is a wonderful IDE for programming with deal.II.

I created the Eclipse project as the dealii Wiki said. 

 Take the example step-22 for example, I got this problem.

and this are yellow swirly lines under those two lines.  Does it mean we should add virtual destructor to the class BoundaryValues and Righthandside?

Description Resource Path Location Type

Class 'BoundaryValues' has virtual method 'vector_value' but non-virtual destructor step-22.cc /step-22 line 174 Code Analysis Problem

Description Resource Path Location Type

Class 'RightHandSide' has virtual method 'vector_value' but non-virtual destructor step-22.cc /step-22 line 215 Code Analysis Problem

Another problem is about compile step-1 with Eclipse.

step-1 run very well and I got the expected results, but the Eclipse shows two problems

Description Resource Path Location Type

Invalid arguments '

Candidates are:

double distance(const dealii::Point<int2,double> &)

' step-1.cc /step-1 line 196 Semantic Error

Description Resource Path Location Type

Invalid arguments '

Candidates are:

void set_boundary(unsigned char, const ? &)

void set_boundary(unsigned char)

' step-1.cc /step-1 line 126 Semantic Error

I donot know why. Anyone can help me? 

在 2015年1月28日星期三 UTC+8下午2:24:57，Martin Steigemann写道：

[Martin Steigemann]

Dear Francis,

the first problem found by Eclipse is C++, if you derive a class from some base class, the base class should always have a virtual deconstructor, there are many explanations about this, just google, in short, if you have a pointer to some base class, the program does not know what class really points to it and just calls the deconstructor of the base class, if the deconstructor is not virtual. There are better explanations ...

 

> Description Resource Path Location Type

> Invalid arguments '

> Candidates are:

> double distance(const dealii::Point<int2,double> &)

> ' step-1.cc /step-1 line 196 Semantic Error

Why "Point<int2,double>", maybe rebuild the index of Eclipse ... no idea, but I would not rely too much on the code analyzer in Eclipse, depending on the version, it has bugs ...

Best,

Martin

[Francis Green]

Dear Matrin,

Thank you for your explanation.

Anther problem is the former . If I install the other package(such as petsc) after the installation of deal.II and I want to compile some examples that need these packages, what should I do? Using Eclipse, it just include the pathes of the packages.

But how do I do with cmake and make/ make run?

Thanks!

在 2015年1月28日星期三 UTC+8下午4:48:03，Martin Steigemann写道：

[Bruno Turcksin]

Francis,

On 01/28/2015 07:57 AM, Francis Green wrote:
> Anther problem is the former . If I install the other package(such as
> petsc) after the installation of deal.II and I want to compile some
> examples that need these packages, what should I do? Using Eclipse, it
> just include the pathes of the packages.
> But how do I do with cmake and make/ make run?
>

if you install petsc after having install deal.II, you need to reinstall
deal.ii and you need to tell cmake where petsc is installed.

Best,

Bruno

[Francis Green]

Thank you very much! 

在 2015年1月28日星期三 UTC+8下午11:14:42，Bruno Turcksin写道：

[Francis Green]

Sorry, I have some other questions. My problem is to solve  nonlinear parabolic equations, depending on time. So I decide to apply the time-step method, theta-Crank-Nicolson method.

The solutions of each time step is depending on former time-step and and coefficients of Laplace_matrix and Mass_matrix are depending on former solutions.

I have not find any solvers for nonlinear PDEs in deal.II(such as quasi-Newton methods) . Would you please tell which step in the examples can enlight me?

Another problem is I donot know why all of the steps of examples are all coded in only one source file.

If I want to solve a problem that is a coupling system, such as dynamic and heat transfering. Or I want to solve a optimzation problem and I need other files for optimzation methods.

What should I do? should I put all the codes in a source file?

Putting a huge lines in a file is awesome!

在 2015年1月28日星期三 UTC+8下午11:14:42，Bruno Turcksin写道：

Francis,

[Thomas Wick]

Hi Francis,

answers below:

01/29/2015 10:06 AM, Francis Green wrote:

Sorry, I have some other questions. My problem is to solve  nonlinear parabolic equations, depending on time. So I decide to apply the time-step method, theta-Crank-Nicolson method.

The solutions of each time step is depending on former time-step and and coefficients of Laplace_matrix and Mass_matrix are depending on former solutions.

I have not find any solvers for nonlinear PDEs in deal.II(such as quasi-Newton methods) . Would you please tell which step in the examples can enlight me?

When you search on:

https://www.dealii.org/8.2.1/doxygen/deal.II/Tutorial.html

you find step-48, step-33, step-25, step-15.

In addition, nonlinear fluid mechanics coupled with nonlinear solid mechanics are solved with a Newton method here, for example:

http://media.archnumsoft.org/10305/

Another problem is I donot know why all of the steps of examples are all coded in only one source file.

If I want to solve a problem that is a coupling system, such as dynamic and heat transfering. Or I want to solve a optimzation problem and I need other files for optimzation methods.

What should I do? should I put all the codes in a source file?

Putting a huge lines in a file is awesome!

Please visit

https://www.dealii.org/developer/index.html

and then 'CMake in user projects',

to learn how you can split and compile your code in several files. You are absolutely right that you
need such an option for large projects otherwise you get crazy with all the code in one file.

