The detection of Gravitational Waves was arguably the biggest science news of 2016. A few weeks ago I met with Dr Chris Messenger, from University of Glasgow’s Institute for Gravitational Research, to find out what things have been like since, and where they’re headed.
Here we discuss the history of gravitational waves, the joy of teaching the scientific method, the ‘Oscars fever’ of the Nobel Prize, what gravitational waves can tell us about neutron stars, black holes and the big bang, and much more besides.
C&Q: So, how’re things?
CM: Things are great. Just at Christmas there I got promoted to lecturer, which is my dream, so the pressure’s on but in some ways the pressure’s off, and within half an hour of receiving that contract I got engaged. We hope to be moving house soon, so life-wise, it’s great.
C&Q: Congratulations! So as a gravitational researcher as well, it’s been an eventful year?
CM: Oh well, in terms of gravitational research (laughs). For us it started on the day. The code for the detection is 15-09-14, so it’s the 14th September, and we all knew that day that it was something exciting, then it was 3-6 months or so before we were allowed to talk about it. We had to keep it very secret.
C&Q: I heard that initially, the signal was deemed too good to be true. People were going around checking to make sure there were no devices artificially producing the signal, with the possibility of someone trying to make a career for themselves. How likely is that?
CM: Very unlikely. Anyone who knows how to do that knows that you would be leaving a trace. The only… well… one way to do it [smirks and glances are exchanged] would be this… we have an injection system, that fires photons at one of the mirrors of the detectors, to move the detector. Now, this is for calibration because we need to know how far that moves and what voltage that gives out to the detector, so we inject signals for testing via that, but that gets recorded in a channel, so there’s always a trace for that, and you’d have to do it in both of our detectors simultaneously and get it exactly right for the corresponding sky position because of the time delay, and it’s just very, very complicated.
C&Q: A case for Sherlock Holmes.
CM: Well yeah, and as you said, it did look too good to be true, but we would’ve done it in any case whether it was a threshold event or something bigger, so it’s part of the checklist.
C&Q: Why the initial secrecy?
Multiple reasons, but mostly because the field has sore memories of the very first detection claim in the late 60s early 70s by Joe Weber. It was seen as a setback to the field since it initiated a perceived undercurrent of distrust in gravitational wave detection claims.
We also needed time to check every single last possible thing that the event could have been. This included thorough checks for malicious tampering of the instruments, deep searches for natural instrumental causes for the event, plus we needed time to extract all of the information from the data.
Related to the last point is that we had to be as sure as possible that the detector didn’t generate events like these randomly every so often. So we needed as much detector data as possible to analyse the background before claiming our 5-sigma event significance.
C&Q: The detection of Gravitational Waves is about 100 years in the making. Can you give us a little bit of the theory, and where are we now?
CM: So, 1915 is the big date for General Relativity, when Einstein published his paper following on from his work on Special Relativity, and it essentially puts mass into the equation, showing how space, time and mass are all related to each other. One year later, after coming up with this beautiful theory and looking at its properties, one of many is that it contains a wave solution, meaning that certain perturbations in space-time will propagate like waves out into the Universe. They travel at the speed of light, they’re transverse waves, meaning they don’t do anything to anything in the direction they’re travelling in, but the things they pass through, they affect sideways.
C&Q: Like in a guitar string?
CM: Yeah! That’s a transverse wave, whereas sound in air would be a longitudinal wave. They have the effect of stretching and squeezing you or an object transverse, or sideways, from the direction they’re travelling. So that was in 1916, but they realised very early on that the effect was going to be so weak that it seemed unlikely they were ever going to be detected. They had no idea of the technology we have now.
Then in the 1960s/70s, a guy called Joe Weber started looking into how we might detect these things using what’s known as a Resonant Bar Detector. Basically a big block of metal, with the idea that as a wave comes along, vibrating space-time, it will also vibrate this bar, and if you can detect how it’s vibrating, you might be able to pick up these gravitational waves. Depending on how you came in this building [The Kelvin Building], you may have noticed the Glasgow one in the lobby. So Weber announced a detection around that time with one of these, and although it wasn’t a true detection (as the sensitivity of these devices is so low), it sparked great excitement. Other RBDs were created (including the one in the lobby of this building).
