Optimized Multi-Frequency Spectra for Applications in Radiative Feedback and Cosmological Reionization

Abstract

The recent implementation of radiative transfer algorithms in numerous hydrodynamics codes has led to a dramatic improvement in studies of feedback in various astrophysical environments. However, because of methodological limitations and computational expense, the spectra of radiation sources are generally sampled at only a few evenly-spaced discrete emission frequencies. Using one-dimensional radiative transfer calculations, we investigate the discrepancies in gas properties surrounding model stars and accreting black holes that arise solely due to spectral discretization. We find that even in the idealized case of a static and uniform density field, commonly used discretization schemes induce errors in the neutral fraction and temperature by factors of two to three on average, and by over an order of magnitude in certain column density regimes. The consequences are most severe for radiative feedback operating on large scales, dense clumps of gas, and media consisting of multiple chemical species. We have developed a method for optimally constructing discrete spectra, and show that for two test cases of interest, carefully chosen four-bin spectra can eliminate errors associated with frequency resolution to high precision. Applying these findings to a fully three-dimensional radiation-hydrodynamic simulation of the early universe, we find that the H II region around a primordial star is substantially altered in both size and morphology, corroborating the one-dimensional prediction that discrete spectral energy distributions can lead to sizable inaccuracies in the physical properties of a medium, and as a result, the subsequent evolution and observable signatures of objects embedded within it.

Radiation-driven Turbulent Accretion onto Massive Black Holes

“…In the Appendix we include a detailed discussion of the SED optimization used in this work, along with convergence tests, and a comparison with SED models published in Mirocha et al (2012). We explore a full set of modeled SEDs in our simulations and mark those that are modified as outlined in this section in Table 1.…”

“…Optimized SEDs for α = 1.5 Power-law BH radiation using Mirocha et al (2012) Note. -Each entry is given in (hνn, Ln) where hνn is in unit of eV and the sum of Ln is not necessarily equal to one.…”

Extracting the 21cm Global Signal using Artificial Neural Networks

“…In this paper, we have explored Artificial Neural Networks (ANNs), as an alternate technique for parameter extraction for the 21cm Global signal. We produce the cosmological signal using tanh parameterization and the Accelerated Reionization Era Simulations(ARES) (Mirocha et al 2012). We then choose a suitable foreground model to represent the bright, dominant foregrounds and create a training dataset for the ANN.…”

Direct parameter inference from global EoR signal with Bayesian statistics

“…The CMB affects spin temperature mainly through Compton scattering, and gas affects spin temperature through collision and Wouthuysen-Field effect (Wouthuysen 1952;Field 1958). The strength of above three effects are mainly controlled by gas neutral fraction, gas temperature, and background Ly α intensity, which are all tightly connected to the evolving history of potential ionizing sources, i.e., black holes and stars (see Mirocha 2014;Mirocha et al 2012, for a quantitative discussion).…”