Pushing 21cm to the (statistical) limit: A first EoR detection with an SKA Precursor

Lead Research Organisation: University of Manchester
Department Name: Physics and Astronomy

Abstract

We propose to undertake the following research projects.

1) We will develop and extend non-linear numerical codes which model the evolution of the universe from inflation to recombination. Using these codes we will calculate precision observable signatures of the early universe including the
bispectrum, magnetic fields and gravitational waves. We will use these observable signatures to confront models of inflation and reheating with forthcoming observational survey data.

2) We explore signatures of GR in galaxy clustering observables; develop optimal statistical methods to extract them from standard contributions; provide simulation data and analysis software to measure and exploit these unique windows on gravity with forthcoming cosmological surveys.

3) We develop the mathematical and numerical tools required to understand gravitational physics in both the mildly nonlinear and strongly non-linear regimes. This work will apply and extend the perturbation theories and parameterizations of dark energy and modified gravity that we have already created for these regimes, and allow us to create the tools that are necessary to exploit them using upcoming data.

4) We extend our state of the art simulations and develop complete data analysis pipelines for the cross-correlation of HI intensity mapping surveys with spectroscopic and photometric optical galaxy surveys. This will enable pioneering HI and cosmological measurements, combining currently available and forthcoming data; provide an open source analysis toolkit for HI intensity mapping and optical galaxy data-sets.

5) We develop and apply statistical analysis methods to solve several important challenges in the analysis of 21cm datasets, e.g. due to foreground leakage and covariance-related signal loss. We will develop fast but highly-accurate simulations to characterise the methods, and then apply them to data from HERA, a key SKA precursor. We will disseminate our tools and expertise to prime the UK community for SKA1.

6) We use plasma wake-field acceleration as a unifying concept to study novel particle acceleration concepts and the origin of ultra high energy cosmic rays . We will perform ground-breaking analytical calculations and cutting-edge plasma fluid and kinetic numerical simulations to address the fascinating, unanswered science questions related to particle acceleration in CERN and black holes.

7) GAIA DR2 can measure pairwise velocity differences for wide binary stars to excellent precision ~ 0.05 km/s. We have selected a sample of > 5000 good candidate wide binaries with low contamination. We propose to follow-up a subset of these with spectroscopy and public imaging to produce a `cleaned' sample, and use this to derive useful new constraints on modified-gravity models.

8) We use HPC simulations to model the gas and dust discs that orbit young stars, which are believed to be the sites of planet formation. We study how the planets that form in these discs interact with them, leading to orbital migration, gap formation and accretion of material. Extrasolar planets are observed to have a broad range of masses and orbital architectures, and we examine the role of disc-planet interactions in determining these properties.

9) We develop novel, canonical hydrodynamic numerical schemes, suitable for inspiral simulations and gravitational-wave detection of neutron binary systems. We obtain formulations of a general class of modified theories of gravity with higher curvature corrections which can be used for inspiral simulations of black hole binaries.

10) We develop a theoretical and computational infrastructure to model gravitational wave signals from extreme mass-ratio inspirals, combining novel numerical methods with time-domain gravitational self-force computation in a radiation gauge to allow calculations
 
Description We have developed a set of statistical tools that allow us to separate out a very weak radio signal from neutral hydrogen gas around the first stars and galaxies from much brighter radio signals that come from our own Galaxy and others nearby. Because there is such a big difference in the strengths of these signals, it is important to very carefully consider all the different uncertainties that could affect the measurement, as otherwise the signals can be mistakenly muddled together. The most significant result of our research is a computer code (and mathematical results to support it) that can perform this separation in a statistically rigorous way, greatly reducing the chances of getting the signals mixed up. As a byproduct, it is also able to solve a number of other issues that are common with this kind of data, such as gaps in the data where radio interference from Earth has been cut out, and accurately estimating the strength and structure of various sources of uncertainty in the data. In addition, we have developed a number of simulation tools to support this work. These make realistic-looking synthetic data that we can use as a laboratory to understand various imperfections in the data, and validate that our code is doing the right thing in the presence of these imperfections. In particular, we have been able to develop simulation codes that are faster and more accurate than their predecessors, which has been very useful for our work.
Exploitation Route The key outputs of this work are open-source computer codes for analysing and simulating data from 21cm radio arrays that can be used by other groups working on similar experiments around the world. Some of these tools are already being used widely within the HERA collaboration, and have been extended to apply to other types of experiment, e.g. 21cm global signal experiments. A next step is to demonstrate that these tools can be used with data from the Square Kilometre Array. To that end, we are currently participating in the SKA Science Data Challenge 3, which will allow our code to compete against other methods in analysing simulated SKA low-frequency data. Of particular use are the 21cm power spectrum estimation and foreground separation code (hydra-pspec), and a very fast and accurate radio interferometer visibility simulator (vis_cpu). Both of these can be used by other experiments with only minimal adaptation.

Beyond this, our methods could be adapted by other applications that require very high dynamic range separation of a stochastic (random) signal from a vastly brighter but much more structured contaminating process. Possible applications could be envisaged in extracting weak telecommunication signals, e.g. over very large distances or challenging environments (e.g. space, under the sea).

(Please note that this project was moved out of a previous consolidated grant, ST/T000341/1, and so some outputs and impacts will be listed as part of that grant.)
Sectors Digital/Communication/Information Technologies (including Software)

URL https://github.com/HydraRadio/hydra-pspec
 
Title Radio interferometer visibility simulator 
Description An important outcome of this research has been a greatly enhanced visibility simulator code that is targeted at the HERA project, but which should find general applicability to other radio interferometers. The main features of the code are that it is very fast to run, which is important for interferometers with many baselines, but highly accurate. In fact, it has been benchmarked against a high-accuracy visibility simulator and produces similar performance, despite being significantly faster to run. 
Type Of Material Improvements to research infrastructure 
Year Produced 2022 
Provided To Others? Yes  
Impact The main impact has been to enable more extensive statistical studies of the performance and accuracy of the HERA data analysis pipeline, as we are now able to generate much larger simulations in a timely fashion. 
URL https://github.com/HERA-Team/vis_cpu
 
Title Hydra-pspec 
Description As part of this project, we have developed a novel Bayesian statistical code to measure the power spectrum of 21cm data, also taking into account foreground contamination and gaps in the data due to RFI filtering. This has significant advantages compared with more traditional methods, as outlined in this paper (Kennedy et al., ApJS accepted: https://arxiv.org/abs/2211.05088) 
Type Of Material Data analysis technique 
Year Produced 2023 
Provided To Others? Yes  
Impact It is still early days, but we expect to be able to perform an enhanced (more accurate/robust) analysis of recent HERA datasets using this method. 
URL https://github.com/HydraRadio/hydra-pspec
 
Description Interviewed for podcast 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Public/other audiences
Results and Impact I was interviewed by the "Better Man" podcast about my research and related fields. The interview ran to around 1h30m and is freely and publicly available. The podcast has a substantial international audience via the Apple Podcasts service. The podcast as a whole has had 20,000 views across all platforms, and so I anticipate this particular episode to have reached a few hundred listeners.
Year(s) Of Engagement Activity 2022
URL https://podcasts.apple.com/gb/podcast/the-better-man-podcast-with-phil-bull-52/id1566852474?i=100057...
 
Description School project interview 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach Local
Primary Audience Schools
Results and Impact One high school student interviewed me on the connections between my research work and people's religious belief, as part of a school project. This only happened recently, but the student seemed very engaged and will go on to write a report. I was also able to answer the student's questions about studying STEM subjects at university.
Year(s) Of Engagement Activity 2023