Then in the ’80s, a guy called Ray Weiss had the idea to use interferometers to do this. The idea of firing light, two beams, at 90 degrees from each other, bouncing them off mirrors, having the light come back, recombine, and if the lengths of these paths change, then we would see in effect an interference pattern, which gravitational waves would cause since they stretch and squeeze things. This in theory would make for a much more sensitive detector. That was the start of LIGO, which went on from the ’80s to the early 2000s I think, and while it didn’t detect any gravitational waves, it got close, which then sparked Advanced LIGO, which is the upgrade, ten times more sensitive. And on day one, we were so, so lucky to make the detection that we did, of two black holes merging into one.
C&Q: Are we entering into an era of ‘hearing’, as well as seeing, the Universe?
CM: *Nods* Yeah, I use the hearing analogy a lot. We’re not actually hearing gravitational waves as such. In space nobody can hear you scream, right? But our detectors on the ground work in the audio frequency band, between a couple of tens of Hz and a couple of kHz, and our detectors are a bit like antennae rather than telescopes. They’re nearly fully omni-directional. They’ll detect a particular gravitational wave in any particular direction at any given time, much like an ear would detect sound within that band. So, to answer your question, yes! (laughs). It’s so cliché to say we’re opening an entirely new window into the Universe, but it really is just that. So far we’ve only detected one type of thing, but we’re really excited about it, and we have n number of things that we know we’re going to find. The truly exciting things are those we don’t expect to find. They’re the hardest to find, because we’re not looking for them.
C&Q: Gravitational waves travel at the speed of light, right? How do you anticipate a signal? Is it just by chance?
CM: Yeah, you have no idea. We can’t see it before it comes. At the moment it works the other way, in the sense that, when we find triggers or candidates events, we as quickly as possible, ideally within a minute, tell all our astronomers who’ve signed an MOU to keep things secret, where in the sky we saw this thing, and if they have time they all point their telescopes at the right part of the sky. So in a sense we tell other people, but we can’t predict where things are going to come from next. This goes back to skepticism about that initial detection. These things are happening all the time, and we’d be just as lucky to find a detection at any time. So far we’re at 2.9, the third wasn’t quite enough to say definitely. We’re 90% sure, but these days you have to be 99.99999% sure.
There are future designs for space-based detectors, like LISA or The Big Bang Observer. There’s a chance that we’ll see a signal in them, slowly in-spiraling for years, and we’d be able to predict when and where that signal would arrive on Earth at what time and in what bandwidth, so we can say, ‘make sure the telescopes are pointing in the right direction and that LIGO is on’. So there is a predictive nature, but we can’t predict from EM observations.
At a particular source, the binaries are evolving, so the one we detected, they’ll have been going around each other for millions of years, initially orbiting quite slowly, but because they’re emitting gravitational radiation, they lose energy and they fall together and go faster. The frequency of gravitational waves we get is about twice this orbital frequency, and so if it’s evolving to get to a higher frequency it will get to our band at some point. At LIGO, that’s in the last few seconds of its life, whereas in the lower bands it might spend years.
C&Q: What’s happening inside a neutron star?
CM: Theorists model neutron stars as being super-extreme but relatively simple objects, in the sense that they are dictated to by only one function, which is how their pressure varies with density. We want to know what’s called the Equation of State of Super-Nuclear Material, the stuff that’s at this stupidly high density (Neutron Stars are something 1.4 times the mass of the Sun, squeezed into something the size of Glasgow), which is spinning potentially a thousand times a second (the only thing more extreme Astrophysically is a Black Hole).
Gravity wants to crush them in further, but they’re held up by neutron degeneracy pressure, which is a quantum effect [by Heisenberg’s uncertainty principle, particles don’t like being confined to one place, and resist it by whizzing around, creating an outward pressure in the star. A similar effect, electron degeneracy pressure, can be found in white dwarves – C&Q], and it’s this pressure vs density that tells you about the structure of these stars. They’re only made of one thing, which is neutrons (though the outside is a bit crusty with other stuff).
One thing gravitational waves can do, is when we see these things merging in the last few seconds of a binary in-spiral, when they merge and hit each other, then form a big blob of stuff before forming a black hole, you get an imprint of what the thing was made of, including the pressure and density. So that’s what I’ve been working on, trying to use this information for various cosmological and astrophysical reasons.
C&Q: What can Gravitational Waves tell us about the Big Bang?
