A little while back I wrote a short post about some research that some colleagues and I did using “open data” from the Large Hadron Collider [LHC]. We used data made public by the CMS experimental collaboration — about 1% of their current data — to search for a new particle, using a couple of twists (as proposed over 10 years ago) on a standard technique. (CMS is one of the two general-purpose particle detectors at the LHC; the other is called ATLAS.) We had two motivations: (1) Even if we didn’t find a new particle, we wanted to prove that our search method was effective; and (2) we wanted to stress-test the CMS Open Data framework, to assure it really does provide all the information needed for a search for something unknown.
Recently I discussed (1), and today I want to address (2): to convey why open data from the LHC is useful but controversial, and why we felt it was important, as theoretical physicists (i.e. people who perform particle physics calculations, but do not build and run the actual experiments), to do something with it that is usually the purview of experimenters.
The Importance of Archiving Data
In many subfields of physics and astronomy, data from experiments is made public as a matter of routine. Usually this occurs after an substantial delay, to allow the experimenters who collected the data to analyze it first for major discoveries. That’s as it should be: the experimenters spent years of their lives proposing, building and testing the experiment, and they deserve an uninterrupted opportunity to investigate its data. To force them to release data immediately would create a terrible disincentive for anyone to do all the hard work!
Data from particle physics colliders, however, has not historically been made public. More worrying, it has rarely been archived in a form that is easy for others to use at a later date. I’m not the right person to tell you the history of this situation, but I can give you a sense for why this still happens today. (more…)
A few days ago I wrote a quick summary of a project that we just completed (and you may find it helpful to read that post first). In this project, we looked for new particles at the Large Hadron Collider (LHC) in a novel way, in two senses. Today I’m going to explain what we did, why we did it, and what was unconventional about our search strategy.
The first half of this post will be appropriate for any reader who has been following particle physics as a spectator sport, or in some similar vein. In the second half, I’ll add some comments for my expert colleagues that may be useful in understanding and appreciating some of our results. [If you just want to read the comments for experts, jump here.]
Why did we do this?
Motivation first. Why, as theorists, would we attempt to take on the role of our experimental colleagues — to try on our own to analyze the extremely complex and challenging data from the LHC? We’re by no means experts in data analysis, and we were very slow at it. And on top of that, we only had access to 1% of the data that CMS has collected. Isn’t it obvious that there is no chance whatsoever of finding something new with just 1% of the data, since the experimenters have had years to look through much larger data sets? (more…)
Today, a small but intrepid band of theoretical particle physicists (professor Jesse Thaler of MIT, postdocs Yotam Soreq and Wei Xue of CERN, Harvard Ph.D. student Cari Cesarotti, and myself) put out a paper that is unconventional in two senses. First, we looked for new particles at the Large Hadron Collider in a way that hasn’t been done before, at least in public. And second, we looked for new particles at the Large Hadron Collider in a way that hasn’t been done before, at least in public.
And no, there’s no error in the previous paragraph.
1) We used a small amount of actual data from the CMS experiment, even though we’re not ourselves members of the CMS experiment, to do a search for a new particle. Both ATLAS and CMS, the two large multipurpose experimental detectors at the Large Hadron Collider [LHC], have made a small fraction of their proton-proton collision data public, through a website called the CERN Open Data Portal. Some experts, including my co-authors Thaler, Xue and their colleagues, have used this data (and the simulations that accompany it) to do a variety of important studies involving known particles and their properties. [Here’s a blog post by Thaler concerning Open Data and its importance from his perspective.] But our new study is the first to look for signs of a new particle in this public data. While our chances of finding anything were low, we had a larger goal: to see whether Open Data could be used for such searches. We hope our paper provides some evidence that Open Data offers a reasonable path for preserving priceless LHC data, allowing it to be used as an archive by physicists of the post-LHC era.
2) Since only had a tiny fraction of CMS’s data was available to us, about 1% by some count, how could we have done anything useful compared to what the LHC experts have already done? Well, that’s why we examined the data in a slightly unconventional way (one of several methods that I’ve advocated for many years, but has not been used in any public study). Consequently it allowed us to explore some ground that no one had yet swept clean, and even have a tiny chance of an actual discovery! But the larger scientific goal, absent a discovery, was to prove the value of this unconventional strategy, in hopes that the experts at CMS and ATLAS will use it (and others like it) in future. Their chance of discovering something new, using their full data set, is vastly greater than ours ever was.
