meeting 2025 08 20 n410 - JacobPilawa/TriaxSchwarzschild_wiki

Context

Some brief tests on the mask that came up in the meeting yesterday, testing the impact of our sigma clipping routine on the kinematics.
Adeg=0 vs. Adeg=-1
- I plotted the fiducial Barth h8 results which use adeg=0 and adeg=-1 (up to now, adeg=0 has been our default). We are seeing a similar trend that we saw with the other galaxies -- the adeg=-1 is systemtatically higher than the adeg=0 cases, depsite the fits not being appreciably better/worse fit. We saw this in the plots from the meeting too, which showed that turning the additive polynomial on in general brought sigma to lower values, though there was a bit of variability when increasing from 0-->1-->2, and so on. Without any better choice, I think sticking to adeg=0 makes the most sense.
Mask Testing
- It doesn't appear like our mask is significantly influencing our kinematics. I've compared our fiducial mask to the no mask case, as well as both the 2.0 and 3.0 sigma clipping case, and the preliminary set of fits all appear to be in rather decent agreement. All the sigma clipping results initially assume no mask, whereas in the past I had used the fiducial BOX mask when running our sigma clipping routine (which also suggested that our mask was not appreicably changing our results).
- Note that the sigma clipping = 3.0 case has very few pixels that end up being masked which explains why the agreement is so nice.
- I think to be consistent with the published kinematics, it makes the most sense to keep the mask from BOX, maybe adding additional masks from 8750 onward.

adeg=0 vs. adeg=-1

To start things off, I've plotted the results from the adeg=-1 vs. the adeg=0 case (it appears like I hadn't run the adeg=-1 case up until now, in part since I had been mostly trying to reproduce what Irina has in her paper).

one-to-one	differences

And here are the RMS vs spectrum number plots to get a sense of which is "better" fit:

adeg=0	adeg=-1

Mask Followups

I've played around with the mask from BOX and our sigma clipping routine to assess how much the mask changes the resulting kinematics. Up to now, I've been using the mask on BOX which is simply a binary 0/1 for whether or not a pixel should be includeed.

	Fiducial Mask vs. No Mask	Fiducial vs. Sigma Clip 2.0	Fiducial vs. Sigma Clip 3.0	No Mask vs. Sigma Clip 2.0	No Mask vs. Sigma Clip 3.0
One to One
Differences

And here are all the spectral fits from the cases above to get a sense of the fiducial mask locations vs. where sigma clipping seems to find places to trim away:
- Note that the sigma clipped versions of the results start without a mask. This is in contrast to tests I've done in the past in which I include the fiducial mask and have pPXF find the pixels to trim away after a fiducial fit with the mask.

Click to expand for spectral fits for different masks