A-SLOTH reveals the nature of the first stars

The first generation of stars (PopIII) are too dim to be observed directly and probably too short-lived to have survived for local observations. Hence, we rely on simulations and indirect observations to constrain the nature of the first stars. In this study, we calibrate the semi-analytical model A-SLOTH (Ancient Stars and Local Observables by Tracing Halos), designed for simulating star formation in the early Universe, using a likelihood function based on nine independent observables. These observables span Milky Way-specific and cosmologically representative variables, ensuring a comprehensive calibration process. This calibration methodology ensures that A-SLOTH provides a robust representation of the early Universe's star formation processes, aligning simulated values with observed benchmarks across a diverse set of parameters. The outcome of this calibration process is best-fit values and their uncertainties for 11 important parameters that describe star formation in the early Universe, such as the shape of the initial mass function (IMF) of PopIII stars or escape fractions of ionizing photons. Our best-fitting model has a PopIII IMF with a steeper slope, d$N$/d$M \propto M^{-1.77}$, than the log-flat models often proposed in the literature, and also relatively high minimum and maximum masses, $M_{\rm min} = 13.6$Msun and $M_{\rm max} = 197$Msun. However, we emphasize that the IMF-generating parameters are poorly constrained and, e.g., the IMF slope could vary from log-flat to Salpeter. We also provide data products, such as delay time distribution, bubble size distributions for ionizing and metal-enriched bubbles at high redshift, and correlation plots between all 11 input parameters. Our study contributes to understanding the formation of early stars through A-SLOTH and provides valuable insights into the intricate processes involved in the early Universe's star formation.