Now don’t all go rushing off to download and analyze terabytes of CMS Open Data; you’d better know what you’re getting into first. It’s worthwhile, but it’s not easy going. LHC data is extremely complicated, and until this project I’ve always been skeptical that it could be released in a form that anyone outside the experimental collaborations could use. Downloading the data and turning it into a manageable form is itself a major task. Then, while studying it, there are an enormous number of mistakes that you can make (and we made quite a few of them) and you’d better know how to make lots of cross-checks to find your mistakes (which, fortunately, we did know; we hope we found all of them!) The CMS personnel in charge of the Open Data project were enormously helpful to us, and we’re very grateful to them; but since the project is new, there were inevitable wrinkles which had to be worked around. And you’d better have some friends among the experimentalists who can give you advice when you get stuck, or point out aspects of your results that don’t look quite right. [Our thanks to them!]
All in all, this project took us two years! Well, honestly, it should have taken half that time — but it couldn’t have taken much less than that, with all we had to learn. So trying to use Open Data from an LHC experiment is not something you do in your idle free time.
Nevertheless, I feel it was worth it. At a personal level, I learned a great deal more about how experimental analyses are carried out at CMS, and by extension, at the LHC more generally. And more importantly, we were able to show what we’d hoped to show: that there are still tremendous opportunities for discovery at the LHC, through the use of (even slightly) unconventional model-independent analyses. It’s a big world to explore, and we took only a small step in the easiest direction, but perhaps our efforts will encourage others to take bigger and more challenging ones.
For those readers with greater interest in our work, I’ll put out more details in two blog posts over the next few days: one about what we looked for and how, and one about our views regarding the value of open data from the LHC, not only for our project but for the field of particle physics as a whole.
There has long been a question as to what types of events and processes are responsible for the highest-energy neutrinos coming from space and observed by scientists. Another question, probably related, is what creates the majority of high-energy cosmic rays — the particles, mostly protons, that are constantly raining down upon the Earth.
As scientists’ ability to detect high-energy neutrinos (particles that are hugely abundant, electrically neutral, very light-weight, and very difficult to observe) and high-energy photons (particles of light, though not necessarily of visible light) have become more powerful and precise, there’s been considerable hope of getting an answer to these question. One of the things we’ve been awaiting (and been disappointed a couple of times) is a violent explosion out in the universe that produces both high-energy photons and neutrinos at the same time, at a high enough rate that both types of particles can be observed at the same time coming from the same direction.
In recent years, there has been some indirect evidence that blazars — narrow jets of particles, pointed in our general direction like the barrel of a gun, and created as material swirls near and almost into giant black holes in the centers of very distant galaxies — may be responsible for the high-energy neutrinos. Strong direct evidence in favor of this hypothesis has just been presented today. Last year, one of these blazars flared brightly, and the flare created both high-energy neutrinos and high-energy photons that were observed within the same period, coming from the same place in the sky.
I have written about the IceCube neutrino observatory before; it’s a cubic kilometer of ice under the South Pole, instrumented with light detectors, and it’s ideal for observing neutrinos whose motion-energy far exceeds that of the protons in the Large Hadron Collider, where the Higgs particle was discovered. These neutrinos mostly pass through Ice Cube undetected, but one in 100,000 hits something, and debris from the collision produces visible light that Ice Cube’s detectors can record. IceCube has already made important discoveries, detecting a new class of high-energy neutrinos.
On Sept 22 of last year, one of these very high-energy neutrinos was observed at IceCube. More precisely, a muon created underground by the collision of this neutrino with an atomic nucleus was observed in IceCube. To create the observed muon, the neutrino must have had a motion-energy tens of thousand times larger than than the motion-energy of each proton at the Large Hadron Collider (LHC). And the direction of the neutrino’s motion is known too; it’s essentially the same as that of the observed muon. So IceCube’s scientists knew where, on the sky, this neutrino had come from.
(This doesn’t work for typical cosmic rays; protons, for instance, travel in curved paths because they are deflected by cosmic magnetic fields, so even if you measure their travel direction at their arrival to Earth, you don’t then know where they came from. Neutrinos, beng electrically neutral, aren’t affected by magnetic fields and travel in a straight line, just as photons do.)
Very close to that direction is a well-known blazar (TXS-0506), four billion light years away (a good fraction of the distance across the visible universe).
