making sense of outliers in EDR

[changhoonhahn]

I identified some outliers (red) quickly from the SpenderQ latent space for the EDR data:

Here's what their spectra look like:

Here's a zoom in to the LyA and LyB region:

Some of them seem like they have the wrong redshifts but others (top gray) seem pretty strange. Are any of these something interesting that I can include in the paper as a quick demo?

[changhoonhahn]

Target IDs are:

[39628243527012437, 39628506174327232, 39632945203381785,
       39628512734218381, 39633282660302938, 39633463514499370,
       39628135402046158, 39628216733794884, 39628302146603290,
       39628363156949622, 39628296643677862, 39633091861416765,
       39633082348733999]

[theabbybault]

In EDR quasars above z > ~1.8 had a known redshift underestimation bias from Redrock. See my paper here: https://arxiv.org/abs/2402.18009. The spectra with wrong redshifts could very easily be affected by this bias. I can try to check against the target IDs later. This was fixed for Year 1 but not EDR. It is probably valid enough to list in the paper a quick sentence or two pointing to this paper and explaining that some of the EDR redshifts are just wrong. I can do this if you point me to the spo

[changhoonhahn]

Oh I didn't know about that. Yes we should mention in the paper that we can confirm the redshift underestimations from your paper. Are there any that aren't redshift misestimates or systematics but something cool?

[theabbybault]

If any of those target IDs have z < 1.8 it could be something cool, but I think its most likely to be systematic.

[changhoonhahn]

Here are some other spectra that are flagged with a crude outlier flagging (farther away from the median in latent-space):

Is there anything interesting about these? Also what's the emission line between CIV and CIII?

[theabbybault]

I think that emission line is either He II or O III] (is this allowed?). Based on rest frame wavelengths from here: http://astronomy.nmsu.edu/drewski/tableofemissionlines.html