Akantu means a little element in Kinyarwanda, a Bantu language. From now on it is also an open-source object-oriented library which has the ambition to be generic and efficient. Even though the code is written to be generic, Akantu strength are in solid mechanics models for fracture and contact simulations.
The full documentation can be found on ReadTheDocs
In order to compile Akantu any compiler supporting fully C++14 should work. In addition some libraries are required:
- CMake (>= 3.5.1)
- Boost (pre-processor and Spirit)
- Eigen3 (if not present the build system will try to download it)
For the python interface:
- Python (>=3 is recommended)
- pybind11 (if not present the build system will try to download it)
To run parallel simulations:
To use the static or implicit dynamic solvers at least one of the following libraries is needed:
- MUMPS (since this is usually compiled in static you also need MUMPS dependencies)
To compile the tests and examples:
- google-test (if not present the build system will try to download it)
sh > sudo apt install cmake libboost-dev zlib1g-dev gmsh libeigen3-dev # For parallel > sudo apt install mpi-default-dev libmumps-dev libscotch-dev # For sequential > sudo apt install libmumps-seq-dev
This works only for sequential computation since mumps from conda-forge is compiled without MPI support
sh > conda create -n akantu > conda activate akantu > conda install boost cmake > conda install -c conda-forge mumps
sh > brew install gcc > brew install firstname.lastname@example.org > brew tap brewsci/num > brew install brewsci-mumps --without-brewsci-parmetis
If it does not work you can edit url to http://graal.ens-lyon.fr/MUMPS/MUMPS_5.3.5.tar.gz using the command:
sh > brew edit brewsci/num
Akantu is a CMake project, so to configure it, you can follow the usual way:
sh > cd akantu > mkdir build > cd build > ccmake .. [ Set the options that you need ] > make > make install
You will need to specify the compiler explicitly:
sh > CC=gcc-12 CXX=g++-12 FC=gfortran-12 cmake ..
Considering the homebrew is installed in /opt/homebrew Define the location of the `Scotch` library path:
sh > cmake .. -DSCOTCH_LIBRARY="/opt/homebrew/lib/libscotch.dylib;/opt/homebrew/lib/libscotcherr.dylib;/opt/homebrew/lib/libscotcherrexit.dylib"
Specify path to all `MUMPS` libraries:
sh > cmake .. -DMUMPS_DIR=/opt/homebrew/opt/brewsci-mumps
In case the above does not work, specify the `MUMPS` path manually using (e.g.):
sh > cmake .. -DMUMPS_LIBRARY_COMMON=/opt/homebrew/opt/brewsci-mumps/lib/libmumps_common.dylib
If compilation does not work change the path of the failing libraries to brew downloads in /opt/homebrew/.
You can install `Akantu` using pip, this will install a pre-compiled version, this works only on Linux machines for now:
sh > pip install akantu
You can then import the package in a python script as:
python import akantu
The python API is similar to the C++ one. If you encounter any problem with the python interface, you are welcome to do a merge request or post an issue on GitLab.
Any contribution is welcome, we are trying to follow a gitflow workflow, so the project developers can create branches named features/<name of my feature> or bugfixes/<name of the fix> directly in the main akantu repository. External fellows can Fork the project. In both cases the modifications have to be submitted in the form of a Merge Request.
If you want to ask for help concerning Akantu's compilation, usage or problem with the code do not hesitate to open an Issue on gitlab. If you want to contribute and don't know where to start, you are also invited to open an issue.
To help getting started, multiple tutorials using the python interface are available as notebooks with pre-installed version of Akantu on Renku. The tutorials can be tested here: