pdlf_reloc
A set of functions to compute time-delays and perform dd relocation. It's a work in progress (meaning, it is not yet working).
Dependencies
- python3 (not tested on python2 but should work)
- A functional MPI 1.x/2.x/3.x implementation like OpenMPI or MPICH built with shared/dynamic libraries
Python modules:
- numpy
- scipy.signal
- h5py
- matplotlib.pyplot
- pyproj
Download the package
To get everything, use the following git command:
git clone https://github.com/eost/pdlf_reloc.git
or download the module directly from https://github.com/eost/pdlf_reloc.git
Define clusters
First compute the correlation matrix using
mpiexec -n NPROC python cc_matrix.py
where NPROC is the number of CPUs that will be used to compute correlation coeffixients (should be at least 2: 1 process for the boss and at least 1 worker).
Then define clusters using
python define_clusters.py
Compute time-delays
Input parameters are indicated in Arguments.py
There are two possibilities to compute time-delays:
1/ Use an MPI version of the script to compute time-delays on multiple CPUs:
mpiexec -n NPROC python dt_computation_mpi.py
where NPROC is the number of CPUs that will be used to compute delays (should be at least 2: 1 process for the boss and at least 1 worker).
2/ Use a sequential version of the script to compute time-delays:
python dt_computation_cc.py
DD reloc
Use:
python ddreloc.py