# 21cmEMU: an emulator of 21cmFAST summary observables

**Authors:** Daniela Breitman, Andrei Mesinger, Steven Murray, David Prelogovic, Yuxiang Qin, Roberto Trotta  
**arXiv:** [2309.05697](https://arxiv.org/abs/2309.05697)  
**Date:** 2023-09-11  
**Summary page:** https://www.summarizepaper.com/en/paper/2309.05697v1/

## Key points

- Recent advancements in the study of the Epoch of Reionization (EoR) have improved understanding of galaxy and intergalactic medium properties during the early stages of the Universe.
- Traditional inference approaches for studying EoR can be time-consuming, requiring up to $10^5$ core hours.
- A team of researchers has developed 21cmEMU, an emulator that predicts summary observables from the 21cmFAST simulation code.
- 21cmEMU takes nine input parameters characterizing EoR galaxies and produces various summary statistics including:
- Mean neutral fraction of the intergalactic medium (IGM)
- Power spectrum of the 21-cm signal
- Average spin temperature of the 21-cm signal
- Sky-averaged global 21-cm signal
- Ultraviolet luminosity functions (LFs)
- Thomson scattering optical depth to the cosmic microwave background (CMB)
- The predictions made by 21cmEMU exhibit sub-percent median accuracy while reducing computational costs by over a factor of $10^4$.
- The researchers validated these inference results and demonstrated several applications such as quantifying relative constraining power between different observational datasets, assessing impacts on previous inferences, and forecasting upcoming constraints from observations made by Hydrogen Epoch of Reionization Array's sixth observing season.
- 21cmEMU is publicly available and included as an alternative simulator in a public sampler called 21CMMC, providing researchers with a valuable tool for studying and analyzing EoR observations more efficiently.

## Summary

Recent years have seen significant advancements in the study of the Epoch of Reionization (EoR), allowing for a better understanding of galaxy and intergalactic medium (IGM) properties during the early stages of our Universe. However, traditional inference approaches that require computing 3D lightcones on-the-fly can be extremely time-consuming, taking up to $10^5$ core hours. To address this issue, a team of researchers has developed 21cmEMU – an emulator that predicts several summary observables from the widely used 21cmFAST simulation code. This emulator takes nine input parameters that characterize EoR galaxies and produces various summary statistics including: (i) the mean neutral fraction of the IGM; (ii) the power spectrum of the 21-cm signal; (iii) the average spin temperature of the 21-cm signal; (iv) the sky-averaged global 21-cm signal; (vi) ultraviolet luminosity functions (LFs); and (vii) Thomson scattering optical depth to the cosmic microwave background (CMB). The predictions made by 21cmEMU exhibit sub-percent median accuracy while reducing computational costs by over a factor of $10^4$. The researchers validated these inference results and demonstrated several applications such as quantifying relative constraining power between different observational datasets, assessing how recent claims regarding a late EoR impact previous inferences, and forecasting upcoming constraints from observations made by Hydrogen Epoch of Reionization Array's sixth observing season. Notably, 21cmEMU is publicly available and has been included as an alternative simulator in public sampler called 21CMMC. This development provides researchers with a valuable tool for studying and analyzing EoR observations more efficiently.