individual filterbank output
siemion opened this issue · 12 comments
Add the ability to output individual filterbank files from multi-antenna input rather than quasi-compliant "multi-antenna filterbank"
As a first approach, we're going to implement a flag to select a single antenna from a raw file to generate the output .fil . Can easily parallelize over antennas to get all the files
rawspec already supports a user-specified subset of coarse channels that can be used to process only channels from a single antenna. No need to add anything special for that. The only drawback is reading the RAW file multiple times.
You can't read a sub-coarse channel on input. On the write side it currently writes out all fine channels for the coarse channels you select. If you want the ability to output a subset of the generated fine channels, it's probably better as a separate issue/feature request since it's orthogonal to selecting a single antenna from a RAW file.
already supports a user-specified subset of coarse channels
So all that would be needed is an interface ontop of the coarse channel selection that works off of antenna selection (i.e. still a new flag, reusing old internals). I can see it
I don't see why the file would have to be read any more times for antenna selection.
This still open. Anyway I could help with this?
@dmkunsman I have implemented it in 2 ways:
#9 provides -a ANT
which enables selective processing of a single antenna's channels. There exists an understanding that there may be a way to bash script parallel rawspec calls to process each antenna within the RAW data. The PR has received a first round of review from @david-macmahon.
#11 provides -s
which instructs rawspec to split the output by antenna, and write to individual files.
I tested the selective antenna processing by generating a RAW file where only one antenna's values were non-zero (populated with random bytes). I tested the split-output by matching the filterbank files against those generated by selective antenna processing.
I believe both are in working order, but perhaps you'd be willing to further verify. Other than that, I think #9 and #11 are just awaiting review->refine->merge.
This is the RAW GUPPI file I used...
6fab771 closes this too.