The IceCube scientists immediately reported their neutrino observation to scientists with high-energy photon detectors. (I’ve also written about some of the detectors used to study the very high-energy photons that we find in the sky: in particular, the Fermi/LAT satellite played a role in this latest discovery.) Fermi/LAT, which continuously monitors the sky, was already detecting high-energy photons coming from the same direction. Within a few days the Fermi scientists had confirmed that TXS-0506 was indeed flaring at the time — already starting in April 2017 in fact, six times as bright as normal. With this news from IceCube and Fermi/LAT, many other telescopes (including the MAGIC cosmic ray detector telescopes among others) then followed suit and studied the blazar, learning more about the properties of its flare.
Now, just a single neutrino on its own isn’t entirely convincing; is it possible that this was all just a coincidence? So the IceCube folks went back to their older data to snoop around. There they discovered, in their 2014-2015 data, a dramatic flare in neutrinos — more than a dozen neutrinos, seen over 150 days, had come from the same direction in the sky where TXS-0506 is sitting. (More precisely, nearly 20 from this direction were seen, in a time period where normally there’d just be 6 or 7 by random chance.) This confirms that this blazar is indeed a source of neutrinos. And from the energies of the neutrinos in this flare, yet more can be learned about this blazar, and how it makes high-energy photons and neutrinos at the same time. Interestingly, so far at least, there’s no strong evidence for this 2014 flare in photons, except perhaps an increase in the number of the highest-energy photons… but not in the total brightness of the source.
The full picture, still emerging, tends to support the idea that the blazar arises from a supermassive black hole, acting as a natural particle accelerator, making a narrow spray of particles, including protons, at extremely high energy. These protons, millions of times more energetic than those at the Large Hadron Collider, then collide with more ordinary particles that are just wandering around, such as visible-light photons from starlight or infrared photons from the ambient heat of the universe. The collisions produce particles called pions, made from quarks and anti-quarks and gluons (just as protons are), which in turn decay either to photons or to (among other things) neutrinos. And its those resulting photons and neutrinos which have now been jointly observed.
Since cosmic rays, the mysterious high energy particles from outer space that are constantly raining down on our planet, are mostly protons, this is evidence that many, perhaps most, of the highest energy cosmic rays are created in the natural particle accelerators associated with blazars. Many scientists have suspected that the most extreme cosmic rays are associated with the most active black holes at the centers of galaxies, and now we have evidence and more details in favor of this idea. It now appears likely that that this question will be answerable over time, as more blazar flares are observed and studied.
The announcement of this important discovery was made at the National Science Foundation by Francis Halzen, the IceCube principal investigator, Olga Botner, former IceCube spokesperson, Regina Caputo, the Fermi-LAT analysis coordinator, and Razmik Mirzoyan, MAGIC spokesperson.
This week, the community of high-energy physicists — of those of us fascinated by particles, fields, strings, black holes, and the universe at large — is mourning the loss of one of the great theoretical physicists of our time, Joe Polchinski. It pains me deeply to write these words.
Everyone who knew him personally will miss his special qualities — his boyish grin, his slightly wicked sense of humor, his charming way of stopping mid-sentence to think deeply, his athleticism and friendly competitiveness. Everyone who knew his research will feel the absence of his particular form of genius, his exceptional insight, his unique combination of abilities, which I’ll try to sketch for you below. Those of us who were lucky enough to know him both personally and scientifically — well, we lose twice.
Polchinski — Joe, to all his colleagues — had one of those brains that works magic, and works magically. Scientific minds are as individual as personalities. Each physicist has a unique combination of talents and skills (and weaknesses); in modern lingo, each of us has a superpower or two. Rarely do you find two scientists who have the same ones.
Joe had several superpowers, and they were really strong. He had a tremendous knack for looking at old problems and seeing them in a new light, often overturning conventional wisdom or restating that wisdom in a new, clearer way. And he had prodigious technical ability, which allowed him to follow difficult calculations all the way to the end, on paths that would have deterred most of us.
One of the greatest privileges of my life was to work with Joe, not once but four times. I think I can best tell you a little about him, and about some of his greatest achievements, through the lens of that unforgettable experience.
[To my colleagues: this post was obviously written in trying circumstances, and it is certainly possible that my memory of distant events is foggy and in error. I welcome any corrections that you might wish to suggest.]
