meeting 2025 04 15 n57 - JacobPilawa/TriaxSchwarzschild_wiki_5 GitHub Wiki
Context
-
Following up from the meeting yesterday, starting off with the GPR and dynesty results after removing the outer 4 bins.
-
Brief Takeaways:
- After removing the outer 4 Mitchell bins from the chi2 calculation, the resulting rho0 and M/L are most impacted but still within the original 1 sigma region. Specifically M/L shifts from ~1.67 with all bins to ~1.77 after removing the outer 4 bins, thouhg the uncertainty on M/L is ~0.1 or so. Rho0 shifts from ~3.1e9 to ~2.9e9. The BH shits from ~9.4 billion to ~9.0 billion, again well within the uncertainty of ~+/- 1 billion or so. The shapes appear to be hardly changed.
-
Moving forward:
- If we're still concerned about these outer 4 Mitchell bins and treating them properly, we could consider reminimizing our latest grid (Grid C) base models after masking the outer 4 properly. The Grid C base models alone resulted in GPR and dynesty posteriors that agreed with all the models, so this could be a quick and cheap sanity check that removing these outer bins is comparable to masking them properly.
Diagnostics
- First, here's a quick side-by-side of the 1d panels with and without the outer 4 bins. These are the NNLS chi2's without dummies.
| All Bins | No Outer 4 Bins |
|---|---|
| [images/250415/all.png]]](/JacobPilawa/TriaxSchwarzschild_wiki_5/wiki/[[images/250415/no_outer_4.png) |
- And here's a quick comparison of the GPR and dynesty results for K=40,60,80, and all using nu=1.5. I've included a vertical plot first followed by all of the posteriors.
- In general there is really nice agreement between the "all bins" and "no outer 4" cases here. There's a bit of movement in rho0 and M/L which I think is to be expected since we're removing our outermost kinematic bin (likely most influenced by rho/the DM halo).
- With that said, all of the parameters are consistent within the 1 sigma confidence intervals of our fiducial case.
| Vertical Plot |
|---|
| images/250415/vertical_plot_no_Outer_4.png |
- And the actual posteriors:
| K=40 | K=60 | K=80 | |
|---|---|---|---|
| All Bins | [images/250410/250410_grids124_best_scales_grids_3_base_grid_alpha_K40_nu1.5-1.png]]](/JacobPilawa/TriaxSchwarzschild_wiki_5/wiki/[[images/250410/250410_grids124_best_scales_grids_3_base_grid_alpha_K60_nu1.5-1.png) | images/250410/250410_grids124_best_scales_grids_3_base_grid_alpha_K80_nu1.5-1.png | |
| No Outer 4 | [images/250415/250414_grids1234_best_scales_no_outer_4_grid_alpha_K40_nu1.5-1.png]]](/JacobPilawa/TriaxSchwarzschild_wiki_5/wiki/[[images/250415/250414_grids1234_best_scales_no_outer_4_grid_alpha_K60_nu1.5-1.png) | images/250415/250414_grids1234_best_scales_no_outer_4_grid_alpha_K80_nu1.5-1.png |
Radial Plots after Emails
- I've plotted the two lowest chi2 models (the points at ~9.7e9 and ~11.5e9 which have chi2=1614.9 and 1615.7 respectively.
- The two sets of kinematics are extremely similar -- I could maybe see the "best fitting" model below actually having a slightly worse fit to the outer 4 Mitchell bins, but that's just me trying to find a difference where there might not be one.
| Best Fitting Model | 2nd Best Fitting Model |
|---|---|
| [images/250415/best_radial.png]]](/JacobPilawa/TriaxSchwarzschild_wiki_5/wiki/[[images/250415/2nd_best_radial.png) |