New in 05/2021: please see this video for an introduction to the code's datastructures: https://www.youtube.com/watch?v=qBBIW2-ktgo
With WABBIT it is possible to solve partial differential equations (PDE) on block-based adaptive grids. Calculations in 2D and 3D are possible and performed fully parallel. The set of PDE is encapsulated from the code handling the adaptive grid, and thus existing monobloc solvers can be adapted easily for this solver. WABBIT can handle PDEs of the following type:
and can be defined. This implementation is handled by the "physics-modules".
How to get a copy of WABBIT and compiling the code:
- Clone from github
git clone [email protected]:adaptive-cfd/WABBIT.git
-
Compile the code running make.
Note that WABBIT requires:
- MPI library
- HDF5 library
- BLAS+LAPACK library
- some variables have to be exported (see below).
make [FC=[mpif90]]
- Run the testfiles with
make test
Make sure that the mpi library which is also used for WABBIT is installed (for example by loading mpich3).
This is a short example (/working practice) of how to install hdf5 libary (Tested for version hdf5-1.10.1).
-
download source code from hdf5
-
open terminal and follow
(mind that path_2_build_dir has to be replaced by the path of the directory of your choice.)
gunzip < hdf5-X.Y.Z.tar.gz | tar xf -
cd hdf5-X.Y.Z
./configure --prefix=path_2_build_dir --enable-fortran --enable-parallel
make
make check # run test suite.
make install
make check-install # verify installation.
- export variables:
HDF_ROOT=path_2_build_dir
export HDF_ROOT
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:${HDF_ROOT}/lib:${HDF_ROOT}/lib64
export LD_RUN_PATH=$LD_LIBRARY_PATH
Recommendation: Add the lines to export the variables to your bashrc-file. Otherwise, the export has to be done every time you open a new terminal to compile the code.
Customize the template .ini-file and rename file to [your_filename.ini], run WABBIT with option for dimension and .ini-file name
mpirun -n 1 ./wabbit [your_filename.ini] --memory=2.0GB
where the --memory options allows you to control how much memory is globally allocated, i.e., on all CPUs. Note that WABBIT does not free memory during runtime. This is because we intent to use clusters, where the available memory is reserved for the execution of the code alone. This is quite typical for supercomputing.
If you are new to WABBIT it is recommended to read the information for newcomers!
In case you have problems with the preparation to use WABBIT, have a look at the informations given in the wiki
For further Information see the documentation. Therefore it is necessary to have Doxygen installed.
make doc
-
"Wavelet adaptive proper orthogonal decomposition for large-scale flow data"; Krah, Engels, Schneider, Reiss; Advances in Comput. Math. Volume 48, Article number: 10, 2022
-
"A wavelet-adaptive method for multiscale simulation of turbulent flows in flying insects"; T. Engels, K. Schneider, J. Reiss and M. Farge; Commun. Comput. Phys., 30, 1118-1149, 2021
-
"An Open and Parallel Multiresolution Framework Using Block-Based Adaptive Grids"; Sroka, Engels, Krah, Mutzel, Schneider, Reiss; Active Flow and Combustion Control 2018