Skip to main content Skip to navigation

Joseph Lyman


I am an assistant professor and UKRI Future Leaders Fellowship holder, based in the Astrophysics group at Warwick working in the Explosive Transients and Stellar Populations groups.


My research is concerned primarily with observing and characterising astrophysical transients - new flashes of light in the night sky that signal the deaths of stars in other galaxies. Study of transients such as supernovae, caused by exploding massive stars, can unlock our understanding of how these stars live and die - this understanding is vital since massive stars have had a profound impact on the evolution of galaxies and chemical elements throughout cosmic time.
I have particular interest in constraining the progenitors stars of core-collapse supernovae through observations of their explosions and the environments they inhabit within their hosts galaxies. Alongside this I study unusual or peculiar subclasses of supernovae, the progenitors of which are poorly understood. I am also involved in the follow-up of electromagnetic counterparts to gravitational wave events, including taking some of the earliest observations GW170817/AT 2017gfo - the first multi-messenger gravitational wave event.

To find the next multi-messenger gravitational wave sources, I am leading the software development and science-exploitation of the Warwick-led GOTO project, and exploiting its capabilities to discover new objects in the night sky.

PhD studentship opportunities

2024: The explosion environments and host galaxies of supernovae

I encourage prospective PhD students interested in transient and/or multi-messenger astronomy to take a look at a flavour of my research below and to get in touch using my email on the right for further discussions. Details of projects available will be just above and also held at the postgraduate opportunties page.


I am co-convener of PX458: The Distant Universe. Please see the module Moodle page for all resources and information.

Publication wordcloudjdl_wordcloud

(made with word cloudLink opens in a new window)


Watching a Magnetar move with the Hubble Space Telescope

When a neutron star harbours an extreme magnetic field, something like >1014 Gauss (for reference, the Earth's magnetic field is about 0.5 Gauss), we call it a Magnetar. Such magnetars are prone to erratic and spontaneous high-energy X-ray and Gamma-ray outbursts. Of the 35 or so we know of in the Galaxy, one such Magnetar stands out: SGR 1935+2154. Since its discovery in 2014 following an outburst, it hasn't shut up. SGR 1935 is the most active magnetar in the Galaxy by far, and yet in 2020 it underwent an entirely unexpected kind of outburst - something that looked similar to, but much fainter than, Fast-radio bursts (FRBs) that have been observed in other galaxies. These are short blips of radio emission and the discovery of a FRB-like emission from the known magnetar SGR 1935 opened up a whole avenue of theoretical and observation enquiry to link magnetars and FRBs.

We know that magnetars are neutrons stars, and as such are linked to the deaths of massive stars when they undergo a core-collapse supernova event. This is a highly-energetic and asymmetric explosion, which can impart a significant velocity into the newly-born neutron star (known as a "supernova-kick"). These kicks can move neutron stars up to 100s of kilometres a second relative to their surroundings! In collaboration with colleagues at Radboud University, SGR 1935 has been monitored with the very stable, and accurate Hubble Space Telescope to try and search for a change in its position, and so measure this kick.

The movement on sky is slight. Over the course of 6 years it moved a grand total of about 20 milliarcseconds - one milliarcsecond is the width of a human hair seen from 10 miles away... The movement is so minute we have to account for the Galaxy's rotation over this time-period, and the motion of the Sun within the Galaxy! The Hubble Space Telescope was essential to do these precise measurements, allowing us to find that SGR 1935 was moving with a velocity of just 100+/-50 km/s with respect to its surroundings. When compared with the small handful of other magnetars with measurements, it is the lowest, and among the lowest for all known neutron stars.

Image of the proper motion of SGR 1935+2154 overlaid to show its birth place in SNR G57.2+08
Image of the proper motion of SGR 1935+2154 overlaid on our Hubble image (left) and a zoom-out to show its birth place in SNR G57.2+08 (right). The yellow contours in the right panel indicate our estimate for the position of SGR 1935 when the supernova remnant was born, which matches the geometric centre of the SNR (pink cross).

Now we know how fast it moves, we can take a look at where we think SGR 1935 was born. In the left panel of the figure we have plotted different realisations for the motion (accounting for all our uncertainties on the direction and speed of travel) on top of our Hubble image, tracing it back 10000 years. It doesn't look too interesting of a site - there is no star cluster or similar. However, if we zoom out (right panel), we can see it is located within a supernova remnant known as SNR G57.2+08 - here we have overplotted a radio image of this remnant. We have an age for the supernova remnant, of about 16000 years, and so an age for our magnetar, if they are linked. Indeed, if we trace back the position of the magnetar these 16000 years we find that it almost exactly overlaps with the geometric centre of the remnant itself - confirming the association between SGR 1935 and SNR G57.2+08.