Our papers between 1999 and 2006 were a sequence of sorts, aimed at understanding more fully the profound connection between quantum field theory — the language of particle physics — and string theory — best-known today as a candidate for a quantum theory of gravity. In each of those papers, as in many thousands of others written after 1995, Joe’s most influential contribution to physics played a central role. This was the discovery of objects known as “D-branes”, which he found in the context of string theory. (The term is a generalization of the word `membrane’.)
I can already hear the polemical haters of string theory screaming at me. ‘A discovery in string theory,’ some will shout, pounding the table, ‘an untested and untestable theory that’s not even wrong, should not be called a discovery in physics.’ Pay them no mind; they’re not even close, as you’ll see by the end of my remarks.
The Great D-scovery
In 1989, Joe, working with two young scientists, Jin Dai and Rob Leigh, was exploring some details of string theory, and carrying out a little mathematical exercise. Normally, in string theory, strings are little lines or loops that are free to move around anywhere they like, much like particles moving around in this room. But in some cases, particles aren’t in fact free to move around; you could, for instance, study particles that are trapped on the surface of a liquid, or trapped in a very thin whisker of metal. With strings, there can be a new type of trapping that particles can’t have — you could perhaps trap one end, or both ends, of the string within a surface, while allowing the middle of the string to move freely. The place wherea string’s end may be trapped — whether a point, a line, a surface, or something more exotic in higher dimensions — is what we now call a “D-brane”. [The `D’ arises for uninteresting technical reasons.]
Joe and his co-workers hit the jackpot, but they didn’t realize it yet. What they discovered, in retrospect, was that D-branes are an automatic feature of string theory. They’re not optional; you can’t choose to study string theories that don’t have them. And they aren’t just surfaces or lines that sit still. They’re physical objects that can roam the world. They have mass and create gravitational effects. They can move around and scatter off each other. They’re just as real, and just as important, as the strings themselves!
Fig. 1: D branes (in green) are physical objects on which a fundamental string (in red) can terminate.
It was as though Joe and his collaborators started off trying to understand why the chicken crossed the road, and ended up discovering the existence of bicycles, cars, trucks, buses, and jet aircraft. It was that unexpected, and that rich.
And yet, nobody, not even Joe and his colleagues, quite realized what they’d done. Rob Leigh, Joe’s co-author, had the office next to mine for a couple of years, and we wrote five papers together between 1993 and 1995. Yet I think Rob mentioned his work on D-branes to me just once or twice, in passing, and never explained it to me in detail. Their paper had less than twenty citations as 1995 began.
In 1995 the understanding of string theory took a huge leap forward. That was the moment when it was realized that all five known types of string theory are different sides of the same die — that there’s really only one string theory. A flood of papers appeared in which certain black holes, and generalizations of black holes — black strings, black surfaces, and the like — played a central role. The relations among these were fascinating, but often confusing.
And then, on October 5, 1995, a paper appeared that changed the whole discussion, forever. It was Joe, explaining D-branes to those of us who’d barely heard of his earlier work, and showing that many of these black holes, black strings and black surfaces were actually D-branes in disguise. His paper made everything clearer, simpler, and easier to calculate; it was an immediate hit. By the beginning of 1996 it had 50 citations; twelve months later, the citation count was approaching 300.
So what? Great for string theorists, but without any connection to experiment and the real world. What good is it to the rest of us? Patience. I’m just getting to that.
What’s it Got to Do With Nature?
Our current understanding of the make-up and workings of the universe is in terms of particles. Material objects are made from atoms, themselves made from electrons orbiting a nucleus; and the nucleus is made from neutrons and protons. We learned in the 1970s that protons and neutrons are themselves made from particles called quarks and antiquarks and gluons — specifically, from a “sea” of gluons and a few quark/anti-quark pairs, within which sit three additional quarks with no anti-quark partner… often called the `valence quarks’. We call protons and neutrons, and all other particles with three valence quarks, `baryons”. (Note that there are no particles with just one valence quark, or two, or four — all you get is baryons, with three.)
In the 1950s and 1960s, physicists discovered short-lived particles much like protons and neutrons, with a similar sea, but which contain one valence quark and one valence anti-quark. Particles of this type are referred to as “mesons”. I’ve sketched a typical meson and a typical baryon in Figure 2. (The simplest meson is called a “pion”; it’s the most common particle produced in the proton-proton collisions at the Large Hadron Collider.)