CM: Yeah, so there are various branches in cosmology. We have the Cosmic Microwave Background, which basically gives us a picture of the Universe 300,000 years after the Big Bang (it seems like a long time but it’s relatively close). We have an equivalent in gravitational waves, and if we could see that map then we’d be able to see much further back in time, on the scale of about 10^-… a very large number, seconds. A very small amount of time. We’re not quite there in terms of the sensitivity we need, but it’s not outside the realm of possibility that we could be with future detectors. At that stage though, it’s just noise, like the CMB, and there’s another noise that would be on top of that and even louder, which is every single thing in the universe emitting gravitational waves (laughs), which forms a background noise, like the overall hum of a party. That’s something we could potentially detect in the next five years with our current detectors.
Slightly less exciting from a cosmological perspective, the thing that I do is more about a thing called Standard Sirens. There’re things called Standard Candles, which are Type IA supernovae, all at the same brightness intrinsically, so if they’re four times less bright then they must be twice as far away by the inverse square law, and then you can start mapping how far things are away vs their redshift, how fast they’re receding. You can then work out things like the Hubble Constant, the energy density of matter, Dark Energy and so on. That’s the side that I look at with the gravitational wave equivalent, which are these binary in-spirals. They allow us to get the distance and hone our maps.
C&Q: The papers you’re a part of are quite varied. Can you tell me a little more about other things you’ve done?
CM: Yeah, I’m a bit of a jack of all trades. I try to do as many things as I can with gravitational waves, so I started in Birmingham, doing my PhD, then I came here the first time round to do the same thing, which is continuous gravitational waves. This is where a neutron star is spinning and has a non-axial symmetry, a big lump on the side. That generates gravitational waves continuously at a constant-ish frequency. They’re even weaker than the other ones, but the bonus is that they’re on all the time, so you can integrate over time to generate a better signal. We’re yet to detect these, but they could be there in many of the neutron stars in the galaxy. People here still do that stuff.
Then I went to Germany, to the Max Planck Institute, working on that, and when I say working on that, we were analysing the data using algorithms we already had or creating new ones, which I spent a lot of time on because that particular problem is highly computationally intractable and you need the computing power of the whole Universe to search for these things properly, so we come up with tricks that sacrifice the sensitivity of our measurements for computational effectiveness.
C&Q: I tried reading some of those papers, and… my brain melted.
CM: (Laughs) Yeah, they’re not the most enlightening papers. It’s very technical. The aim of course is to write papers that involve interesting science and physics, but there’s also data analysis.
C&Q: In a sense it’s a new field, so I suppose you’re able to try for everything.
CM: Yeah, and we’re stealing from other fields as well, (laughs), that’ve done this stuff before. So, then I went to Cardiff and got the chance to work on compact binary coalescences, which is how I got into the Cosmology side of things that I mentioned earlier. Then I came back here on a fellowship, still on compact binaries, keeping in with continuous waves still, but also my colleague [Dr Siong Heng], is the chair of the Burst group, so I’ve started doing a lot of burst things as well. CBC is where the glamour is because that’s where we’re finding stuff, but burst is the next biggest, and I think the most exciting, because there’s so much potential for new physics. So many unknowns. What if we found a population of signals around the galactic centre unlike anything we’d predicted, and they kept arriving? The potential is vast with that.
What else… [looks up at his board]. I can’t seem to get away from continuous waves really, and at the moment we have a very bright PhD student working on some machine learning techniques for analysing the data. Convolutional/ Artificial Neural Networks. Old hat for computer scientists, Google have been using it for years [Amazon as well – this is the same sort of programming that arranges recommendations based on what you browse/ buy – C&Q], like if you type into Google, ‘show me a picture of a cat’, it’ll do that, because it’s learned what cats are. We’re trying to do that with gravitational wave data. Daniel Williams is currently working on machine learning techniques to tell numerical relativity people, who simulate these waveforms for us on supercomputers (which costs millions of CPU hours and dollars), where they should do their next signal in parameter space. We want one of these masses with this spin and so on. Whatever will help us the most.
There’s a lot to this stuff. Have I used the word ‘Bayesian’ yet? I can’t believe I haven’t (laughs). For myself it’s the true way of doing probabilistic data analysis statistics, and it fits in well with my view of how science should be and how it operates. There’s the frequentist approach as well, which suits others and is perfectly valid, but the Bayesian revolution begun here about ten years ago and we’ve gone out into LIGO and spread the disease [where frequentists hold to 0 or 1 probability, ie, true or false, Bayesian statisticians use probability values from 0 to 1 based on an increasingly developed picture, where posteriors (after new data) are calculated, which then become a prior (before new data) in the next stage of testing – C&Q].
