A scalable Bayesian framework for galaxy emission line detection and redshift estimation

Abstract

Estimating galaxy redshifts is crucial for constraining key physical quantities like those in the equation of state of dark energy. Modern telescopes such as the James Webb Space Telescope, the Euclid Space Telescope, and the NASA Nancy Grace Roman Space Telescope are producing massive amounts of spectroscopic data that enable precise redshift estimation. However, a galaxy's redshift can be estimated only when emission lines are present in the observed spectrum, which is unknown a priori. A novel Bayesian approach to estimating redshift and simultaneously testing for the presence of emission lines is developed. Although modern spectroscopic surveys involve millions of spectra and give rise to highly multimodal posterior distributions, the proposed framework remains computationally efficient, admitting a parallelizable implementation suitable for large-scale inference.

Figures (29)

• Figure 1: Discrete approximation to the posterior marginal density (grid resolution of 0.001) for two simulated datasets, one with a strong line present, and the other with no lines present. Both are normalized so that the resulting probability mass functions sum to one.
• Figure 2: An observed galaxy spectrum from JWST. The data are binned corresponding to the flat rectangular nature of the plot. The grey dotted lines represent $+/-$ two standard deviations around the observed flux values using the reported measurement errors from JWST.
• Figure 3: A comparison of observed emission line locations at redshift 0 (top) and redshift 2 (bottom). The light blue rectangles represent the observable range of the JWST detector. The OII, OIII, and SII doublets are visualized as single lines for simplicity.
• Figure 4: Discrete approximation to the posterior marginal density for spectrum 00004. The dashed orange line indicates the redshift estimate provided by astronomers.
• Figure 5: A posterior predictive plot assessing model fit for spectrum 00004 in Figure \ref{['fig:spec']}. The black dots show the observed flux values per wavelength. Here, 5000 replicate datasets are drawn from the posterior predictive distribution. The pointwise median per wavelength is shown in dark blue, and the pointwise 2.5% and 97.5% pointwise quantiles as the lower and upper bounds of the light blue region, respectively.