The OSA scripts jemx_science_analysis and ibis_science_analysis take in input the og_group that you created with the og_create command, and apply various analysis steps on each SCW, one after the other. This process can be long, especially if your analysis involves hundreds of SCWs. Moreover if one of the SCWs in your list is problematic for any of the analysis step, the script might be interrupted and no analysis will be performed also for the subsequent SCWs.
With some manipulation, experienced users can however trivially parallelise part of the analysis process. We provide here a brief discussion, showing one possible approach to conduct the analysis in parallel over your entire list of SCWs, instructions are given for JEM-X, but can be transferred to IBIS with ease. This is optimally achieved in a cluster, as described below, but it can be done also on single machines.
In order to submit separate jobs, you will have to produce two shell scripts: one is the script you will actually run from the head node of your cluster, the other is the ``cluster script'' containing the analysis commands (setup of the OSA envoironment variables, og_create and jemx_science_analysis commands with the parameters required for your analysis).
The first script should read the list of SCWs (for example you read one line at a time from a txt file similar to the mos.lst file that we used in Sect. ), and include a loop over each SCWs. In this loop, you will submit a job to your cluster, the job being the ``cluster script'' . You are therefore submitting the same job for each SCW in your list, and this job will setup the OSA environment in a cluster node, and run the og_create and jemx_science_analysis commands from the appropriate subdirectories. The exact command you will need to submit the jobs may depend on the cluster you are working on. Although you will be running the same cluster script for each SCW, it is a good practice to store the pfiles in a temporary directory, one for each SCW. Each of these directories containing the pfiles can be deleted soon after the analysis for the given SCW ends. The list of parameters used for the analysis is stored in the log file (defined through the COMMONLOGFILE environment variable, as explained in Sect. ).
Following the procedure above you will end up with one og_group for each SCWs in your list. Each og_group will contain only a single SCW, as shown in the scheme of Fig. . Once the analysis has completed for all the desired SCWs, you can merge the products to obtain the mosaic image, a single spectrum or lightcurve for your dataset by following the instructions in Sect. .
|
|