C&Q: I don’t suppose you can give us a laymen-friendly summary of Bayesian techniques in under a minute? (Smiles)
CM: (Laughs) Okay, right… Bayesian stuff asks and answers reasonably clear questions, where once you state what the question is you want to ask, it’s relatively straightforward. The main difference from the frequentists is that it allows you to ‘explicitly quote your priors’, so say your mate did an experiment last week. You’re able to fold in your analysis with their result, so that the final product includes all the accumulated knowledge up to that point. What you generate is called a posterior distribution on what you’re looking for, which you can then hand to somebody else and they can call it their prior, and so on in a chain, and in that way, knowledge and information gets refined as you go along.
This happens in the frequentist approach, but it’s hidden. Many frequentists will read this interview and be very angry I’ve said this!
C&Q: I’m not sure how many I really have in my audience, to be honest!
CM: It’s a religious thing, really.
C&Q: Which gets the most/ best results?
CM: Eh, we’re at the crossroads at the moment.
C&Q: The frequentists are the old guard?
CM: Definitely, definitely the old guard. There’s a certain element to doing something over and over again and counting how many times something happened, which is the frequentist approach. It’s also more in line with an experimental way of doing things. I’m an experimentalist, ‘I’m going to try it, I’m going to do this’. Bayesians are more… you don’t have to repeat the experiment, you just have to know what the conditions were and do the correct mathematics.
Right now we have a split in LIGO, which is parameter estimation. What are the masses, what are the spins, where in the sky was that sourced, that’s all Bayesian, everybody’s happy to do that Bayesian. When we state how significant we are at detecting something, we’re all frequentists, because the old guard are very strong in saying we need to use standard statistical techniques that are known throughout the rest of science, that everyone will agree with and understand, which is perfectly reasonable, but there is a move now, now that we have a detection and everyone’s happy, from the younger generation who want to change things. There’re emails flying around about that right now.
C&Q: So in the future we might all be Bayesians, with one frequentist in the corner as a back-up?
CM: I think so, but then these things happen in cycles and I think in 20 years there might be a revolution of frequentists to kick out all the old guard Bayesians probably (Laughs).
C&Q: Speaking of which, it seems science has become increasingly communal since Einstein’s time, where a single guy could be both theorist and experimenter in his basement. What’re the main differences?
CM: Yeah exactly, and you could write a paper, send it off, and not hear anything in three months because no one’s read it (Laughs). As an academic, especially with the career change recently, there’s more admin, there’s a lot more happening, whereas at postdoc you’re just given the research you want to do with minimal admin, but now I’m given the responsibility of teaching, and supervising students with projects, which is one of my passions here, getting the students in their final years involved in gravitational waves or whatever their interest is, getting them involved in the scientific method, how to do things logically, rigorously and be asking questions all the time. It’s surprising how much it doesn’t come naturally to people, and it’s the most important thing, more than knowing how to code, how to do equations, but using those practically and coming up with new ideas. If I want to test this, I’ll have to do that. Those kinds of questions give the most enjoyment while teaching. Plus, I’ve forgotten a lot of what I learned at undergrad by now, and a great way to relearn all that physics is to teach it.
C&Q: You mentioned the importance of the scientific method. What skills are required in science, as both a pursuit and as a career?
CM: One element is that you have to be honest, not just with your collaborators and colleagues, other scientists, but with yourself about your results. There’s a conflict with time constraints, where you might end up with something you’re not totally proud of or necessarily believe in, whether you wish you’d run more tests… and one of the things I try to get into the heads of the students I mentor is, don’t publish anything you don’t trust, but also be aware that you can’t test everything. When you state your results, be honest, quote your results, state what the restrictions were, the limitations, anything you didn’t do. To do that without going crazy and falling into a big research hole you have to be organised and have a plan. Write lots of documents about what you’re doing.
Also, make sure you have a good supervisor, and a good relationship with the people mentoring you. That’s from a student’s perspective. In terms of having a successful career, I’d love to give a list of things other than doing good quality science and loving what you do, but my negative statement is going to be that this game is so lucky. You have to be lucky. On day one that you choose your PhD you might choose something that’s really exciting to you, but it might not be exciting to anyone else, especially funders. You might find yourself to be the greatest scientist in the world, but there are no postdocs for your project. It’s just not exciting right now. We in this field have been so lucky to come along as gravitational waves did, but so many have fallen by the wayside prior to getting a detection. They’re not homeless or anything, they’re smart people who found other careers and things, but following your dreams and being a scientist is in some ways a fool’s game. You just have to keep telling yourself that it’s not you. It really is so lucky. Those who are successful will acknowledge that. Of course there are smarter ones, but you have to be smart and lucky.
