| Note . | Parameter . | Description . | Prior . |
|---|---|---|---|
| 1 | zred | Redshift | Uniform: zspec ± 0.01 |
| 2 | log (M/M⊙) | Total mass formed | MZR: clipped normal, min = 8, max = 15 |
| 3 | |$\hat{\tau }_{\lambda ,2}$| | Diffuse dust optical depth | Uniform: min = 0, max = 4 |
| 4 | log (Z/Z⊙) | Stellar metallicity | MZR: clipped normal, min = −2, max = 0.19 |
| 5 | |$\log \left(\frac{\mathrm{SFR}(t)}{\mathrm{SFR}(t+\Delta t)}\right)$| | Ratio of the SFR ratios in adjacent age bins | Student t: μ = 0, σ = 0.3, 2 dof |
| 6 | spec_norm | Normalization of the spectra | Uniform: min = 0, max = 100 |
| 7 | p1, p2, p3 | Continuum shape correction polynomial coefficients | Uniform: min = −0.1/(n + 1), max = 0.1/(n + 1) |
| 8 | spec_jitter | Spectra white noise model | Uniform: min = 1, max = 3 |
| 9 | |$f_\mathrm{outlier,\, spec}$| | Spectra outlier fraction | Uniform: min = 10−5, max = 0.5 |
| Note . | Parameter . | Description . | Prior . |
|---|---|---|---|
| 1 | zred | Redshift | Uniform: zspec ± 0.01 |
| 2 | log (M/M⊙) | Total mass formed | MZR: clipped normal, min = 8, max = 15 |
| 3 | |$\hat{\tau }_{\lambda ,2}$| | Diffuse dust optical depth | Uniform: min = 0, max = 4 |
| 4 | log (Z/Z⊙) | Stellar metallicity | MZR: clipped normal, min = −2, max = 0.19 |
| 5 | |$\log \left(\frac{\mathrm{SFR}(t)}{\mathrm{SFR}(t+\Delta t)}\right)$| | Ratio of the SFR ratios in adjacent age bins | Student t: μ = 0, σ = 0.3, 2 dof |
| 6 | spec_norm | Normalization of the spectra | Uniform: min = 0, max = 100 |
| 7 | p1, p2, p3 | Continuum shape correction polynomial coefficients | Uniform: min = −0.1/(n + 1), max = 0.1/(n + 1) |
| 8 | spec_jitter | Spectra white noise model | Uniform: min = 1, max = 3 |
| 9 | |$f_\mathrm{outlier,\, spec}$| | Spectra outlier fraction | Uniform: min = 10−5, max = 0.5 |
Note. 1 – Spectroscopic redshift. 2 – Total mass is the sum of total stellar mass and mass lost to outflows; see note 3 for a comment on the prior. 3 – We assume a Milky Way extinction curve (Cardelli et al. 1989). 4 – We assume a prior on the stellar mass–metallicity relation (MZR) according to the local trend reported by Gallazzi et al. (2005), where we add the systematic offset between parametric and non-parametric stellar mass estimates (see Appendix C). 5 – Ratio of the SFRs in adjacent bins of the 10-bin non-parametric SFH. The age bins are spaced in lookback time: 0, 30, 100, 500 Myr, and 1 Gyr, five equally spaced bins, and lastly 0.95× the age of our Universe at the observed redshift. For N age bins, there are N − 1 free parameters. 6 – The normalization of the spectra is a free parameter to account for systematics in the relative flux calibration. 7 – The shape of the spectral continuum can be adjusted by a third degree Chebyshev polynomial to account for systematics in the relative flux calibration. 8 – The uncertainty on the spectra can be increased by a given factor, with a likelihood penalty for factors giving reduced χ2 < 1. 9 – An outlier pixel model can increase the errors for individual pixels by a factor of 50, to accommodate for poor matches between the data and spectral templates.
