Go to the source code of this file.
Functions/Subroutines | |
| program | mpi_scatci |
| MPI-SCATCI main program. More... | |
Function/Subroutine Documentation
◆ mpi_scatci()
| program mpi_scatci |
MPI-SCATCI main program.
- Date
- 2017 - 2019
MPI-SCATCI diagonalizes Hamiltonians for one or more irreducible representations. The input for each is read from the single input file as the pair of standard SCATCI namelists &input and &cinorn. If there are fewer MPI processes than Hamiltonians to diagonalize, each process is given several Hamiltonians to diagonalize sequentially. If there are more processes than Hamiltonians, some or all Hamiltonians will be diagonalized in a distributed regime. All processes that participate on diagonalization of the same matrix are part of a "diagonalization group" which has separate BLACS context and their own MPI communicator.
The eigenvectors are kept in the memory and, if desired, are then used to calculate transition dipole moments and further derived data.
- Note
- 16/01/2019 - Jakub Benda: Unifom coding style and expanded documentation.
- 25/01/2019 - Jakub Benda: Extension to support multiple subsequent diagonalizations.
- 30/01/2019 - Jakub Benda: Evaluation of properties using libcdenprop, including cross-symmetry ones.
Definition at line 41 of file mpi_scatci.f90.
