search

willettk / rgz-analysis

Analysis pipeline for Radio Galaxy Zoo
MIT License
3 stars 5 forks source link

Repeats of individual FIRST components within catalogue #35

Open ivyw opened 8 years ago

ivyw commented 8 years ago

Hi Kyle (& Avery),

Thanks again for your help.

All the work you see here is based on the catalogue (rgz_75 which has about 33,000 lines) used for the WISE color-color map in Paper 1. Using the same selection as Paper 1, I am able to produce the same map as per the paper without a problem. Once I filter out all the rgz_75 sources where the SNR >5 for W1, W2 & W3 and matched to GalaxyZoo 1 catalogue ; I am left with ~1100 sources. I then created an ascii catalogue of these ~1100 sources and ran a simple "find_repeats.pl" perl script which checks for a repeat in the FIRST ID and found 30 repeated entries.

As a result of this, this was how we ended up with a "dusty spiral" peak in the color-color plot.

Therefore I'm wondering if you or Avery can check if your the current catalogue contains repeats of the same FIRST components?

Thanks heaps, Ivy

willettk commented 8 years ago

Hi @ivyw -

Sorry for taking a couple days on this. I now have data, so let's see if they match!

Short version: I've looked again at the current version of the catalog on the UMN servers. I looked at duplicate entries both by doing cone searches around the position and for identical WISE sources. For radio sources that had both a WISE and SDSS counterpart, about 10% have the same WISE counterpart for two different radio sources in the list. This is expected behavior if either:

I then made the same versions of the WISE color-color that you described above, matching to Galaxy Zoo 1 (spec-z galaxies) with a 6" radius for morphologies, requiring SNR>5 in W1,W2,W3 and 75% consensus. I did this for both the older catalogue (rgz_75) that you've been using and the most updated version. I then split the results into:

Here are the results for the older set of data that you've been using:

gz1_olddata

And here is the most recent version, including RGZ data as of this week (12 Apr 2016):

gz1_newdata

So: my analysis of the data does show a strong split in the WISE colors based on the optical morphology of the hosts; ellipticals have very blue (W2-W3) colors and are clustered more towards "normal galaxies". Spirals have much redder (W2-W3) and are mixed in where we'd expect LIRGs and starbursts. It'd fit with these being star-forming galaxies and picking up thermal emission rather than the AGN.

However, there are a couple puzzling things:

  1. I tried to duplicate your selection method exactly, but I get 2214 sources using the old file, as opposed to the ~1100 that you quoted. Why are we getting different numbers?
  2. I don't see evidence that the duplicates are dominating selection of the dusty spirals. I will note that the newer data does seem to have more total RGZ sources in the lower right corner than the older version did (see below). Given the random selection of sources, that's something I don't understand.

compare_density

ivyw commented 8 years ago

The matching is certainly very strange. If I did a straight TOPCAT match to 6 arcsecs between the initial catalogue that you gave me to GZ1's Table 2, I only get 1997 matches. This is before the WISE SNR cull and is a simple positional match between GZ1 and the WISE coordinates. I am very puzzled too. My suspicion is that my version of RGZ_75 may be different from the older one that you used.

In terms of the duplicates, the dusty spiral population is the peak that you see around (W2-W3)=1. You see a fairly large population of these in your first set of panels too. Do you mind posting the fits version of the latest RGZ catalogue (as of April 12) in the fits format so that I can check these again?

Thanks again, Ivy

ivyw commented 8 years ago

Hi Kyle, I've just shared a Dropbox link containing the version of the catalogue that I was using. Could you share your latest catalogue through Dropbox too?
Thanks. cheers, Ivy