meeting 2025 06 23 n57 - JacobPilawa/TriaxSchwarzschild_wiki_6 GitHub Wiki
Context
Continuing on to N57 after the N315 referee report -- bit of a work in progress as I gather material from past bullets/remind myself of whree things stood.
I've tried to summarize where we left off below/pull some of the relevant plots from the summary page.
Polynomials:
We had been experimenting a bit with the dependence on the additive and multiplicative polynomials.
It seems like N57 needs something like adeg=1, 2, or 3 for there to be "stable" results, and likewise needs mdeg>3 or so for there to be stable results as a function of multiplicative degree.
This seems to be consistent with what Emily had found for M87, where small mdegs were leading to instability. However I still don't have a good understanding of why (physically) the spectra with adeg=-1 appear to be inflated (beyond the argument where the spectra gets "squished" when you allow for the additive constant, thus broadening the features).
Sigma Clipping:
As a follow-up to some of the masking tests, I ran the sigma clipping routine Emily had on the N57 spectra. One concern we had is that the original mask on BOX might not have been the most optimal. It had appeared like a few sky lines were either not masked or over masked, so we wanted to try the sigma clipping approach.
It seems like sigma clipping = 2/2.5/3 sigma results are in decent agreement with the fiducial spectral fits, suggesting that the mask is not what is causign the issue. Note that I tried these using the Barth star subet of the library.
Libraries:
We had a bit of tensions when fitting N57 with the Barth library only vs. the "trimmed library" Emily had passed my way.
I had previously been trying to reconcile the libraries by iteratively trimming out the highest weight templates from the pPXF fits, but we decided it was likely better to take a physical approach to trimming away the templates. To that end, I've extracted the spectral types for all the stars in the library and filtered them into two sets: GKM stars only, and GKM main sequence stars only.
While it does see
Diagnostics
Additive/Multiplicative Polynomial Follow Ups
First, here's a quick test in which I vary the multiplicative polynomial in a similar way of varying the additive polynomial from last time. I ran this for both N57 and N410 using both adeg=0 and adeg=-1, but the results are quite consistent throughout. It seems like we need a multiplicative order >=3 or so (which we're currently using) for the results to stabilize. Using lower orders seems to cause quite a bit of variability (and the fits themselves were not great as I glanced at them):
Sigma as a function of adeg for the N57 spectra
Plot
Zoomed
Sigma as a function of mdeg: The plots for this test are a bit large, so they may take a minute to load.
N57
adeg=0
adeg=-1
Sigma Clipping on N57
I've also run the sigma clipping routine on the N57 spectra so we can better assess whether or not the mask is causing us any issues. It seems like there is really nice agreement between my original mask from box and the sigma clipping routine for sigma = 2.0, 2.5, 3.0.
Here's both the spectra fits and the resulting 1-to-1 plots of the kinematics:
Spectral fits using the sigma clipping routine and adeg=0
Original Mask
Sigma = 2.0
Sigma = 2.5
Sigma = 3.0
Spectral fits using the sigma clipping routine and adeg=-1
Original Mask
Sigma = 2.0
Sigma = 2.5
Sigma = 3.0
And the corresponding 1-to-1 plots (with the fiducial kinematics using Barth + no sigma clipping + adeg=0 on the x-axis):
Sigma=2.0
Sigma=2.5
Sigma=3.0
adeg=0
adeg=-1
Library Follow Up
Last time, I had been trimming away from the "trimmed library" suboptimally -- I had been discarding the highest weight spectra until I had started to see some of the Barth stars appear in the "preferred" templates, but it's probably better to trim away bad stars first, and then assess the fits.
So to do this, I've taken a closer look at the templates and their reported spectral types, and have some follow ups below.
First, here's a large plot containing all 209 stars from the trimmed library
Trimmed Library Spectra
And here's what the distribution of spectral types look like for the "trimmed" library:
Note that there are two categories which appear in the plot which are not actually spectral types. There are 10 stars which are not matched on SIMBAD (I can check these out more carefully but wanted to move forward for now), which get labelled as "N" due to being "Not Found" in the database. Theres another star that is listed as a type "kA9hF2mF2(IV)" star, which is the lowercase "k" entry.
Trimmed Library Spectral Type Distribution
And for reference, here's the Barth stars along with their spectral types:
Filename
Star
Spectral Type
0
scan399.fits
HD121146
K2IV
1
scan411.fits
HD125560
K3III
2
scan419.fits
HD129312
G7IIIa
3
scan437.fits
HD136028
K2/3III
4
scan574.fits
HD188056
K3IIICN2
5
scan603.fits
HD199580
K1IV
6
scan612.fits
HD203344
K0III
7
scan669.fits
HD221148
K3-IIIbFe2
8
scan194.fits
HD12929
K2-IIIbCa-1
9
scan216.fits
HD19476
G9.5IIIb
10
scan219.fits
HD20893
K3III
11
scan272.fits
HD49293
K0+III
12
scan274.fits
HD51440
K2III
13
scan286.fits
HD58207
G9IIIb
14
scan304.fits
HD69267
K4IIIBa0.5:
ONLY GKM Stars
My first pass at trimming this library is to select only the G/K/M stars but include all non-main sequence stars:
Here's what the actual spectra look like for this subset:
GKM Spectra
And here's the spectral type distribution for this case:
GKM Stars Spectral Type Distribution
And using only these GKM stars for the spectral fitting, here is the resulting 1-to-1 plot:
Barth vs. GKM Stars
ONLY GKM Main Sequence Stars
As a second pass, I've excluded non-main sequence stars from the library (which isn't necessarily what we should do, but I wanted to test this anyway):
Here's what the actual spectra look like for this subset:
GKM Main Sequence Spectra
And here's the spectral type distribution for this case:
GKM Main Sequence Stars Spectral Type Distribution
And using only these GKM Main Sequence Stars, here's the resulting 1-to-1 plot of the kinematics:
