pyuvdata
pyuvdata copied to clipboard
beamfits format partial read
It would be super if we could do partial reads, at least for beamfits. The use case I have in mind is our current attempt at H4C simulations, where the beam file has a whole bunch of frequencies in it, but we only simulate one frequency at a time. To run as many frequencies as possible on a cluster, we run each on one CPU via a slurm task, and we try to minimize the requested RAM. However, just reading in the beam file takes quite a bit of RAM because it reads in all the frequencies. To do the cubic spline interpolation down to a single frequency, we'd need at most I guess 4 frequencies, which would reduce RAM considerably. This is not possible currently, it seems.