Our results are given in the paper The Fast Radio Burst-emitting magnetar SGR 1935+2154 – proper motion and variability from long-term Hubble Space Telescope monitoring.

Chasing gravitational waves with GOTO

In 2017, the Nobel Prize for Physics was awarded to the team behind the first direct detection of Gravitational Waves, thus settling a century-old prediction by Einstein. These first detections were caused by pairs of black holes spiralling inwards and merging with each other. As an observational astronomer, the real excitement came with the first gravitational wave signature associated with neutron stars, which gives us a chance to see them in electromagnetic radiation as well, creating a "multi-messenger" event (see below). We still (as of April 2020!) only have one example of a gravitational wave multi-messenger event. To overcome the challenges with finding these electromagnetic counterparts, I am developing and exploiting the GOTO project, which is lead by Danny Steeghs at Warwick.

GOTO telescopes at sunset
The two domes of the GOTO project on the island of La Palma. The prototype configuration is shown here, with 4 wide-angle telescopes mounted in a single dome. The final project, which has been funded by STFC as well as partner contributions, will have 16 telescopes, 8 in each dome, alongside a further 16 telescopes in Australia. At full capabilities GOTO will be able to continuously monitor the entire night sky, and react to gravitational wave triggers immediately.

One of the main challenges comes from the fact that unfortunately the gravitational wave detectors can only reliably tell us when an event has occurred, but determining where (and so knowing where to point our telescopes!) is more difficult. To combat this GOTO is designed explicitly as a wide-angle set of telescopes that can quickly swathe across large areas of the night-sky to look for any changes that could be associated with the new event. As yet we (nor anyone else) have not been successful in finding these electromagnetic counterparts - we hope this will change in 2022 when new upgrades to the gravitational wave detectors, as well as our telescopes, will make the challenge surmountable.

In order to make GOTO a world-leading survey my team and I will be using machine-learned image and time-series analysis, alongside novel data pipeline development, to expedite the turn around time from GOTO observations to humans being notified that something interesting has been detected. No easy feat when we are searching for a single object in a sea of hundreds of millions in the night sky.

The results of our search for gravitational-wave counterparts in 2019/2020 are given in:
Observational constraints on the optical and near-infrared emission from the neutron star–black hole binary merger S190814bvLink opens in a new window and Searching for Electromagnetic Counterparts to Gravitational-wave Merger Events with the Prototype Gravitational-wave Optical Transient Observer (GOTO-4)Link opens in a new window

The unusual late-time behaviour of GW170817

GW170817 was the first multi-messenger gravitational wave event. It was caused by the merger of two neutron stars about 130 million light years away and in August 2017 astronomers detected its presence in both gravitational waves and photons (two 'messengers' of information). The first of these photons was a burst of extremely high energy gamma-rays known as a GRB, occuring about 1.7 seconds after the neutron stars merged, as extremely fast-moving material (almost the speed of light) was launched in a jet from the merger. This was followed over the next week or so in optical and infrared light by a kilonova - the site where heavy elements like gold and platinum are made. GW170817 then went behind the Sun and most telescopes were unable to continue monitoring its evolution.

Hubble Space Telescope observations

At 110 days after the merger first occured, as soon as the Hubble Space Telescope was again able to point at its location, we immediately took very deep images to see what was going on. Remarkably we still saw an optical signature from the source (see figure). This signature was far too bright to be due to the kilonova. Instead, we interpreted this as a result of the GRB. When the fast-moving material launched in the merger collides with the surrounding gas and dust in the circumstellar environment, it gives rise to an afterglow. This afterglow is visible across the electromagnetic spectrum, from X-rays, through optical, to radio waves.

Late time HST detection of GW170817
Hubble Space Telescope observations of GW170817 at 110 days after the neutron star merger. The place of the merger within its galaxy is shown in panel a. We see the object in optical light (panels b and c), but not in near-infrared light (d and e). This optical signature is due to the collision between the fast-moving material expelled in the merger with the surrounding material in the circumstellar environment.
Jet or cocoon?

When we looked more closely at how this afterglow was evolving, however, it did not seem to match what we have seen in previous GRBs. Although very faint, it continued to slowly get brighter for months after the merger. The GRB itself we saw 1.7 seconds after GW170817 was itself quite faint. (The fact that is was remarkably nearby (in GRB terms) meant we were still able to detect and characterise it.) With a rising afterglow, and a faint GRB, it was generally accepted that the jet launched by the merger was not pointed at us. There are currently two competing scenarios for marking the afterglow. One scenario suggests the jet began punching through the slower moving material of the merger and was 'choked' causing it to transfer its energy to the material creating a cocoon of material accelerated to mildly relativistic velocities. Another suggests that the jet indeed sucessfully punched through this slower material and instead we are viewing this event in the wings of this jet.

