A consistent and robust measurement of the thermal state of the IGM at 2 ≤ z ≤ 4 from a large sample of  Ly α forest spectra: evidence for late and rapid He ii reionization

Abstract

We characterise the thermal state of the intergalactic medium (IGM) in ten redshift bins in the range 2 ≤ z ≤ 4 with a sample of 103 high resolution, high S/N Lyα forest spectra using four different flux distribution statistics. Our measurements are calibrated with mock spectra from a large suite of hydrodynamical simulations post-processed with our thermal IGM evolution code cite, finely sampling amplitude and slope of the expected temperature-density relation. The thermal parameters inferred from our measurements of the flux power spectrum, Doppler parameter distribution, as well as wavelet and curvature statistics agree well within their respective errors and all clearly show the peak in temperature and minimum in slope of the temperature density relation expected from He ii reionization. Combining our measurements from the different flux statistics gives T0 = (14750 ± 1322)K for the peak temperature at mean density and a corresponding minimum slope γ = 1.225 ± 0.120. The peak in the temperature evolution occurs around z ≈ 3, in agreement with previous measurements that had suggested the presence of such a peak, albeit with a large scatter. Using cite, we also calculate the thermal state of the IGM predicted by five widely used (spatially homogeneous) UV-background models. The rather rapid thermal evolution inferred by our measurements is well reproduced by two of the models, if we assume (physically well motivated) non-equilibrium evolution with photo-heating rates that are reduced by a moderate factor of ∼0.7 − 0.8. The other three models predict He ii reionization to be more extended with a higher temperature peak occurring somewhat earlier than our measurements suggest.

Improving IGM temperature constraints using wavelet analysis on high-redshift quasars

“…The precision from this method can then be directly compared to the canonical approach using 𝑃 F (𝑘). We will also consider the precision achieved when combining the wavelet amplitude PDFs and power spectrum as has recently been attempted in the literature (Gaikwad et al 2021).…”

“…Several statistics have been used to measure the thermal state of the IGM, including the flux probability density (Becker et al 2007;Bolton et al 2008;Viel et al 2009;Calura et al 2012;Lee et al 2015), the curvature (Becker et al 2011;Boera et al 2014;Gaikwad et al 2021), the Doppler parameter distribution (Schaye et al 1999(Schaye et al , 2000Ricotti et al 2000;Bryan & Machacek 2000;McDonald et al 2001;Rudie et al 2012;Bolton et al 2010Bolton et al , 2012Bolton et al , 2014Rorai et al 2018;Gaikwad et al 2021), and the joint distribution of the Doppler parameters with the Hydrogen Column Density (Hiss et al 2018). One of the most commonly used statistics used to measure the structure of the Ly𝛼 forest is the 1D flux power spectrum (𝑃 F (𝑘)) Zaldarriaga et al 2001;Yèche et al 2017;Walther et al 2017;Boera et al 2019;Gaikwad et al 2021). The reduction in small-scale structure in the Ly𝛼 forest leads to a cut-off in the power at high 𝑘 values.…”

“…Wavelet analysis provides an alternative statistical method to measure the structure of the Ly𝛼 forest over a range of characteristic scales (Lidz et al 2010;Garzilli et al 2012;Gaikwad et al 2021) (though see also Theuns & Zaroubi (2000); Theuns et al (2002b); Zaldarriaga (2002); Meiksin (2000)). Wavelets are localized in both frequency and real space, which allows them to encode Fourier information while remaining in configuration space.…”

“…Our work builds off and improves upon the previous implementation of wavelet analysis done by Lidz et al (2010) and Gaikwad et al (2021). The work done in Lidz et al (2010) used one of the two characteristic wavelet scales explored to constrain the thermal state of the IGM.…”

“…Lidz et al (2010) also ignored correlations between the bins in the wavelet amplitude PDFs, potentially significantly affecting the resulting error bars. Gaikwad et al (2021) used eight wavelet scales in their analysis and included the correlations between the bins within each wavelet amplitude PDF. Their method still ignored the correlations between the bins for wavelet amplitude PDFs of different scales, again potentially effecting the resulting error bars.…”