Fig. 2: Baryons (such as protons and neutrons) and mesons each contain a sea of gluons and quark-antiquark pairs; baryons have three unpaired “valence” quarks, while mesons have a valence quark and a valence anti-quark. (What determines whether a quark is valence or sea involves subtle quantum effects, not discussed here.)
But the quark/gluon picture of mesons and baryons, back in the late 1960s, was just an idea, and it was in competition with a proposal that mesons are little strings. These are not, I hasten to add, the “theory of everything” strings that you learn about in Brian Greene’s books, which are a billion billion times smaller than a proton. In a “theory of everything” string theory, often all the types of particles of nature, including electrons, photons and Higgs bosons, are tiny tiny strings. What I’m talking about is a “theory of mesons” string theory, a much less ambitious idea, in which only the mesons are strings. They’re much larger: just about as long as a proton is wide. That’s small by human standards, but immense compared to theory-of-everything strings.
Why did people think mesons were strings? Because there was experimental evidence for it! (Here’s another example.) And that evidence didn’t go away after quarks were discovered. Instead, theoretical physicists gradually understood why quarks and gluons might produce mesons that behave a bit like strings. If you spin a meson fast enough (and this can happen by accident in experiments), its valence quark and anti-quark may separate, and the sea of objects between them forms what is called a “flux tube.” See Figure 3. [In certain superconductors, somewhat similar flux tubes can trap magnetic fields.] It’s kind of a thick string rather than a thin one, but still, it shares enough properties with a string in string theory that it can produce experimental results that are similar to string theory’s predictions.
Fig. 3: One reason mesons behave like strings in experiment is that a spinning meson acts like a thick string, with the valence quark and anti-quark at the two ends.
And so, from the mid-1970s onward, people were confident that quantum field theories like the one that describes quarks and gluons can create objects with stringy behavior. A number of physicists — including some of the most famous and respected ones — made a bolder, more ambitious claim: that quantum field theory and string theory are profoundly related, in some fundamental way. But they weren’t able to be precise about it; they had strong evidence, but it wasn’t ever entirely clear or convincing.
In particular, there was an important unresolved puzzle. If mesons are strings, then what are baryons? What are protons and neutrons, with their three valence quarks? What do they look like if you spin them quickly? The sketches people drew looked something like Figure 3. A baryon would perhaps become three joined flux tubes (with one possibly much longer than the other two), each with its own valence quark at the end. In a stringy cartoon, that baryon would be three strings, each with a free end, with the strings attached to some sort of junction. This junction of three strings was called a “baryon vertex.” If mesons are little strings, the fundamental objects in a string theory, what is the baryon vertex from the string theory point of view?! Where is it hiding — what is it made of — in the mathematics of string theory?
Fig. 4: A fast-spinning baryon looks vaguely like the letter Y — three valence quarks connected by flux tubes to a “baryon vertex”. A cartoon of how this would appear from a stringy viewpoint, analogous to Fig. 3, leads to a mystery: what, in string theory, is this vertex?!
[Experts: Notice that the vertex has nothing to do with the quarks. It’s a property of the sea — specifically, of the gluons. Thus, in a world with only gluons — a world whose strings naively form loops without ends — it must still be possible, with sufficient energy, to create a vertex-antivertex pair. Thus field theory predicts that these vertices must exist in closed string theories, though they are linearly confined.]
The baryon puzzle: what is a baryon from the string theory viewpoint?
No one knew. But isn’t it interesting that the most prominent feature of this vertex is that it is a location where a string’s end can be trapped?
Everything changed in the period 1997-2000. Following insights from many other physicists, and using D-branes as the essential tool, Juan Maldacena finally made the connection between quantum field theory and string theory precise. He was able to relate strings with gravity and extra dimensions, which you can read about in Brian Greene’s books, with the physics of particles in just three spatial dimensions, similar to those of the real world, with only non-gravitational forces. It was soon clear that the most ambitious and radical thinking of the ’70s was correct — that almost every quantum field theory, with its particles and forces, can alternatively be viewed as a string theory. It’s a bit analogous to the way that a painting can be described in English or in Japanese — fields/particles and strings/gravity are, in this context, two very different languages for talking about exactly the same thing.