We will continue to monitor GW170817 all through 2018 in order to determine exactly what caused the unusual behaviour we have seen thus far, and to answer the question of whether this is related to the higher-energy GRBs we have seen before.

Update: During 2018, with further Hubble observations, we saw a steep decline in the brightness of the afterglow, confirming our predictions and making the afterglow fully consistent with a successful jet scenario. These results are in Lamb, Lyman et al. (2019), linked below.

Further info

Researchers wait over 100 days to see beam of light from first confirmed neutron star merger emerge from behind the sunLink opens in a new window

Scientists observe first visual evidence of devastating collision between two neutron stars, as powerful beam of light reaches Earth after 130 million yearsLink opens in a new window

The results are contained in the following papers: The optical afterglow of the short gamma-ray burst associated with GW170817Link opens in a new window and The Optical Afterglow of GW170817 at One Year Post-mergerLink opens in a new window

The dawn of multi-messenger gravitational wave astrophysics

On the 17th August 2017, the LIGO-Virgo consortium announced the detection of a binary neutron star merger. Within two seconds high-energy gamma ray emission was seen coming from the same region of the sky. What followed was one of the largest coordinated observing campaigns in astronomy.

Classification spectrum of AT 2017gfo
The first multi-messenger gravitational wave source detection. The object is at 11 o'clock from the centre of its galaxy. The three panels on the right show an image before the new object exploded, when it was young and blue, and when it was older and red (top to bottom). Credit ESO/N. Tanvir/A. Levan/VIN-ROUGE collaboration.

Almost every telescope able to search the southern sky for the eletromagnetic counterpart set about this task. Within a day a candidate was found in the galaxy NGC 4993, around 40 Mpc (130 million lightyears) away. I was lucky enough to be at the European Southern ObservatoryLink opens in a new window at this time, observing explosive transients for ePESSTOLink opens in a new window. At the telescope, with David Homan (U. Edinburgh), we took one of the first spectra of this candidate (named AT 2017gfo) and were the first to classify it as a kilonova and alert the astronomical community to its uniqueness. Because it looked like nothing we'd seen before, we were condfident that the gravitational wave counterpart had been found.

Classification spectrum of AT 2017gfo
The first ePESSTO spectrum of AT 2017gfo showing some features assigned as caesium and tellurium (Smartt et al. 2017).

Our followup campaign withing ePESSTO revealed a somewhat blue, fastly-evolving transient (Smartt et al. 2017). This suggested to us that the merger was not producing so-called 'lanthanide' elements, which include gold and platinum. If the merger had produced these lanthanides we would expect the transient to have been much redder (appearing mainly in the near infrared) and evolve slower, taking perhaps a week to reach its maximum brightness. This is because lanthanides have extremely high opacitiesLink opens in a new window. Instead it appeared to show evidence for relatively lighter (but still heavier than iron) elements such as caesium and tellurium.

Near infrared spectra of AT 2017gfo
Later-time near infrared spectra revealed a lanthanide-rich component (Tanvir et al. 2017).

At late times however the near infrared stayed brighter than expected for a lanthanide-free merger. This was further investigated in Tanvir et al. (2017) where deep near infrared observations were taken from both the ground, with ESO, and space, with the Hubble Space Telescope (HST). Our observations here showed that the merger also had a lanthanide-rich component that emerged after several days. Spectra in the near infrared showed good agreement with models of lanthanide-rich mergers.

Stellar population maps of NGC 4993
Maps of the stellar populations in NGC 4993. Top row (left to right): The line of sight velocity, the velocity spread (dispersion) and extinction of the stars. Bottom row (left to right): The fraction of light contributed by young (younger than 500 million years), intermediate age (between 500 million years and 5 billion years) and old (older than 5 billion years) stars in NGC 4993.

In Levan et al. (2017) we set about investigating where this event had occured. Studying the host galaxy, NGC 4993, and the merger's local environment within that galaxy can provide additional information about the merger. (This is much like investigating what cars were involved in a crash based on where the crash occured - if they crashed turning the corner in Monaco, you would probably think differently to a crash on a local High Street.) We found that the galaxy has recently (in cosmological terms) undergone a merger event - NGC 4993 is the result of two galaxies colliding. This may have happened around a billion years ago as we find a significant population of stars of that age in the galaxy. Because we find no younger stars, we can say that the merger system was at least a billion years old, and perhaps older. From studying the light of the transient compared to the light of the galaxy in the vicinity (specifically looking for absorption features associated with the light having travelled though gas and dust), we find tentative evidence that the merger occured on the near-side of the galaxy. This may indicate that the 'kickLink opens in a new window' that occured during the birth of the neutron stars was directed towards us.

