madeleine interpolation tool (MIT)

MIT is a simple C++ interpolation tool based on bitpit with Python bindings. MIT provides user with a C++ library for interpolation between two meshes (defined as discipline meshes) via a buffer mesh (defined neutral mesh). This library can be used in Python thanks to a Cythonization of the C++ library.

C++ library dependencies:

cmake
bitpit (compiled as a shared library with flag "-fPIC")
MPI (optional)

Python wrapping dependencies:

Cython
Numpy
other Python modules: distutils, libc, libcpp, traceback, imp, os, re, sys

Configuring C++ library compilation:

In MIT's root folder make a building folder, e.g. build

    madeleine$ mkdir build

Enter the build folder

    madeleine$ cd build

In order to configure it, run:

    madeleine/build$ ccmake ../

By this way, MIT can be configured for C++ and/or Python usage. Setting some variable in ccmake interface you can customize a bit your configuration. The only mandatory cmake variable that has to be set is

    BITPIT_DIR = [/bitpit/installation/dir]/lib/cmake/bitpit-{bitpit_version}

where bitpit-version should be 1.5 (previous versions have not been tested). In order to use the Python wrappers the following cmake variable has to be set:

    BUILD_SHARED_LIBS = ON

C++ examples are compiled if the following bariable is set

    BUILD_EXAMPLES = ON

Building C++ library:

Once configured the C++ library can be compiled with

    make

C++ library example:

If BUILD_EXAMPLES variable has been set to ON, in [build_folder]/example an exacutable named example_00002 is produced.

Just move to the above folder and launch it as follows

    ./example_00002

By this way the example will use to default meshes and analtical data to perform interpolation from neutral mesh to discipline mesh, a very easy computation and an interpolation from discipline mesh to neutral mesh. The example procudes vtu files to check the result.

The user can customize the example run by providing 2 stl files containing 2 different triagulations of a unit sphere.

    ./example_00002 /path/to/first/stl/file /path/to/second/stl/file

Building Python wrappers:

Once the MIT C++ library has been compiled, in order to build Cythonized Python wrappers, move to the

 python-wrapper

folder into the MIT's root folder and launch

    python setup.py build_ext --inplace --bitpit-path=[/path/to/bitpit/installation/folder] \
    --madeleine-path=[/path/to/madeleine/root/folder]/build/src/ --extensions-source=coupling.pyx

This string will produce a Python module named

    coupling.so

In order to be able to use it the following environment variable should contain at least

    LD_LIBRARY_PATH=$LD_LIBRARY_PATH:[/path/to/bitpit/dynamic/library]:[/path/madeleine/root/folder]/build/src

Optionally

    LD_LIBRARY_PATH=$LD_LIBRARY_PATH:[/path/to/bitpit/dynamic/library]:[/path/madeleine/root/folder]/build/src:[/path/to/MPI/library]

Running Python wrappers example:

In the python-wrapper folder, an example, named py_example_00002.py is provided. Just simply run

    python py_example_00002.py

to launch it. This script reproduces exactly the same computation of the C++ example (example_00002)

optimad / madeleine