meeting 2026 03 10 gw - JacobPilawa/TriaxSchwarzschild_wiki_6 GitHub Wiki

Context

  • Following up on some things that came up in the meeting on Friday and in the meeting yesterday with Emily.

Diagnostics

  • One plot that came up in the meeting -- we wanted to look at simply Cluver14 vs. LM24 masses plotted against each other, colored by the same clouds of points we've been seeing the last few weeks:
Mass Difference vs. LM24 Mass Difference vs. Cluver Mass vs. Mass
[images/260310/mass_residuals_lm24_x_single.png]] ](/JacobPilawa/TriaxSchwarzschild_wiki_6/wiki/[[images/260310/mass_residuals_cluver14_x_single.png) images/260310/mass_residuals_single.png
  • And here's the same thing but split into panels:
Mass Difference vs. LM24 Mass Difference vs. Cluver Mass vs. Mass
[images/260310/mass_residuals_lm24_x_zbins.png]] ](/JacobPilawa/TriaxSchwarzschild_wiki_6/wiki/[[images/260310/mass_residuals_cluver14_x_zbins.png) images/260310/mass_residuals_zbins.png

GSMF Fitting

  • I'm starting to feel like our incomplete knowledge of the mass completeness might be what's hampering us here, given how large of a fraction of data are below the completeness limit of Leja, for example.

  • I've also tried a number of different changes to the GSMF fitting routine after meeting with Emily yesterday. I'm still a bit stuck here. Some of the changes/thing I've played around with:

    • Cutting my catalog so that I'm only using local (z<0.05 sources)
    • Cutting my catalog so that I'm only using the most massive sources (maybe the large number of low mass sources are throwing the fit off?)
    • Adjusting the number of Leja points per dex used as constraints
    • Adjusting the prior ranges (this is quite fruitless; minimal impact on anything compared to the other changes)
    • Adjusting the number of points being sampled from my catalog and used as observations (also very minimal impact, and difficult to raise to to a large number of galaxies without dynesty hanging up)

  • First, one thing Emily wanted to see was the (M, z) plane, or more specifically, the (M, comoving volume enclosed by that galaxy) plane:

M vs. z M vs. V
[images/260310/redshift.png]]](/JacobPilawa/TriaxSchwarzschild_wiki_6/wiki/[[images/260310/volume.png)
  • And here are some of the results of the testing I've been doing. They're not really informative, but wanted to include them here for completeness:
Case Posteriors GSMF Samples
1k points, z<0.05, with Leja constraints (20 per dex) [images/260310/260309_testing_1k_low_z_posterior.png]]](/JacobPilawa/TriaxSchwarzschild_wiki_6/wiki/[[images/260310/260309_testing_1k_low_z_gsmf.png)
1k points, z<0.05, without Leja constraints [images/260310/260309_testing_1k_low_z_no_leja_posterior.png]]](/JacobPilawa/TriaxSchwarzschild_wiki_6/wiki/[[images/260310/260309_testing_1k_low_z_no_leja_gsmf.png)
1k points, z<0.05, with Leja constraints (50 per dex) [images/260310/260309_testing_1k_low_z_adjusted_prior_100_leja_per_dex_gsmf.png]]](/JacobPilawa/TriaxSchwarzschild_wiki_6/wiki/[[images/260310/260309_testing_1k_low_z_adjusted_prior_100_leja_per_dex_posterior.png)
1k, low z, 50 leja per dex, mstar > 1e9 [images/260310/260310_1k_low_z_50_leja_per_dex_mstar_greater_than_9_posterior.png]]](/JacobPilawa/TriaxSchwarzschild_wiki_6/wiki/[[images/260310/260310_1k_low_z_50_leja_per_dex_mstar_greater_than_9_gsmf.png)
1k, low z, 50 leja per dex, mstar > 1e10 [images/260310/260310_1k_low_z_50_leja_per_dex_mstar_greater_than_10_posterior.png]]](/JacobPilawa/TriaxSchwarzschild_wiki_6/wiki/[[images/260310/260310_1k_low_z_50_leja_per_dex_mstar_greater_than_10_gsmf.png)
1k, ALL z, 50 leja per dex, mstar >1e10 [images/260310/260310_1k_50_leja_per_dex_mstar_greater_than_10_posterior.png]]](/JacobPilawa/TriaxSchwarzschild_wiki_6/wiki/[[images/260310/260310_1k_50_leja_per_dex_mstar_greater_than_10_gsmf.png)
1k, ALL z, no leja, mstar >1e10 [images/260310/260310_1k_no_leja_per_dex_mstar_greater_than_10_posterior.png]]](/JacobPilawa/TriaxSchwarzschild_wiki_6/wiki/[[images/260310/260310_1k_no_leja_per_dex_mstar_greater_than_10_gsmf.png)
1k, ALL z, no leja, mstar >1e10, adjusted prior [images/260310/260310_1k_no_leja_per_dex_mstar_greater_than_10_adjusted_prior_posterior.png]]](/JacobPilawa/TriaxSchwarzschild_wiki_6/wiki/[[images/260310/260310_1k_no_leja_per_dex_mstar_greater_than_10_adjusted_prior_gsmf.png)
  • The best result I have so far (just prior to the meeting as I play around) uses:
    • z < 0.05, Mstar > 1e10
    • No leja constraint
    • 1000 galaxies sampled
    • For now I've added a fixed 0.5 dex mass (doing some additional testing)
Case Posteriors GSMF Samples
Best [images/260310/260310_1k_low_z_no_leja_per_dex_mstar_greater_than_10_adjusted_prior_TAKE2_WITH_MASS_ADJUSTMENT_posterior.png]]](/JacobPilawa/TriaxSchwarzschild_wiki_6/wiki/[[images/260310/260310_1k_low_z_no_leja_per_dex_mstar_greater_than_10_adjusted_prior_TAKE2_WITH_MASS_ADJUSTMENT_gsmf_TEST_TEST_TEST.png)