Hopefully this helps.

Best Thomas

++--------------------------------------------------------------------++
Dr. Thomas Wick
Research Scientist
Johann Radon Institute for Computational and Applied Mathematics (RICAM)
Austrian Academy of Sciences (ÖAW)
Altenberger Straße 69
4040 Linz, Austria 

Email:  thoma...@ricam.oeaw.ac.at
www:    http://people.ricam.oeaw.ac.at/t.wick/
++--------------------------------------------------------------------++
--

在 2015年1月28日星期三 UTC+8下午11:14:42，Bruno Turcksin写道：

Francis,

On 01/28/2015 07:57 AM, Francis Green wrote:
> Anther problem is the former . If I install the other package(such as
> petsc) after the installation of deal.II and I want to compile some
> examples that need these packages, what should I do? Using Eclipse, it
> just include the pathes of the packages.
> But how do I do with cmake and make/ make run?
>
if you install petsc after having install deal.II, you need to reinstall
deal.ii and you need to tell cmake where petsc is installed.

Best,

Bruno

--
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+un...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Francis Green]

I noticed some steps of examples assemble Laplace and Mass matrices by the function provided by deal.II, but some other assemble them by code( loop over cells ), even some of these big problems.

So in which kinds of occassions, I should assemble these matrices by myself(like loop over all cells)?

Thanks!

[Thomas Wick]

On 01/29/2015 10:18 AM, Francis Green wrote:

I noticed some steps of examples assemble Laplace and Mass matrices by the function provided by deal.II, but some other assemble them by code( loop over cells ), even some of these big problems.

So in which kinds of occassions, I should assemble these matrices by myself(like loop over all cells)?

This of course depends on your problem. If you have additional coefficients or additional equations terms that can not be handled by the build-in functions,
you need to do it by yourself.

Honestly, I am always doing it by myself to have all the flexibility if I want to change something in the equation.

Best Thomas

-- 
++--------------------------------------------------------------------++
Dr. Thomas Wick
Research Scientist
Johann Radon Institute for Computational and Applied Mathematics (RICAM)
Austrian Academy of Sciences (ÖAW)
Altenberger Straße 69
4040 Linz, Austria 

Email:  thoma...@ricam.oeaw.ac.at

www:    http://people.ricam.oeaw.ac.at/t.wick/
++--------------------------------------------------------------------++
--

Thanks!

[Francis Green]

Deal Thomas

Thank you for your detailed explanation. About the CMake, if I use Eclipse to compile my project, can I split the project to several files, then plus #include XXs  in the main source file?

Thanks!

在 2015年1月29日星期四 UTC+8下午5:16:40，Thomas Wick写道：

[Thomas Wick]

Dear Francis,

Deal Thomas

Thank you for your detailed explanation. About the CMake, if I use Eclipse to compile my project, can I split the project to several files, then plus #include XXs  in the main source file?

In general yes; of course, you need to include all your files. But I never used eclipse but just simply emacs.

Best Thomas

[Francis Green]

Thank you very much. I am not familiar with vi/vim or emacs. A new tool always takes a long time to learn.

I notice that in step-33, Sacado class in Trilinos is used to get the approximation of Jacobi matrix.

Are there any other methods or classes in deal.II to solve this approximation, such as forward-diffrerence method?

Thanks !

Are 

在 2015年1月29日星期四 UTC+8下午8:37:41，Thomas Wick写道：

[Thomas Wick]

Dear Francis,

On 01/29/2015 02:11 PM, Francis Green wrote:

Thank you very much. I am not familiar with vi/vim or emacs. A new tool always takes a long time to learn.

But this is exactly the pleasure and fun in research: to discover and learn new things and tools.

I notice that in step-33, Sacado class in Trilinos is used to get the approximation of Jacobi matrix.

Are there any other methods or classes in deal.II to solve this approximation, such as forward-diffrerence method?

This is again as in my email before: you have to do it by yourself - but honestly it is not so difficult to
implement a finite difference scheme of your Jacobian.

Best Thomas

[Francis Green]

Thanks a lot!

vi/vim and emacs are old tools now! haha~. Of course, it is new to me. Maybe I will learn emacs someday.

在 2015年1月29日星期四 UTC+8下午9:16:26，Thomas Wick写道：

[Thomas Wick]

On 01/29/2015 02:35 PM, Francis Green wrote:

Thanks a lot!

vi/vim and emacs are old tools now! haha~. Of course, it is new to me. Maybe I will learn emacs someday.

Of course, eclipse offers more.

Let me know if it works out with several files and if you are able to compile them.

Thomas

[Wolfgang Bangerth]

On 01/28/2015 02:48 AM, 'Martin Steigemann' via deal.II User Group wrote:
> no idea, but I would not rely too much on the code analyzer in Eclipse,
> depending on the version, it has bugs ...

Indeed. Eclipse tries to understand the source code, but it is not a compiler
-- it makes mistakes, and it sometimes annotates things that it doesn't
understand. That's annoying, but it doesn't mean that you code won't compile:
it's only eclipse that makes the mistake, not the compiler.

Best
W.


--
------------------------------------------------------------------------
Wolfgang Bangerth email: bang...@math.tamu.edu
www: http://www.math.tamu.edu/~bangerth/