/parallelDMD

A parallel dynamic mode decomposition routine

Primary LanguageC++GNU General Public License v3.0GPL-3.0

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.out
  • make 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