NuTo Wiki

Status: Planning

Brought to you by: beinersd, eckardt4, ttitscher, unger3

Installation

Authors:

Attachments

ANN-cmake.patch (1688 bytes)

ARPACK-cmake.patch (3593 bytes)

Installation & Build a NuTo environment

Build environment

Requirements

SWIG >=1.3.40 (aptitude install swig2.0 python-instant)
CMAKE >= 2.8.0
Python >= 2.6 (aptitude install python2.7-dev)
Doxygen >= 1.5.8
BOOST >= 1.36.0
BLAS
Eigen >= 3.0
ANN
ARPACK
MUMPS: http://graal.ens-lyon.fr/MUMPS: sparse direct solver (How to install MUMPS) -> needed for all explicit FE solutions
PARDISO (Parallel solver direct or iterative)

Optional packages

Intel Math Kernel Library (Intel MKL) http://software.intel.com/en-us/intel-mkl: sparse direct solver, blas, lapack (binding with python requires static linking with MKL libraries)

Instruction for compilation (Unix OS)

checkout NuTo?

svn co --username=USERNAME svn+ssh://USERNAME@svn.code.sf.net/p/nuto/code/ nuto

or update using

svn up

Build and Install external packages (optional)
a. Download individually from the relevant homepages and build static libraries (on 64-bit systems the -fPIC option is required during compilation) or build shared libraries -> recommended to get the newest versions
b. Download via ssh (login required) repository location has changed

svn co svn+ssh://parrot.bauing.uni-weimar.de/home/projects/NUTO/externals

External Libraries

Eigen3

Homepage: http://eigen.tuxfamily.org

cd eigen
mkdir eigen3
tar xzf 3.0.6.tar.gz --strip 1 -C eigen3
cd eigen3
mkdir objdir
cd objdir
cmake -DCMAKE_INSTALL_PREFIX=/opt/nuto ..
make
make install (eventually as root)

BOOST

Homepage: http://www.boost.org/

cd boost
tar xzf boost_1_54_0.tar.gz
cd boost_1_54_0
./bootstrap.sh --prefix=/opt/nuto
./bjam --layout=versioned cxxflags="-fPIC -std=c++0x" --with-serialization --with-filesystem --with-python link=static
./bjam --layout=versioned cxxflags="-fPIC -std=c++0x" --with-serialization --with-filesystem --with-python link=static install (eventually as root)

In order not to use the standard compiler, do e.g.

./bjam --toolset=gcc-4.4 --layout=versioned cxxflags="-fPIC -std=c++0x" --with-serialization --with-filesystem --with-python link=static

In order to use debug version do

 ./bjam debug --layout=versioned cxxflags="-fPIC -std=c++0x" --with-serialization --with-filesystem --with-python link=static

BLAS

Homepage: http://github.com/xianyi/OpenBLAS/archive/v0.2.14.tar.gz.

cd openblas
tar xzf OpenBLAS-0.2.14.tar.gz
cd OpenBLAS-0.2.14
make USE_OPENMP=1 (if openMP parallelization is used else just make)        
make PREFIX=/opt/nuto/openblas install (eventually as root)

PARDISO

Homepage: http://www.pardiso-project.org/branch. Download the appropriate PARDISO library. "libpardiso500-GNU472-X86-64.so" means you are using at least gcc version 4.7.2.

cp libpardiso500-GNU<gcc_version>-X86-64.so /opt/nuto/lib/libpardiso.so (eventually as root)

ANN

Homepage: http://www.cs.umd.edu/~mount/ANN
Patches: ANN-cmake.patch

cd ann
tar xzf ann_1.1.2.tar.gz
cd ann_1.1.2
patch -p1 -i ../ANN-cmake.patch
mkdir objdir
cd objdir
cmake -DCMAKE_INSTALL_PREFIX=/opt/nuto ..
make
make install (eventually as root)
mkdir /opt/nuto/include/ANN (eventually as root)
mv /opt/nuto/include/ANN.h /opt/nuto/include/ANN (eventually as root)

Arpack

Homepage: http://www.caam.rice.edu/software/ARPACK/
Patches: Arpack-cmake.patch

cd ARPACK

tar xzf arpack96.tar.gz
cd ARPACK
patch -p1 -i ../ARPACK-cmake.patch
mkdir objdir
cd objdir
cmake -DCMAKE_INSTALL_PREFIX=/opt/nuto ..
make
make install (eventually as root)

