color-magnitude relation with dust model

[changhoonhahn]

Without any prescription for dust attenuation, all of the simulations struggled to reproduce the optical and UV color-magnitude relation SDSS. With the dust model at the median ABC posterior values, the simulations are able to roughly reproduce SDSS observations. Let's examine how the dust model is producing better agreement.

optical color-magnitude

Here's a plot of SFR-M relation (left) and optical color-magnitude relations without dust (center) and with dust (right) We mark galaxy populations with different SSFRs on the SFR-M diagram. Focusing on TNG and EAGLE, which have more converged posteriors:

• The color-magnitude relation is shifted overall by ~0.25 mag
• Intrinsically blue/star-forming galaxies are significantly attenuated: they are ~0.5 mag fainter.
• Without dust, star-forming, green valley, and quiescent galaxies had distinct colors, but the variation in the attenuation curves causes mixing among star-forming, green valley, and quiescent galaxies.

UV color magnitude

Now here's a plot of the SFR-M* relation (left) and UV color-magnitude relations without dust (center) and with dust (right)

• Without dust, a subpopulation of quiescent galaxies with very low SSFRs and very red UV colors. With the dust model, this quiescent galaxy population is highly attenuated. They are fainter by ~1 mag.
• Besides the low SSFR quiescent population, there's an overall reddening of the color-magnitude relation with the dust model.
• Again, the variation in the attenuation curves causes the star-forming, green valley, and quiescent galaxies to mix in UV color.

main take-away

• both the UV and optical color-magnitude relations are reddened by the dust model.
• the optical color-magnitude relation drives the significant attenuation of luminous intrinsically blue/starforming galaxies
• the UV color-magnitude relation drives the significant attenuation of low SSFR quiescent galaxies

remaining questions

• [ ] Why do quiescent galaxies with low SSFR have such red UV color?
• [x] What's the deal with a few low SSFR quiescent galaxies in TNG with blue colors?

[changhoonhahn]

The low SSFR quiescent galaxies in TNG with blue colors are caused by quiescent galaxies with flat/positive slopes.

[TStarkenburg]

If the attenuation curve is flat/positive for some low SSFR galaxies then that should be only the case for specific Mstar values, right (because in the model the slope has a Mstar and SSFR dependency and no flexibility otherwise)? Are the blue low-SSFR galaxies dwarfs?

[TStarkenburg]

BTW this may sound a bit silly, but could you double check that part of the purple population is not reddened out of the NUV-FUV range (i.e. NUV-FUV > 4), especially for SIMBA? It seems there are just way less purple points in the dust-added panels, are those dropped out because of the large overal attenuation (i.e. dropped in the magnitude axis), or because of reddening (dropped out in the color axis)?

[changhoonhahn]

If the attenuation curve is flat/positive for some low SSFR galaxies then that should be only the case for specific Mstar values, right (because in the model the slope has a Mstar and SSFR dependency and no flexibility otherwise)? Are the blue low-SSFR galaxies dwarfs?

The attenuation curve is flat/positive for low SSFR TNG galaxies with log M > 10.5. Most log M < 10.5 quiescent galaxies are attenuated out of the luminosity cut. Blue is the ~starburst population.

BTW this may sound a bit silly, but could you double check that part of the purple population is not reddened out of the NUV-FUV range (i.e. NUV-FUV > 4), especially for SIMBA? It seems there are just way less purple points in the dust-added panels, are those dropped out because of the large overal attenuation (i.e. dropped in the magnitude axis), or because of reddening (dropped out in the color axis)?

I wouldn't focus on the SIMBA plots yet. The ABC posteriors for SIMBA definitely haven't converged yet. For TNG and EAGLE, the low M* and low SSFR galaxies are attenuated out of the sample.

[TStarkenburg]

If the attenuation curve is flat/positive for some low SSFR galaxies then that should be only the case for specific Mstar values, right (because in the model the slope has a Mstar and SSFR dependency and no flexibility otherwise)? Are the blue low-SSFR galaxies dwarfs?

The attenuation curve is flat/positive for low SSFR TNG galaxies with log M > 10.5. Most log M < 10.5 quiescent galaxies are attenuated out of the luminosity cut. Blue is the ~starburst population.

Sorry, I meant the low SSFR quiescent galaxies in TNG with blue colors, but you answered that anyway. Thanks for checking the mass range. I think its interesting that a for high mass galaxies and low SSFR the attenuation curve flattens/is positive. Have to think a bit about how to interpret that.

BTW this may sound a bit silly, but could you double check that part of the purple population is not reddened out of the NUV-FUV range (i.e. NUV-FUV > 4), especially for SIMBA? It seems there are just way less purple points in the dust-added panels, are those dropped out because of the large overal attenuation (i.e. dropped in the magnitude axis), or because of reddening (dropped out in the color axis)?

I wouldn't focus on the SIMBA plots yet. The ABC posteriors for SIMBA definitely haven't converged yet. For TNG and EAGLE, the low M* and low SSFR galaxies are attenuated out of the sample.

Ok, thanks! For TNG especially, that still means an attenuation of >~2 magnitudes for that population, right?

[changhoonhahn]

comaprison to literature

Trayford+(2015)

z = 0.1 galaxies from the EAGLE simulation suite using the GALAXEV population synthesis models ... using a two-component screen model

Trayford+(2017)

EAGLE hydrodynamical simulation at redshift z = 0.1, modelling dust with the SKIRT Monte Carlo radiative transfer code

Nelson+(2018)

optical colors using a dust model that includes attenuation due to dense gas birth clouds surrounding young stellar populations and also due to simulated distribution of neutral gas and metals