mirkwood: Fast and Accurate SED Modeling Using Machine Learning

Gilda, Sankalp; Lower, Sidney; Narayanan, Desika

doi:10.3847/1538-4357/ac0058

“…In this section we perform a series of sanity checks of our simulated galaxies by examining their SFRs, masses (stellar and molecular), and metallicities in the context of known empirical correlations between these quantities. Before doing so, we stress that observed integrated properties such as stellar and molecular gas masses are likely to be uncertain by at least a factor of two (Lower et al 2020;Gilda et al 2021).…”

Section: Integrated Propertiesmentioning

confidence: 99%

Tracing Molecular Gas Mass in z ≃ 6 Galaxies with [C ii]

Vizgan

¹

,

Greve

²

,

Olsen

³

et al. 2022

ApJ

Self Cite

View full text Add to dashboard Cite

We investigate the fine-structure [C ii] line at 158 μm as a molecular gas tracer by analyzing the relationship between molecular gas mass (M mol) and [C ii] line luminosity (L [C II]) in 11,125 z ≃ 6 star-forming, main-sequence galaxies from the simba simulations, with line emission modeled by the Simulator of Galaxy Millimeter/Submillimeter Emission. Though most (∼50%–100%) of the gas mass in our simulations is ionized, the bulk (>50%) of the [C ii] emission comes from the molecular phase. We find a sublinear (slope 0.78 ± 0.01) log L [ C II ] – log M mol relation, in contrast with the linear relation derived from observational samples of more massive, metal-rich galaxies at z ≲ 6. We derive a median [C ii]-to-M mol conversion factor of α [C II] ≃ 18 M ⊙/L ⊙. This is lower than the average value of ≃30 M ⊙/L ⊙ derived from observations, which we attribute to lower gas-phase metallicities in our simulations. Thus, a lower, luminosity-dependent conversion factor must be applied when inferring molecular gas masses from [C ii] observations of low-mass galaxies. For our simulations, [C ii] is a better tracer of the molecular gas than CO J = 1–0, especially at the lowest metallicities, where much of the gas is CO-dark. We find that L [C II] is more tightly correlated with M mol than with star formation rate (SFR), and both the log L [ C II ] – log M mol and log L [ C II ] – log SFR relations arise from the Kennicutt–Schmidt relation. Our findings suggest that L [C II] is a promising tracer of the molecular gas at the earliest cosmic epochs.

show abstract

“…In this section we perform a series of sanity checks of our simulated galaxies by examining their star formation rates, masses (stellar and molecular), and metallicities in the context of known empirical correlations between these quantities. Before doing so, we stress that observed integrated properties such as stellar and molecular gas masses are likely to be uncertain by at least a factor of two (Lower et al 2020;Gilda et al 2021).…”

Section: Integrated Propertiesmentioning

confidence: 99%

Tracing Molecular Gas in z $\simeq$ 6 Galaxies with [C${\rm \scriptsize II}$]

Vizgan,

Greve,

Olsen

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

We investigate the fine-structure [CII] line at 158 µm as a molecular gas tracer by analyzing the relationship between molecular gas mass (M mol ) and [CII] line luminosity (L [CII] ) in 11,125 z 6 starforming, main sequence galaxies from the simba simulations, with line emission modeled by Sígame. Though most (∼ 50−100 %) of the gas mass in our simulations is ionized, the bulk (> 50 %) of the [CII] emission comes from the molecular phase. We find a sub-linear (slope 0.78 ± 0.01) log L [CII] − log M mol relation, in contrast with the linear relation derived from observational samples of more massive, metalrich galaxies at z < ∼ 6. We derive a median [CII]-to-M mol conversion factor of α [CII] 18 M /L . This is lower than the average value of 30 M /L derived from observations, which we attribute to lower gas-phase metallicities in our simulations. Thus, a lower, luminosity-dependent, conversion factor must be applied when inferring molecular gas masses from [CII] observations of low-mass galaxies. For our simulations, [CII] is a better tracer of the molecular gas than CO J = 1 − 0, especially at the lowest metallicities, where much of the gas is 'CO-dark'. We find that L [CII] is more tightly correlated with M mol than with star-formation rate (SFR), and both the log L [CII] − log M mol and log L [CII] − log SFR relations arise from the Kennicutt-Schmidt relation. Our findings suggest that L [CII] is a promising tracer of the molecular gas at the earliest cosmic epochs.

show abstract

“…deriving the galaxy and dust properties from a SED. Such models can be found in Lovell et al (2019); Dobbels et al (2020); Gilda et al (2021). This work also gives an example of using the deep learning model in SED-fitting.…”

Section: Introductionmentioning

confidence: 98%

Starduster: A multi-wavelength SED model based on radiative transfer simulations and deep learning

Qiu,

Kang

2021

Preprint

0

View full text Add to dashboard Cite

We present starduster, a supervised deep learning model that predicts the multi-wavelength SED from galaxy geometry parameters and star formation history by emulating dust radiative transfer simulations. The model is comprised of three specifically designed neural networks, which take into account the features of dust attenuation and emission. We utilise the skirt radiative transfer simulation to produce data for the training data of neural networks. Each neural network can be trained using ∼ 4000 − 5000 samples. Compared with the direct results of the skirt simulation, our deep learning model produces 0.1 − 0.2 mag errors in FUV to FIR wavelengths. At some bands, the uncertainty is only 0.01 mag. As an application, we fit our model to the observed SEDs of IC4225 and NGC5166. Our model can reproduce the observations, and successfully predicts that both IC4225 and NGC5166 are edge-on galaxies. However, the predicted geometry parameters are different from image-fitting studies. Our analysis implies that the inconsistency is mainly due to the degeneracy in the star formation history of the stellar disk and bulge. In addition, we find that the predicted fluxes at 20 µm − 80 µm by our SED model are correlated with bulge radius. Our SED code is public available and can be applied to both SED-fitting and SED-modelling of galaxies from semi-analytic models.

show abstract

mirkwood: Fast and Accurate SED Modeling Using Machine Learning

Cited by 23 publications

References 98 publications

Tracing Molecular Gas Mass in z ≃ 6 Galaxies with [C ii]

Tracing Molecular Gas Mass in z ≃ 6 Galaxies with [C ii]

Tracing Molecular Gas in z $\simeq$ 6 Galaxies with [C${\rm \scriptsize II}$]

Starduster: A multi-wavelength SED model based on radiative transfer simulations and deep learning

Contact Info

Product

Resources

About