File names

A special problem arises when the parallel processes use the same directory to create their files in. Whether or not this is the case depends on the platform. Consult the Installation Manual or ask the person who has installed the program, or someone else who is familiar with parallelization on your machine.

If the parallel processes use the same directory a conflict of names would arise if the different processes all tried to create for instance a file called TAPE21. Therefore, the file names are modified in this situation and all normal file names are appended by _n where n is an integer between 0 and n (n+1 being the total number of parallel processes. The _1 files correspond to kid #1, etc. The program run scripts automatically take care of this aspect by renaming or copying files as required. The program run scripts are part of the ADF package and should be used, as described later in this manual (see also the Examples document), in particular for parallel runs.

Standard output is a special case here: the parent writes its normal ('print') output to standard output while the kids each write to a file KidOutput. When file names are modified as discussed above, the kid output files become KidOutput_n. There is also a KidOutput_0 (from the parent) but this file is empty (apart from a dummy record): the parent writes its output to standard output.