About

Python bindings to tucanos

Install

• Install Rust
• Optionally enter your prefered conda or venv or virtualenv
• Run:

pip install git+https://github.com/tucanos/pytucanos.git

To install tucanos in debug mode:

pip install -C debug=true git+https://github.com/tucanos/pytucanos.git

To install Tucanos with libmeshb support:

pip install -C meshb=true git+https://github.com/tucanos/pytucanos.git

Benchmarks

.meshb/.solb I/O

libMeshb is required to run the benchmarks with the reference codes below. It can be built as explained in the tucanos README.

Reference codes

MMG

A minimal version can be installed with:

cmake .. -DCMAKE_BUILD_TYPE=Release -DCMAKE_INSTALL_PREFIX=$PREFIX -DUSE_VTK=OFF -DUSE_ELAS=OFF -DUSE_SCOTCH=OFF
make -j$(nproc) install

NB: the performance will be better with scotch installed

Omega_h

Before building Omega`_h you shall ensure that sandialabs/omega_h#408 is merged in you sources. Also make sure that libMeshb is installed. A minimal version can then be installed with:

cmake .. -DCMAKE_INSTALL_PREFIX=$PREFIX -DCMAKE_BUILD_TYPE=Release -DBUILD_SHARED_LIBS=ON -DOmega_h_USE_libMeshb=ON
make -j$(nproc) install

Refine

A minimal version can be installed with:

cmake .. -DCMAKE_INSTALL_PREFIX=$PREFIX -DCMAKE_BUILD_TYPE=Release -DBUILD_SHARED_LIBS=ON
make -j$(nproc) install

Avro

Detailed installation instructions are available here

Test cases

• Isotropic remeshing in a 2D square domain
• Anisotropic remeshing in a 2D square domain
• Isotropic remeshing in a 3D cubic domain
• Anisotropic remeshing in a 3D cubic domain
• polar-1 benchmark from the UGAWG