Prof. Stephen Serjeant

Research

Infrared extragalactic astronomy

I have a long-standing interest in infrared survey astronomy since our 1998 landmark SCUBA-1 observations of what was then called the Hubble Deep Field, uncovering an unexpected population of hyperluminous starbursts at Cosmic Noon, which was then irreconcilable with semi-analytic models of galaxy evolution. I was Co-PI of Active Galaxies and Rare Objects working group in Herschel ATLAS key project (one of the six science areas). This was the largest Open Time project on the European Space Agency Herschel mission, with 141 members covering 60 institutions from 12 countries. I was part of the AKARI far-infrared space telescope team, including its premier survey of the North Ecliptic Pole. I led the Herschel survey of the North Ecliptic Pole and am UK lead for the ongoing James Clerk Maxwell Telescope survey of the same NEP field. The NEP will shortly be one of the Deep Fields in the forthcoming optical/near-infrared ESA Euclid mission, as well as being a deep eROSITA field and having naturally excellent coverage/visibility for many other space telescopes including JWST. Within Euclid I co-lead the Primeval Universe / Strong Lensing joint working group. I am the deputy UK project scientist for the proposed ESA/JAXA SPICA mission, shortlisted for the ESA M5 slot (see our special issue in PASA).

Strong gravitational lensing

Gravitational lensing is one of the very few means of measuring the dark matter distribution, and magnifies the angular sizes and flux densities of background galaxies. I have been interested in strong gravitational lensing since being one of the 1995 discoverers of the strong lensing of the prototypical hyperluminous galaxy IRAS F10214+4724. Our ESA Herschel ATLAS survey uncovered an unprecedented, extremely efficient route to discovering strong gravitational lens events (Negrello+10, Science), which in turn has led to many high-profile results (e.g. Zavala+18, Nature Astronomy). I have led and been involved in several redshift campaigns for the background lensed sources (using e.g. ALMA, NOEMA, etc.) and the foreground lensing galaxies (using e.g. ESO NTT, SALT, etc.). I am currently leading a dark matter halo substructure project on ALMA and a Cycle 1 strong lensing proposal for JWST (drop me a line if you would like to join the latter). I have also found that both Euclid and the Square Kilometer Array will be rich sources of reliable strong gravitational lenses (Serjeant+14, ApJL) possibly including at ultra-high redshifts (Marchetti+17, MNRAS).

Data mining, machine learning and citizen science

I am interested in many aspects of data processing and data mining, including matched filtering (e.g. my early submm papers such as Serjeant+03, or more recently in e.g. Serjeant19), genetic algorithms (e.g. collaborative work with computer scientist Alex Freitas while I was at Kent), and deep learning including convnets (currently in collaboration with OU computer scientist Jane Bromley, lead author of the pioneering Siamese Networks paper). Within the ESA Euclid Strong Lensing working group my team has participated in the strong lensing challenges using convnets. A surprising result of the first challenge was that human experts visually inspecting the challenge data did not win, so we ran a citizen science experiment to see if this was just fatigue or if other humans could do better. This was funded through our ASTERICS Horizon 2020 project for multimessenger / multiwavelength astronomy and astroparticle physics, in which one goal was to involve the general public directly in scientific discovery. ASTERICS also facilitated the creation of several other crowdsourcing experiments. Citizen science is sometimes regarded as public engagement, but that is not its main reason for existence. Citizen science is not outreach. Rather, it is a crowdsourcing tool for conducting scientific experiments with the help of (usually) the general public, for whom a secondary goal can sometimes be outreach and education. With the help of our subsequent ESCAPE Horizon 2020 project to build the astronomy/physics cell of the European Open Science Cloud, my team have recently launched or help launch two new crowdsourcing experiments: Galaxy Zoo: Clump Scout (led by OU postdoc Hugh Dickinson, with machine-learning backend tools under development for deciding when subjects are well characterised and for volunteer classification reliability) and Radio Galaxy Zoo: LOFAR. Public participation in scientific discovery is part of the ESCAPE vision for the European Open Science Cloud. We are also participating in the REINFORCE Horizon 2020 project to build four new citizen science experiments in astronomy/physics, where we are tasked with sharing best practices between the H2020 projects, avoiding duplications, and chairing the REINFORCE Advisory Board.

Other fun stuff

I have a lot of interdisciplinary interests, many of which follow from my involvement in the STFC Food Network+, the second phase of which now involves my friends at the Ethiopian Space Science and Technology Institute. The objectives are to use STFC technology, data science and facilities in the service of food supply and food security, particularly (in phase 2) in DAC-list countries. This has gotten me involved in lots of fun stuff, such as using astronomical infrared sensing and feature detection to find when cows are on heat (not a metaphor, it turns out) in collaboration with Niamh Forde in Leeds, and using the Zooniverse platform for measuring food perceptions in collaboration with Christian Reynolds. I also co-ran a national workshop in deploying STFC technology and data science to soil science, in collaboration with Chris Collins in Reading. You can read our report on how the STFC Food Network+ can have an impact on climate change.

Last update: 6 March 2020