In terms of technical skills, trial and error works, it just might not be the quickest way. Again, I’ve been really lucky to have good mentors over the years. My PhD supervisor Professor Alberto Vecchio just threw me in. Just said, ‘we’re going to write this paper, a bit like this one, we’re gonna build up from this one, we’re gonna write some code to do it, use this code that I’ve already written and build on that, there’s a load of code over there, read that, have a look what that is, we’ve got a conference next week, you’re going to that conference and you’re going to give a talk’, and I was just like… *intake of breath*… okay, I’ve got to do it! So when you’re working in big places like LIGO or a big University, there’s always someone you can ask a question. Just ask, don’t be afraid. Finding the right resources to learn is another skill, finding the right papers to learn…
C&Q: The right news sources?
CM:(Laughs) It’s very equivalent to that, yeah.
C&Q: How might the current political climate affect research?
CM: This is tricky since I’m a low level researcher and don’t much get involved in the higher level funding discussions. All I can say with reasonable confidence is that just like most global institutions, uncertainty is bad for us. There was lots of worry during the Scottish independence referendum since an independent Scotland (however you may feel about it) would result in large funding changes in research.
With the global rise of more right leaning politics it becomes less attractive for foreign students and researchers to travel for PhDs, postdocs and conferences. In my opinion however, the most dangerous elements of recent political events are the anti-fact (or post-fact) attitude displayed by global leaders (or one leader in particular). Having these people ignore scientific fact and the scientific method is deeply worrying.
C&Q: What’re your plans going forward?
CM: Our real hope now is that we start detecting things other than black holes. This sounds quite negative, but… the world was amazed by our first discovery. The world was kind of reasonably happy with our second discovery. We had that 0.9, but we have no idea how they’re going to feel about a 3rd or a 4th or a 5th… they all have different properties, but they’re all black holes. We want to try for something new.
C&Q: Tell that to the people who make the Fast & Furious movies.
CM: (Laughs), I think we’ve got a slightly different audience, but… we need to detect neutron stars. That’s the next thing. Very similar waveforms but very different objects. After that, in any order, I don’t care, continuous gravitational waves (Matt Patkin, my colleague next door is doing THE search for that, for LIGO), bursts, detect more things, excite people, and… we need to start using these detections, cos it’s not just about detections, it’s not just saying, ‘these exist, gravitational waves exist, so haha, you’ve been telling us for all these years we’d never find them and now we have’, we need to do something with them. There’s a whole area in our field about taking populations of detections, and computating/ calculating important physical quantities of our universe. For example, making measurements of the Hubble Constant, mass distributions of black holes, working out how stars die, how they get to be black holes in the first place. Gravitational waves will be used to answer a lot of these questions that can’t be answered via other methods.
C&Q: Potentially cheeky question, but there are rumours going around outside IGR that Glasgow might win the Nobel this year for the detection of gravitational waves. Can you confirm or deny?
CM: (Laughs) Um… it’s not cheeky. I mean, no one knows at all. We got really excited about the last one, but we were told that because we kept it secret for so long we were too late for the deadline. I was also told a rumour that we weren’t, and somebody had leaked it to the panel, or that we weren’t too late but we just didn’t win it. I had to give a summer school in Germany in place of Jim Huff on the day of the ceremony, because he realised he had to be in the country in case we won.
We at LIGO were so biased, we reckoned we’d definitely get it. Even if somebody does something better this year, they’d just wait til a year where there was a lull and then we’ll get it, but the big discussion is who gets it. That’s where all the gossip is, choosing three people from a 30 year project. None of this is secret or anything because it’s not official, but the three names that are bandied around are Ray Weiss, founder of LIGO at MIT, I forget the second guy off the top of my head, and then Ron Drever, at Caltech but who started out here and was Jim Huff’s supervisor. There are other prizes and Ron won one of those, so there’s kind of a Golden Globes/ Oscar buzz going around. We’d love it, it’d be great, but we don’t know.
C&Q: Ending on a scientific note. Cheers Chris!
CM: Yeah, thanks, it’s been fun!