[Paper Review] Measuring the History of Cosmic Reionization using the 21-cm Difference PDF

Tony Pan, Rennan Barkana|arXiv (Cornell University)|Sep 25, 2012

Radio Astronomy Observations and Technology3 citations

TL;DR

This paper proposes using the 21-cm difference PDF—derived from brightness temperature differences between paired pixels—as a statistical tool to probe the history of cosmic reionization. Using maximum likelihood analysis on mock MWA data, it demonstrates that one-year observations can constrain the late-stage reionization history, while second-generation arrays are needed to measure the full shape of the difference PDF across scales and redshifts.

ABSTRACT

During cosmic reionization, the 21-cm brightness fluctuations were highly non-Gaussian, and complementary statistics can be extracted from the distribution of pixel brightness temperatures that are not derivable from the 21-cm power spectrum. One such statistic is the 21-cm difference PDF, the probability distribution function of the difference in the 21-cm brightness temperatures between two points, as a function of the distance between the points. Guided by 21-cm difference PDFs extracted from simulations, we perform a maximum likelihood analysis on mock observational data, and analyze the ability of present and future low-frequency radio array experiments to estimate the shape of the 21-cm difference PDF, and measure the history of cosmic reionization. We find that one-year data with an experiment such as the Murchison Wide-field Array should suffice for probing large scales during the mid-to-late stages of reionization, while a second-generation experiment should yield detailed measurements over a wide range of scales during most of the reionization era.

Motivation & Objective

• To develop and test a statistical method based on the 21-cm difference PDF to extract reionization history beyond the power spectrum.
• To assess the feasibility of measuring the difference PDF under realistic thermal noise conditions for current and future low-frequency radio arrays.
• To evaluate whether the difference PDF can provide independent, non-Gaussian constraints on ionization topology and bubble sizes during reionization.
• To compare the performance of first-generation (e.g., MWA) and second-generation (e.g., SKA-like) experiments in reconstructing the difference PDF.
• To determine the minimum observational times and signal-to-noise levels required to detect evolving non-Gaussian features in the difference PDF.

Proposed method

• Uses maximum likelihood fitting to reconstruct a binned 21-cm difference PDF from mock observational data, treating each bin as a statistical model parameter.
• Applies equations for thermal noise in 21-cm interferometers (Furlanetto et al. 2006) to estimate noise levels per voxel, scaling with redshift and comoving pixel size.
• Defines the difference PDF as the probability distribution of brightness temperature differences $\Delta T_b = |T_2 - T_1|$ as a function of separation $r$, with bins in both $\Delta T_b$ and $r$.
• Performs simulations to generate mock data with known reionization histories and injects thermal noise to test reconstruction accuracy.
• Considers three noise levels: MWA (1-year), MWA/2 (4-year equivalent), and MWA/10 (second-generation array), to assess sensitivity scaling.
• Analyzes both a 1-bin model (focusing on correlation function) and a 10-bin model (full shape of the difference PDF) to assess reconstruction capability.

Experimental results

Research questions

• RQ1 Can the 21-cm difference PDF be reliably reconstructed from thermal-noise-limited data using maximum likelihood techniques?
• RQ2 How well can one-year MWA data constrain the shape of the difference PDF during mid-to-late reionization?
• RQ3 What observational integration time and noise level are required to measure the detailed shape of the difference PDF across multiple separation bins?
• RQ4 How does the difference PDF evolve during reionization, and what does its shape reveal about ionization topology and bubble sizes?
• RQ5 To what extent can the difference PDF provide independent constraints on the reionization history compared to the power spectrum?

Key findings

• One-year MWA data with standard thermal noise levels can detect the characteristic flattening and bimodal structure of the difference PDF at the end of reionization.
• The 1-bin model (equivalent to the correlation function) can be measured with high accuracy even in one year, due to reduced degeneracy and large number of pairs.
• Four-year MWA data (MWA/2) reduce errors by a factor of about four, not two, due to reduced noise alleviating partial degeneracies in the PDF reconstruction.
• Second-generation experiments (e.g., MWA/10) are required to recover the full shape of the difference PDF across a wide range of separations and redshifts during most of the reionization era.
• The number of pairs at large separations (e.g., $r = 120$ Mpc) is extremely high (~$3 \times 10^{15}$ at $z=8$), enabling high-precision measurement of the asymptotic PDF shape, which reflects the cosmic mean ionized fraction.
• The difference PDF at large separations is nearly constant and probes the one-point PDF with exquisite precision, offering a complementary constraint to the power spectrum.