The saga of the baryon vertex took a turn in May 1998, when Ed Witten showed how a similar vertex appears in Maldacena’s examples. [Note added: I had forgotten that two days after Witten’s paper, David Gross and Hirosi Ooguri submitted a beautiful, wide-ranging paper, whose section on baryons contains many of the same ideas.] Not surprisingly, this vertex was a D-brane — specifically a D-particle, an object on which the strings extending from freely-moving quarks could end. It wasn’t yet quite satisfactory, because the gluons and quarks in Maldacena’s examples roam free and don’t form mesons or baryons. Correspondingly the baryon vertex isn’t really a physical object; if you make one, it quickly diffuses away into nothing. Nevertheless, Witten’s paper made it obvious what was going on. To the extent real-world mesons can be viewed as strings, real-world protons and neutrons can be viewed as strings attached to a D-brane.
The baryon puzzle, resolved. A baryon is made from three strings and a point-like D-brane. [Note there is yet another viewpoint in which a baryon is something known as a skyrmion, a soliton made from meson fields — but that is an issue for another day.]It didn’t take long for more realistic examples, with actual baryons, to be found by theorists. I don’t remember who found one first, but I do know that one of the earliest examples showed up in my first paper with Joe, in the year 2000.
Working with Joe
That project arose during my September 1999 visit to the KITP (Kavli Institute for Theoretical Physics) in Santa Barbara, where Joe was a faculty member. Some time before that I happened to have studied a field theory (called N=1*) that differed from Maldacena’s examples only slightly, but in which meson-like objects do form. One of the first talks I heard when I arrived at KITP was by Rob Myers, about a weird property of D-branes that he’d discovered. During that talk I made a connection between Myers’ observation and a feature of the N=1* field theory, and I had one of those “aha” moments that physicists live for. I suddenly knew what the string theory that describes the N=1* field theory must look like.
But for me, the answer was bad news. To work out the details was clearly going to require a very difficult set of calculations, using aspects of string theory about which I knew almost nothing [non-holomorphic curved branes in high-dimensional curved geometry.] The best I could hope to do, if I worked alone, would be to write a conceptual paper with lots of pictures, and far more conjectures than demonstrable facts.
But I was at KITP. Joe and I had had a good personal rapport for some years, and I knew that we found similar questions exciting. And Joe was the brane-master; he knew everything about D-branes. So I decided my best hope was to persuade Joe to join me. I engaged in a bit of persistent cajoling. Very fortunately for me, it paid off.
I went back to the east coast, and Joe and I went to work. Every week or two Joe would email some research notes with some preliminary calculations in string theory. They had such a high level of technical sophistication, and so few pedagogical details, that I felt like a child; I could barely understand anything he was doing. We made slow progress. Joe did an important warm-up calculation, but I found it really hard to follow. If the warm-up string theory calculation was so complex, had we any hope of solving the full problem? Even Joe was a little concerned.
And then one day, I received a message that resounded with a triumphant cackle — a sort of “we got ’em!” that anyone who knew Joe will recognize. Through a spectacular trick, he’d figured out how use his warm-up example to make the full problem easy! Instead of months of work ahead of us, we were essentially done.
From then on, it was great fun! Almost every week had the same pattern. I’d be thinking about a quantum field theory phenomenon that I knew about, one that should be visible from the string viewpoint — such as the baryon vertex. I knew enough about D-branes to develop a heuristic argument about how it should show up. I’d call Joe and tell him about it, and maybe send him a sketch. A few days later, a set of notes would arrive by email, containing a complete calculation verifying the phenomenon. Each calculation was unique, a little gem, involving a distinctive investigation of exotically-shaped D-branes sitting in a curved space. It was breathtaking to witness the speed with which Joe worked, the breadth and depth of his mathematical talent, and his unmatched understanding of these branes.
[Experts: It’s not instantly obvious that the N=1* theory has physical baryons, but it does; you have to choose the right vacuum, where the theory is partially Higgsed and partially confining. Then to infer, from Witten’s work, what the baryon vertex is, you have to understand brane crossings (which I knew about from Hanany-Witten days): Witten’s D5-brane baryon vertex operator creates a physical baryon vertex in the form of a D3-brane 3-ball, whose boundary is an NS 5-brane 2-sphere located at a point in the usual three dimensions. And finally, a physical baryon is a vertex with n strings that are connected to nearby D5-brane 2-spheres. See chapter VI, sections B, C, and E, of our paper from 2000.]
