I'm submitting jobs to a Sun Grid Engine using the qsub command. The -t option to qsub enables me to specify the datasets upon which I want to call my script -- e.g.,
$ qsub . . . -t 101-103 my_script.sh
My question is, is it possible to specify non-consecutive datasets with the -t option? For example, say I wanted to run the script on 101 and 103, but not 102. How would I accomplish that?
And, more generally, how would I select arbitrarily numbered datasets?
I would like an answer that works in practice for a large number of datasets -- far beyond the two used in this toy example.
If the goal is to run regularly spaced datasets -- e.g., 1, 3, 5, . . . or 10, 15, 20, . . . -- then @chrk's answer is the one to use.
For arbitrarily numbered datasets, using -t is not possible. The same functionality can be attained, however, using the submit command (with the -f option) instead of qsub.
$ submit . . . -s my_script.sh -f my_datasets.txt
The file my_datasets.txt contains one dataset per line, as in
101
103
I'm not sure how specific this solution is to the particular configuration of my computing environment.