meeting 2025 03 26 n315 - JacobPilawa/TriaxSchwarzschild_wiki_5 GitHub Wiki

Context

  • Cleaning up some results/open questions from our last meeting. Specifically we wanted to see the effect of the starting guess on the N315 spectra. In short, it seems like adjusting the starting guess fixes the issue we had with three of the N315 bins. Correcting this doesn't appreciably differ from our currrent results, obtained by running a subset of the Barth library as the templates.

Diagnostics

  • The first thing we wanted to check was the impact of the starting guess for the three elevated-sigma bins from our reduction. As a quick reminder, we found that 3 of the GMOS bins preferred very large sigmas when we did the initial fitting of the pPXF spectra, but this was assuming a sigma starting guess of 1 km/s. I've updated the code to specify a starting guess of (0, 200) km/s for V and sigma, and this fixed the fitting issue without having to modify the template library.

  • First, here's a quick comparison of the spectra for the two starting guesses:

Guess = (0,1) km/s Guess = (0,1) km/s, subset of Barth Library Guess = (0, 200) km/s
Bin 148 [images/250326/og_pPXF_plot_00148-1.png]]](/JacobPilawa/TriaxSchwarzschild_wiki_5/wiki/[[images/250326/sub_pPXF_plot_00148-1.png) images/250326/pPXF_plot_00148-1.png
Bin 238 [images/250326/og_pPXF_plot_00238-1.png]]](/JacobPilawa/TriaxSchwarzschild_wiki_5/wiki/[[images/250326/sub_pPXF_plot_00238-1.png) images/250326/pPXF_plot_00238-1.png
Bin 247 [images/250326/og_pPXF_plot_00247-1.png]]](/JacobPilawa/TriaxSchwarzschild_wiki_5/wiki/[[images/250326/sub_pPXF_plot_00247-1.png) images/250326/pPXF_plot_00247-1.png

  • So one question is what to do about the three strange bins now that adjusting the starting guess seems to fix the issue. I personally think we're fine to remove that statement from the paper since rerunning pPXF with only an updated guess gives virtually identical kinematics. Below I've included the full kinematics from the three bins, and you can see things are extremely consistent:

  • Here's the original kinematics which are produced by limiting the template library to 13/15 of the Barth stars:

Bin v v_err sigma sigma_err h3 h3_err h4 h4_err h5 h5_err h6 h6_err h7 h7_err h8 h8_err
149 -23.995 4.634 328.224 4.567 0.031 0.011 0.021 0.011 0.002 0.010 0.048 0.014 0.047 0.013 -0.061 0.012
239 -1.316 5.357 334.215 5.397 -0.003 0.011 0.030 0.011 0.047 0.010 -0.042 0.015 0.006 0.018 0.067 0.014
248 1.665 5.867 324.947 5.504 0.029 0.015 0.013 0.014 0.071 0.014 0.001 0.019 -0.049 0.017 0.022 0.016
  • And the kinematics produced by updating the initial guess:
Bin v v_err sigma sigma_err h3 h3_err h4 h4_err h5 h5_err h6 h6_err h7 h7_err h8 h8_err
149 -21.41717 4.52668 326.38801 4.33452 0.02728 0.01058 0.01647 0.01022 -0.00263 0.01023 0.05222 0.01403 0.04954 0.01326 -0.06113 0.01180
239 3.34680 5.43670 333.08982 5.19644 -0.00684 0.01166 0.03129 0.01134 0.04587 0.01057 -0.04548 0.01484 0.01131 0.01647 0.06748 0.01428
248 4.11479 5.57898 325.20928 5.14857 0.02576 0.01363 0.01038 0.01284 0.06936 0.01266 -0.00724 0.01790 -0.04269 0.01509 0.02607 0.01561
  • Thus it seems like we're fine to report what we currently have and leave out the discussion in the text of the pPXF subset vs. non-subset of the Barth star fits.

Original Kinematics vs. Corrected Kinematics Plot

  • One thing we also wanted to have on the Github for future reference is a comparison of the unsymmetrized N315 kinematics (corrected) vs. the original set of kinematics used in the minimizations. Note that the "corrected" kinematics here are the case where we fit almost all the spectra with the full Barth library except for the three stars which had elevated sigma values. For these cases, we trimmed out the two templates which were contributing to the weird spectral fits, and this seemed to fix the issue.
  • Here's the final (non-point symmetric, non-median subtracted V):
Plot
images/250326/unsym_comparison.png