search

ratt-ru / CubiCal

A fast radio interferometric calibration suite.
GNU General Public License v2.0
18 stars 13 forks source link

disable PA (de)rotation for (some) dual-corr modes #389

Open o-smirnov opened 4 years ago

o-smirnov commented 4 years ago

PA [de]rotation of RR+LL data is a no-op, so it should just ignore itself.

PA derotation of XX+YY data is not well-defined and should throw an error.

The only dual-corr regime where I see any point bothering with PA is when we have a full-pol model (i.e. an LSM not a column), and we want to compute a rotated version of that to calibrate XX+YY data against. Derotation should still not be allowed.

IanHeywood commented 4 years ago

I just saw the remark about this being a pointless memory guzzler. Is there an argument for it being disabled in the default parset?

o-smirnov commented 4 years ago

I guess there is. 99% of the time you're not doing polarimetry, right?

IanHeywood commented 4 years ago

100% of the time if I can get away with that.

o-smirnov commented 4 years ago

image

IanHeywood commented 4 years ago

100% of the time I expected that to be your "You may not be interested in the polarisation..." slide.

o-smirnov commented 4 years ago

That happens the other 100% of the time (which also explains CASA's flagging statistics...)