search

absey22 / SpectroZ

Spec1d/2d Pipelines
0 stars 0 forks source link

Sz v. Pz #11

Open absey22 opened 7 years ago

absey22 commented 7 years ago

Rudimentary plots, can redo in matplotlib . These are to highest confidence. 454 18

anjavdl commented 7 years ago

Well done!!! :-)

And yikes! I guess this showcases why we cannot just use the "z_best" estimate. I think we will need to compare to the full p(z). In the interest of time, I can do that if you send me list of object IDs, spec-z's, etc.

What are the different colors in the plots?

absey22 commented 7 years ago

:) I threw in the masks all together per cluster, so colors are masks. I hope that's not fundamentally invalid. (Interesting, does the density at z~0.8 have any relevance? Unless that just reflects the choice behind cluster.)

I can send you those, however laptop top is dead. As soon as I get out of class, I will. Is csv format ok by you?

I maybe want to check my z-matching scheme again...

anjavdl commented 7 years ago

So are these all Q=4 ?

We do expect the peak of the distribution to be around ~0.8, so so far, so good.

Maybe I'll start with matching up the catalogs as a cross-check.

absey22 commented 7 years ago

Yes, Q=4 for those two. I had made a total, but it corrupted. some incorrect GNUplot syntax or something.

Ok, that's good to know about the relative density. I am fairly confident in my scheme (for M0454, the spectroz outlier z~1.2 did get matched to a corresponding photoz outlier.) I checked the process midway through from obj_info.fits again; the SeqNr was indeed matching the proper slitno.

anjavdl commented 7 years ago

The "serendipitous" redshifts need to be included in the list, as well. In those cases, we don't know which object was photo-z selected, so they need to be ignored for now.