Build the Makefiles with cmake (check CMakeLists.txt for additional build options) An out-of-source is recommended
a. create a build directory (it is recommended that the build directory is not a subdirectory of the nuto source directory)

    mkdir nuto_build
    cd nuto_build

b. call cmake to create unix make files

cmake -DNUTO_EXTERNAL_LIBRARIES_ROOT=/opt/nuto -DBOOST_ROOT=/opt/nuto mysourcedirectory

here mysourcedirectory = 'home/username/develop/nuto'
or to create unix make files and an eclipse project for debugging and only python tests turned on

cmake -G"Eclipse CDT4 - Unix Makefiles" -DNUTO_EXTERNAL_LIBRARIES_ROOT=/opt/nuto -DBOOST_ROOT=/opt/nuto  -DENABLE_DEBUG:BOOL=ON -DENABLE_C++_TESTS:BOOL=OFF  -DENABLE_EXAMPLES:BOOL=OFF ../nuto

or to create unix make files and an eclipse project with optimization and python and c++ tests turned on

cmake -G"Eclipse CDT4 - Unix Makefiles" -DNUTO_EXTERNAL_LIBRARIES_ROOT=/opt/nuto -DBOOST_ROOT=/opt/nuto -DENABLE_DEBUG:BOOL=OFF -DENABLE_C++_TESTS:BOOL=ON ../nuto

cmake -G"Eclipse CDT4 - Unix Makefiles"  -DCMAKE_ECLIPSE_VERSION=3.8 -DNUTO_EXTERNAL_LIBRARIES_ROOT=/opt/nuto -DBOOST_ROOT=/opt/nuto -DPYTHON_INCLUDE_DIR=/usr/include/python2.7 -DPYTHON_LIBRARY=/usr/lib/python2.7/config-x86_64-linux-gnu/libpython2.7.so -DENABLE_DEBUG:BOOL=ON -DENABLE_C++_TESTS:BOOL=OFF -DENABLE_EXAMPLES:BOOL=OFF ../nuto

this one allows you to specify the python version you are using (in case python2 and python3 are installed in parallel) and the eclipse version

call the build process (in your working directory)
```
make
```

set search path for python modules

export PYTHONPATH=$(pwd)/src:$PYTHONPATH

test the build process
```
make test
```
a single test file (e.g. Brick8N) can be executed
```
ctest -V -R Brick8N
```
build documentation
```
make doc
```
if you have problems with the Fortran-compiler, set the path to the Fortran-compiler by the environment variable FC or the cmake variable CMAKE_Fortran_COMPILER.

Instruction to include the new NuTo-modules to your Python-module environment (Unix OS)

add module path to your environment

export PYTHONPATH=$PYTHONPATH:<path to>/mynuto/src

GUI requirements

The NuTo sources contain a simple GUI (“NuToGUI”) to try out NuTo and view results.

It has the following additional requirements:

wxWidgets >= 2.8 (2.8.11 or 2.9.1 recommended)
VTK >= 5.2
Boost.Python

wxWidgets

wxWidgets (http://www.wxwidgets.org/) is a cross-platform GUI toolkit, used to create the interface for the NuToGUI application.

For Linux, wxWidgets is usually available as a package in the distribution. For Windows, binaries are available.

A comprehensive overview of the different ways to acquire wxWidgets can be found on it's Wiki: http://wiki.wxwidgets.org/Guides_%26_Tutorials#Installing_wxWidgets_and_Setting_up_Your_IDE

When configuring NuTo for building you may have to tell CMake where to find wxWidgets. The CMake documentation lists the different variables that control how wxWidgets is found: http://cmake.org/cmake/help/cmake-2-8-docs.html#module:FindwxWidgets

VTK

VTK is a visualization toolkit and used for the results display.

VTK uses CMake for building; so building it is essentially just a matter of doing cmake <path to="" vtk_sources="">; make. (On Windows you should provide an appropriate generator options.) When building NuTo you need to specify where VTK can be found - this is done through the VTK_DIR variable, like cmake -DVTK_DIR=<path where="" vtk="" was="" built=""> <path to="" nuto="">.

</path></path></path>

Boost.Python

On Linux, there are probably distribution packages available that contain Boost.Python.

If you built Boost yourself, Boost.Python should usually be built as a part of it, if the Python headers+libraries are available.

Running the GUI

You need a NuTo python module set up, as described above. You can run the GUI application itself with ./nutogui from the build directory. (Without a correct NuTo environment you will get an error message along the lines of “could not initialize Python ... could not import module nuto”.)

Small walkthrough:

Click the “Open file” button, locate and open Brick8N.py in the NuTo sources.
Click “Run script”.
A new tab with the results activates.
You could view different result data (drop down that starts with “Solid color”).
You could visualize the displacement in various ways - click one of the “displacement” buttons in the toolbar. The scale is also inviting for playing around.

Using a different compiler with cmake

Instead of just calling cmake do

 CC=/usr/bin/gcc-4.4 CXX=/usr/bin/g++-4.4 FC=/usr/bin/gfortran-4.4 cmake [...]

Attachments

dSFMT-src-2.1-DSFMT_MEXP.patch (483 bytes) - added by eckardt4 4 years ago.
dSFMT-src-2.1-cmake.patch (1.3 KB) - added by eckardt4 4 years ago.
ANN-cmake.patch (1.6 KB) - added by unger3 3 years ago.