Parallel DMD
This is a parallel implementation for the classic SVD-based dynamic mode decomposition, where the Cholesky decomposition is used to compute the SVD matrices.
Output results are only with respect to the un-tweaked amplitudes, eigenvalues, and spatial modes, and no selection criterion is specified within the code since the selection criterion can be different and flexible.
Dependency
- Eigen3
- MPI library with C++ bindings
Compiling
make
: compiling source code to ./bin/dmd_mpi.outmake clean
: remove compiled binaries
Usage
mpirun -n <cores> dmd_mpi.out <inputFile> <xn> <tn>
- cores is the number of processors to be requested
- inputFile should be an xn by tn+1 matrix
- xn is the number of rows
- tn is the number of columns