A new analysis of fine-structure constant measurements and modelling errors from quasar absorption lines

Abstract

We present an analysis of 23 absorption systems along the lines of sight towards 18 quasars in the redshift range of 0.4 z abs 2.3 observed on the Very Large Telescope (VLT) using the Ultraviolet and Visual Echelle Spectrograph (UVES). Considering both statistical and systematic error contributions we find a robust estimate of the weighted mean deviation of the fine-structure constant from its current, laboratory value of ∆α/α = (0.22 ± 0.23) × 10 −5 , consistent with the dipole variation reported in Webb et al. (2011) and King et al. (2012). This paper also examines modelling methodologies and systematic effects. In particular we focus on the consequences of fitting quasar absorption systems with too few absorbing components and of selectively fitting only the stronger components in an absorption complex. We show that using insufficient continuum regions around an absorption complex causes a significant increase in the scatter of a sample of ∆α/α measurements, thus unnecessarily reducing the overall precision. We further show that fitting absorption systems with too few velocity components also results in a significant increase in the scatter of ∆α/α measurements, and in addition causes ∆α/α error estimates to be systematically underestimated. These results thus identify some of the potential pitfalls in analysis techniques and provide a guide for future analyses.

Precise limits on cosmological variability of the fine-structure constant with zinc and chromium quasar absorption lines

“…Using only the statistical errors in Table 2, the weighted mean ∆α/α is ∆α/α w = 1.12±1.67 ppm, which is clearly consistent with no cosmological variation in α. Note that this precision is similar to that of the large absorber samples from Keck and VLT (≈1.2 ppm; Webb et al 2001;Murphy et al 2004;Webb et al 2011;King et al 2012) and somewhat better than that of the 23-absorber VLT sample of Chand et al (2004) reanalysed by Wilczynska et al (2015), ≈2.3 ppm. However, the uncertainties quoted for these previous studies were not just statistical and included some systematic effects.…”

“…Interestingly, spectra from the Keck telescope's High Resolution Echelle Spectrometer (HIRES) indicated some evidence for a smaller α at redshifts z abs = 0.2-4.2 than the current laboratory value Murphy et al 2001a), with the final sample of 143 absorbers indicating ∆α/α = −5.7 ± 1.1 parts per million (ppm; Murphy et al 2003Murphy et al , 2004. Spectra from the Very Large Telescope's (VLT's) Ultraviolet and Visual Echelle Spectrograph (UVES) did not replicate this result (Chand et al 2004, though see Murphy et al 2008;Wilczynska et al 2015), with the most recent large sample of 153 absorbers indicating ∆α/α = +2.1 ± 1.2 ppm (King et al 2012). Nevertheless, when combined, the large HIRES and UVES samples supported self-consistent, 4σ evidence for a coherent, dipole-like variation in α across the sky (Webb et al 2011;King et al 2012).…”

Cosmological constraints onΛ(α)CDM models with time-varying fine structure constant

“…In addition, we include 313 data points of ∆α/α from the absorption systems in the spectra of distant quasars with 0.2223 z abs 4.1798 in the analysis. Note that among these data, 293 were published in 2012 [23] 2 , while 20 of them are the new ones [36]. It is interesting to emphasize that ∆α/α = O(10 −5 ) for all data points.…”

High-precision limit on variation in the fine-structure constant from a single quasar absorption system

“…It therefore appears that previous analyses significantly underestimated the number of components, even taking into account the lower S/N of the spectra used in those studies (see also discussion of Q04's fit in Murphy et al 2008b). Murphy et al (2008b,a) and Wilczynska et al (2015) have shown that under-fitting in this way often leads to spurious shifts in varying-constant measurements from absorption profiles. On the other hand, Murphy et al (2008b) showed that overfitting, to the very small extent allowed by χ 2 minimization codes like vpfit, yields accurate ∆α/α values with slightly overestimated statistical errors.…”