This first object has confirmed that binary neutron stars create gravitational waves, cause short-duration gamma-ray bursts, produce an electromagnetic signature well described by a kilonovaLink opens in a new window and are a dominant source of heavy element production in the universe.

Further info

The results are presented in the following papers (plus many, many others):
A kilonova as the electromagnetic counterpart to a gravitational-wave sourceLink opens in a new window
The emergence of a lanthanide-rich kilonova following the merger of two neutron starsLink opens in a new window
The environment of the binary neutron star merger GW170817Link opens in a new window
Multi-messenger observations of a binary neutron star mergerLink opens in a new window
The Distance to NGC 4993: The Host Galaxy of the Gravitational-wave Event GW170817Link opens in a new window
The Diversity of Kilonova Emission in Short Gamma-Ray BurstsLink opens in a new window

Some example press coverage of the event:
Yorkshire PostLink opens in a new window, BBCLink opens in a new window, European Southern ObservatoryLink opens in a new window, GuardianLink opens in a new window, IndependentLink opens in a new window, TelegraphLink opens in a new window, NASALink opens in a new window.

Investigating the peculiar and diverse supernovae type Iax

Supernovae type Ia (SNe Ia) have been instrumental in our current understanding of cosmology through their use as standard(isable) candles. There are, however, imitators. In the process of searching for SNe Ia, many events have been found that share characteristics of SNe Ia, but have subleties in the explosions that make them distinct. SNe Ia are thought to arise from the thermonuclear detonation of a massive white dwarf, but it is likely these peculiar subtypes are hinting to us the many different ways there are to blow up a star. Of these peculiar types the most numerous are the supernovae type Iax (SNe Iax).

Although SNe Iax look similar to SNe Ia, clues from their spectra (which show lower-velocity, narrower lines) and light curves (which have fainter peaks and evolve faster) suggest they are fainter, weaker cousins to the SNe Ia. So much so that it is difficult to explain their observational properties with models of SNe Ia explosions. Instead, 'failed' explosions may occur, in which the whole of the white dwarf is not detonated and instead a remnant ('zombie star'Link opens in a new window?) is left behind post-explosion. Alternatively, there have been suggestions instead that SNe Iax may be due to more massive stars (around 7-9 times the mass of the sun) that undergo an electron-capture supernova. Even very massive stars (more than 25 times the mass of the sun) have been suggests as progenitors - these would explode and produce a blackhole upon collapse. If a large fraction of the star that is exploding falls into this black hole then we would see a weak energy supernova with little material being thrown off, the traits of a SN Iax.

We set about investigating where these weird supernovae are exploding using spectroscopic data of their galaxy environments from the Very Large Telescope (VLT) and Nordic Optical Telescope (NOT), in order to test predictions of these different progenitor models.


Further info

The results are presented in the following paper:
Investigating the diversity of supernovae type Iax: A MUSE and NOT spectroscopic study of their environmentsLink opens in a new window

Calcium-rich supernovae hosts galaxies and explosion sites

Supernovae are seen to explode in or around the bright discs and bulges of their host galaxies. This is expected since these bright regions constain the vast majority of the stars in a galaxy. Calcium-rich supernovae, a peculiar subclass of supernovae, however, have a strong tendancy to explode in the remote out-reaches of their host galaxies where there are very few stars. Quite why they prefer these solitary locations has been a puzzle. Suggestions have been made that they could be formed in very faint (and thus difficult to detect) systems at these remote locations.

Using observations taken with the Hubble Space Telescope (HST) and the Very Large Telescope (VLT), I have investigated the explosion sites of these unusual explosions to search for signs of any potential birth places, such as faint dwarf galaxies, that may have been undetected by other observations.

SN2005E and it's explosion site


In the first panel, SN2005E, a member of the Ca-rich supernova class, is seen in the bottom right. Returning to observe this location with HST long after the supernova has faded (second panel), we find no evidence for any birth site (last panel - a zoom in of the square on the middle panel).

Ca-rich supernovae are truly lonely

Thanks to the extreme depth of the observations taken with HST and VLT, the fact that we see no sources at the explosion sites of any of the Ca-rich supernovae we have looked at allows us to rule out their formation in faint underlying systems (such as globular clusters or dwarf galaxies).