| Note . | Parameter . | Description . | Prior . |
|---|---|---|---|
| 1 | zred | Redshift | Uniform: zspec ± 0.01 |
| 2 | log (M/M⊙) | Total mass formed | MZR: clipped normal, min = 8, max = 15 |
| 3 | |$\hat{\tau }_{\lambda ,2}$| | Diffuse dust optical depth | Uniform: min = 0, max = 4 |
| 4 | log (Z/Z⊙) | Stellar metallicity | MZR: clipped normal, min = −2, max = 0.19 |
| 5 | |$\log \left(\frac{\mathrm{SFR}(t)}{\mathrm{SFR}(t+\Delta t)}\right)$| | Ratio of the SFR ratios in adjacent age bins | Student t: μ = 0, σ = 0.3, 2 dof |
| 6 | spec_norm | Normalization of the spectra | Uniform: min = 0, max = 100 |
| 7 | p1, p2, p3 | Continuum shape correction polynomial coefficients | Uniform: min = −0.1/(n + 1), max = 0.1/(n + 1) |
| 8 | spec_jitter | Spectra white noise model | Uniform: min = 1, max = 3 |
| 9 | |$f_\mathrm{outlier,\, spec}$| | Spectra outlier fraction | Uniform: min = 10−5, max = 0.5 |
| Note . | Parameter . | Description . | Prior . |
|---|---|---|---|
| 1 | zred | Redshift | Uniform: zspec ± 0.01 |
| 2 | log (M/M⊙) | Total mass formed | MZR: clipped normal, min = 8, max = 15 |
| 3 | |$\hat{\tau }_{\lambda ,2}$| | Diffuse dust optical depth | Uniform: min = 0, max = 4 |
| 4 | log (Z/Z⊙) | Stellar metallicity | MZR: clipped normal, min = −2, max = 0.19 |
| 5 | |$\log \left(\frac{\mathrm{SFR}(t)}{\mathrm{SFR}(t+\Delta t)}\right)$| | Ratio of the SFR ratios in adjacent age bins | Student t: μ = 0, σ = 0.3, 2 dof |
| 6 | spec_norm | Normalization of the spectra | Uniform: min = 0, max = 100 |
| 7 | p1, p2, p3 | Continuum shape correction polynomial coefficients | Uniform: min = −0.1/(n + 1), max = 0.1/(n + 1) |
| 8 | spec_jitter | Spectra white noise model | Uniform: min = 1, max = 3 |
| 9 | |$f_\mathrm{outlier,\, spec}$| | Spectra outlier fraction | Uniform: min = 10−5, max = 0.5 |
Note. 1 – Spectroscopic redshift. 2 – Total mass is the sum of total stellar mass and mass lost to outflows; see note 3 for a comment on the prior. 3 – We assume a Milky Way extinction curve (Cardelli et al. 1989). 4 – We assume a prior on the stellar mass–metallicity relation (MZR) according to the local trend reported by Gallazzi et al. (2005), where we add the systematic offset between parametric and non-parametric stellar mass estimates (see Appendix C). 5 – Ratio of the SFRs in adjacent bins of the 10-bin non-parametric SFH. The age bins are spaced in lookback time: 0, 30, 100, 500 Myr, and 1 Gyr, five equally spaced bins, and lastly 0.95× the age of our Universe at the observed redshift. For N age bins, there are N − 1 free parameters. 6 – The normalization of the spectra is a free parameter to account for systematics in the relative flux calibration. 7 – The shape of the spectral continuum can be adjusted by a third degree Chebyshev polynomial to account for systematics in the relative flux calibration. 8 – The uncertainty on the spectra can be increased by a given factor, with a likelihood penalty for factors giving reduced χ2 < 1. 9 – An outlier pixel model can increase the errors for individual pixels by a factor of 50, to accommodate for poor matches between the data and spectral templates.
This PDF is available to Subscribers Only
View Article Abstract & Purchase OptionsFor full access to this pdf, sign in to an existing account, or purchase an annual subscription.
