Z < 0.01 Filter:

• To start, I've removed the z<0.01 filter that I had placed on the galaxies, so we now have that "low mass" tail of sources we had seen before. I've also tracked down the MASSIVE galaxies inside of this "no-cut" catalog, and as we had seen before, we have 95/115 of the MASSIVE galaxies present in our data.
  • The remaining 20 galaxies are missing, likely due to lack of detection in either the W3/W4 channels, or more likely the SuperCOMSOS I band values. When the parent samples were being created, non-uniform sensitivity and sky coverage of these channels limited which sources were detected.
    • For NGC 410 for example, I searched through all sources within 1 arcmin of NGC 410's coordinates, which returned this table:
    • In the 2MPZ paper, they say that they only consider sources with meanClass=1 (galaxies), and sources which have all quality flags less than 2048 ("no strong warnings, nor severe defects, Hambly et al. 2001b"). The issue is that the I-band data has a ton of quality errors. There are only 2 rows for which the I-band data is good, and for those two rows, there are problems with the other photometric bands. So there are no valid sources for NGC 410, and thus this doesn't get passed into 2MPZ due to the lack of I-band coverage.
    • I suspect this is true for the other MASSIVE galaxies, but it's a bit clunky to check the data in this way (since the SUPERCOMSOS data is basically a scan of photographic plates containg 1.9 billion sources).
• One other consequence of this is that the fits to the residuals in general seem "flattened" in part because of the low mass tail which generally has LM24 << W1 masses.

Joint Cappellari13-LM24 K-band Relation:

• We wanted to test how big of an effect switching to the Cappellari+13 relation would have for some of the less massive galaxies. In the LM24 paper, the value of M_k~-25 was referenced as the point beyond which LM24 does not constrain the relation, so for objects fainter than M_k = -25, I have some plots where I use the C13 relation in stead of LM24.
• The results of this is what we had expected - in general the mass difference as a function of the Cluver14 mass gets flattened in particular the less massive end.
• While I think that it's probably best to actually move forward with this approach (since using LM24 alone likely miscalibrates the low end), the results are not different enough to change any of our conclusions so far, I think.

Updated GPR Fits to the data:

• There were a few things we wanted to test with the GPR fits:
  • First, I've created a second set of GPR fits that use both a white noise kernel and larger length scales for the fit. This resulted in a much "flatter" correction, and probably more in line with what we're looking for.
  • Next, I've added a set of plots where I fit a Gaussian process (a) to each panel individually, and (b) to only the lowest z bin and extrapolated outward from there.
  • Third, I've also taken the 2D GPR fits and created a cleaner plot of their predicted surface, as well as some predictions for each individual redshift bin (in the 4 panel style). These are really nice predictions! Note that the GP assumes the midpoint of the bin, hence why the last panel looks a bit "offset" since most of the data are near the lower boundary of that redshift slice.

Extremely short takeaways:

• z < 0.01 filter has not appreciably changed any of our results; if anything it results in a flatter GPR correction prediction which is what we'd want. Removing this also guarantees that we retain as many MASSIVE galaxies as possible.
• The joint C13-LM24 K-band relation also "flattens" the residuals, but not to a large extent. I do think that moving forward, we should use this joint relation for calibration since we know it's more reliable at low masses.
• The "smoother" GPR kernel results look a bit better to me than the once we had last meeting. They are generally flatter and more "linear", and appear to agree better with the data.