Runaway couples?

These supernovae really are exploding where they have no business to be. Since there are no obvious birth sites for the supernovae, and the fact that the number of stars in these remote locations is so small, we can consider if these explosions are the result of high-velocity pairs of stars that have been flung from their galaxies at hundreds of kilometres per second.

These systems could comprise of a neutron star and a white dwarf. The neutron star is formed when a very massive star collapses under its own gravity and makes a supernova of its own. When it is formed it undergoes a 'kick', accelerating it to large velocities. Since most stars are in binary systems, a companion star can be dragged along for the ride. The supernovae would then be a result of the companion being eventually ripped apart by the neutron star, after the pair have travelled a significant distance to these remote out-reaches of their host galaxy. To the right is an artist's impression of such a system having being ejected from its galaxy.

Further info

Lonely Supernovae May Have Been Kicked Out Of Their GalaxiesLink opens in a new window

Hubble observes calcium-rich supernovaeLink opens in a new window

The results are contained in two papers: The progenitors of calcium-rich transients are not formed in situLink opens in a new window and Hubble Space Telescope observations of the host galaxies and environments of calcium-rich supernovae.Link opens in a new window

Core-collapse supernova light curves and progenitors

Core-collapse supernovae (CCSNe) are thought to arise from the deaths of stars at least 8 times more massive than the Sun. Beyond this cutoff, there is debate as to exactly how different kinds of massive stars die and specifically how their varied deaths produce the varied types of supernova we see.

I have used literature data of CCSN to produce a method of creating the bolometric light curve of a CCSN (which have traditionally been observationally expensive to create) from relatively little data. This method was then applied to a large sample of CCSN in order to determine the properties of their explosions, and thus inform on their progenitor stars.

Bolometric corrections for CCSNe

Bolometric corrections for CCSNe B-Ic

With just an optical colour, one can estimate the bolometric light curve of a CCSN over a wide range of epochs with a typical rms of the scatter about the relations $< 0.1$ mag. The case for $B-I$ is shown above.

Bolometric lightcurves of CCSNe

Bolometric light curves of CCSNe

Using the bolometric correction method, a large catalogue of bolometric light curves has been created for stripped-envelope CCSNe. These are types of CCSN that show little or no hydrogen in their specta - it is thought that the massive hydrogen envelopes of the progenitor stars has been largely lost before exploding, resulting in an absence of hydrogen.

There are two (main) mechanisms that can strip the hydrogen envelope from a progenitor star. In this first case the progenitor star is simply that massive and luminous that it sheds these outer layers itself, through very strong stellar winds, over the course of its life.This requires very massive stars, typically at least 20 to 30 times as massive as the Sun. In the second case the star can be more modestly massive and it is the presence of a binary companion that strips this envelope through its gravitational influence. This second mechanism can work for lower mass progenitors (8-20 times the mass of the Sun). Thus if we can determine the masses of the proenitor stars we can distinguish between these mechanisms.

Modelling the bolometric light curves of CCSNe

By employing a simple model to the catalogue of bolometric light curves, one can extract estimates for the explosion parameters of the supernovae. One such parameter is the mass of material that is ejected during the supernova, $M_\textrm{ej}$. This is intrinsically linked to the mass of the star when it was born and we can thus use it, by comparing to results from models of stellar evolution, to constrain the mass range of stripped-envelope CCSN progenitors.


Above is a plot of the distributions of $M_\textrm{ej}$ for different types of stripped-envelope CCSNe (IIb, Ib, Ic, Ic-BL). The ranges for $M_\textrm{ej}$ from stellar evolution modelling are indicated by gray bars. As can be seen, the distributions are best decribed by binary stars that are between 8 and 20 times the mass of the Sun when formed ($M_\textrm{init}$). There is very little contribution from more massive stars, either single or binary.

This indicates that the progenitors of stripped-envelope CCSNe are less massive than previously thought, with the binary interaction mechanism being mainly responsible for stripping the hydrogen envelopes of the progenitor stars. A lack of large values for $M_\textrm{ej}$, which would be expected from more massive stars, has dramatic consequences for the fates of these more massive stars. It may indeed be the case that these stars do not produce a luminous supernova, instead directly collapsing to a black hole.

Further info

The results are presented in the following papers:
Bolometric corrections for optical light curves of core-collapse supernovaeLink opens in a new window and Bolometric light curves and explosion parameters of 38 stripped-envelope core-collapse supernovaeLink opens in a new window

recent arxiv feed



A1.13 (Millburn House - see the campus map)

Profile picture