search

IQcollaboratory / galpopFM

forward modeling galaxy populations
MIT License
1 stars 0 forks source link

SSFR model reparameterization #47

Open changhoonhahn opened 4 years ago

changhoonhahn commented 4 years ago

As of 10/21/2020, the model is parameterized using M and SFR. However, I am concerned that parameterizing based on SFR is imposing an unwanted M dependence on the dust attenuation. Here's how the galaxy subpopulations map onto the optical and UV color-magnitude relation without any dust attenuation where I've color-coded galaxy populations with different SSFRs

image

Galaxies with similar SSFRs have similar colors. Therefore, I've included new dust models parameterized by SSFR a383d182f27ebf42491d2a414150fd073cebf58f.

changhoonhahn commented 4 years ago

Here are the preliminary color-magnitude relations from the ABC posteriors from the new SSFR parameterization: image

Overall, the SSFR parameterization is able to reproduce observations at a similar level as the SFR parameterization. Since SSFR is easier to interpret, let's favor SSFR over SFR.

Also, excluding nbar from the distance metric does not seem to have a significant impact on the final posteriors.

changhoonhahn commented 4 years ago

Run ABC until convergence for SSFR parameterization: