Artificial intelligence applied to the automatic analysis of absorption spectra. Objective measurement of the fine structure constant [IMA]

A new and fully-automated method is presented for the analysis of high-resolution absorption spectra (GVPFIT). The method has broad application but here we apply it specifically to the problem of measuring the fine structure constant at high redshift. For this we need objectivity and reproducibility. GVPFIT is also motivated by the importance of obtaining a large statistical sample of measurements of $\Delta\alpha/\alpha$. Interactive analyses are both time consuming and complex and automation makes obtaining a large sample feasible.
Three numerical methods are unified into one artificial intelligence process: a genetic algorithm that emulates the Darwinian processes of reproduction, mutation and selection, non-linear least-squares with parameter constraints (VPFIT), and Bayesian model averaging. In contrast to previous methodologies, which relied on a particular solution as being the most likely model, GVPFIT plus Bayesian model averaging derives results from a large set of models, and helps overcome systematic uncertainties associated with model choice.
In this paper we illustrate the method using a test-case, the $z_{abs} = 1.8389$ absorber towards the $z_{em} = 2.145$ quasar J110325-264515 (chosen because there are several previous analyses of the same system for comparison). When applied to this system, we show that GVPFIT performs better than a human. The derived constraint of $\Delta\alpha/\alpha = 3.5 \pm 2.5 \times 10^{-6}$ is consistent with no variation and also consistent with the tentative spatial variation reported in Webb et al (2011) and King et al (2012).

Read this paper on arXiv…

M. Bainbridge and J. Webb
Fri, 24 Jun 16

Comments: 33 pages, 10 figures, 16 tables. Submitted to MNRAS