search

EoRImaging / eppsilon

eppsilon - error propagated power spectrum with interleaved observed noise
BSD 2-Clause "Simplified" License
5 stars 4 forks source link

Dr. Seuss's revenge #31

Closed nicholebarry closed 6 years ago

nicholebarry commented 6 years ago

There appears to be a low-level Dr. Seuss effect still happening, similar to what Matt saw awhile ago. I tried generating a simulation of bubbles (and also of hash, though harder to see there) and can see the Dr. Seuss effect pretty clearly at high k, shown below. fhd_nb_sim_bubble_only_obs_id_6296_averemove_bh_dencorr_2dkpower So, I checked the frequencies for round-off error. There is round-off error, but pretty minor stuff at floating-point level. Should I pursue the frequencies further, or does anyone else have a better idea?

I've been talking to Adam Lanman...it appears that he sees it in his sims as well, which use all 384 input frequencies (instead of averaging down to 192 like I have been).

Need this for the thesis, so help is greatly appreciated.

aelanman commented 6 years ago

Just to clarify --- This hasn't come up in my simulations, but only when I tried to run firstpass on some data.

nicholebarry commented 6 years ago

This was a settings issue. I had some flagged frequencies that crept through, which I successfully squashed with keywords. The amount of contamination seen in the harmonics is proportional to the intrinsic foregrounds, so it was showing up at really low levels in my simulation.