-
Notifications
You must be signed in to change notification settings - Fork 1
MLalign2D_combine
The calculations of ml_align2d are easily parallelizable. The integration over all experimental images can be split in as many subsets as desired. Within one iteration (!), these calculations are indpendent and each of these subsets can be calculated on a different processor. This program serves to combine all output from the parallel runs with a simple command. This program explicitly uses the specific structure of the ml_align2d-logfiles, and the specific output filenames of the ml_align2d program. Therefore, these should not be changed.
If a cluster of processors is available that shares an (NFS-mounted) hard disc, then MlAlign2DCshScript may be used to parallelize the ml_align2d-calculations.
In case of a cluster of processors that uses a LSF queuing system (with commands bsub, bjobs, etc.), MlAlign2DCshScriptLsf may be used to parallelize the ml_align2d-calculations. These scripts can be executed with a command like:
$ nohup ./script_name.csh &
$ ml_align2d_combine ...
Parameters
- `` Input selfile containingMLalign2D-logfiles of all parallel runs
- `` Rootname for all output files
This program is used as follows:
$ selfile_create "processor_*.log" > all_logs.sel
$ ml_align2D_combine -i all_logs.sel -o ML_it002
Which creates the same output files as theMLalign2D program, now withML_it002as rootname:ML_it002_ref0000[1-3].xmp,ML_it002.sel,ML_it002.log, and if the-output_assignments flag was used on the different parallel runs also:ML_it002_ref0000[1-3].sel andML_it002.doc.
--Main.AlfredoSolano - 25 Jan 2007