Throughout our years of collaboration, it was always that way when we needed to go head-first into the equations; Joe inevitably left me in the dust, shaking my head in disbelief. That’s partly my weakness… I’m pretty average (for a physicist) when it comes to calculation. But a lot of it was Joe being so incredibly good at it.
Fortunately for me, the collaboration was still enjoyable, because I was almost always able to keep pace with Joe on the conceptual issues, sometimes running ahead of him. Among my favorite memories as a scientist are moments when I taught Joe something he didn’t know; he’d be silent for a few seconds, nodding rapidly, with an intent look — his eyes narrow and his mouth slightly open — as he absorbed the point. “Uh-huh… uh-huh…”, he’d say.
But another side of Joe came out in our second paper. As we stood chatting in the KITP hallway, before we’d even decided exactly which question we were going to work on, Joe suddenly guessed the answer! And I couldn’t get him to explain which problem he’d solved, much less the solution, for several days!! It was quite disorienting.
This was another classic feature of Joe. Often he knew he’d found the answer to a puzzle (and he was almost always right), but he couldn’t say anything comprehensible about it until he’d had a few days to think and to turn his ideas into equations. During our collaboration, this happened several times. (I never said “Use your words, Joe…”, but perhaps I should have.) Somehow his mind was working in places that language doesn’t go, in ways that none of us outside his brain will ever understand. In him, there was something of an oracle.
Looking Toward The Horizon
Our interests gradually diverged after 2006; I focused on the Large Hadron Collider [also known as the Large D-brane Collider], while Joe, after some other explorations, ended up thinking about black hole horizons and the information paradox. But I enjoyed his work from afar, especially when, in 2012, Joe and three colleagues (Ahmed Almheiri, Don Marolf, and James Sully) blew apart the idea of black hole complementarity, widely hoped to be the solution to the paradox. [I explained this subject here, and also mentioned a talk Joe gave about it here.] The wreckage is still smoldering, and the paradox remains.
Then Joe fell ill, and we began to lose him, at far too young an age. One of his last gifts to us was his memoirs, which taught each of us something about him that we didn’t know. Finally, on Friday last, he crossed the horizon of no return. If there’s no firewall there, he knows it now.
What, we may already wonder, will Joe’s scientific legacy be, decades from now? It’s difficult to foresee how a theorist’s work will be viewed a century hence; science changes in unexpected ways, and what seems unimportant now may become central in future… as was the path for D-branes themselves in the course of the 1990s. For those of us working today, D-branes in string theory are clearly Joe’s most important discovery — though his contributions to our understanding of black holes, cosmic strings, and aspects of field theory aren’t soon, if ever, to be forgotten. But who knows? By the year 2100, string theory may be the accepted theory of quantum gravity, or it may just be a little-known tool for the study of quantum fields.
Yet even if the latter were to be string theory’s fate, I still suspect it will be D-branes that Joe is remembered for. Because — as I’ve tried to make clear — they’re real. Really real. There’s one in every proton, one in every neutron. Our bodies contain them by the billion billion billions. For that insight, that elemental contribution to human knowledge, our descendants can blame Joseph Polchinski.
Thanks for everything, Joe. We’ll miss you terribly. You so often taught us new ways to look at the world — and even at ourselves.
Yesterday’s post on the results from the LIGO/VIRGO network of gravitational wave detectors was aimed at getting information out, rather than providing the pedagogical backdrop. Today I’m following up with a post that attempts to answer some of the questions that my readers and my personal friends asked me. Some wanted to understand better how to visualize what had happened, while others wanted more clarity on why the discovery was so important. So I’ve put together a post which (1) explains what neutron stars and black holes are and what their mergers are like, (2) clarifies why yesterday’s announcement was important — and there were many reasons, which is why it’s hard to reduce it all to a single soundbite. And (3) there are some miscellaneous questions at the end.
First, a disclaimer: I am *not* an expert in the very complex subject of neutron star mergers and the resulting explosions, called kilonovas. These are much more complicated than black hole mergers. I am still learning some of the details. Hopefully I’ve avoided errors, but you’ll notice a few places where I don’t know the answers … yet. Perhaps my more expert colleagues will help me fill in the gaps over time.
Please, if you spot any errors, don’t hesitate to comment!! And feel free to ask additional questions whose answers I can add to the list.
BASIC QUESTIONS ABOUT NEUTRON STARS, BLACK HOLES, AND MERGERS
What are neutron stars and black holes, and how are they related?
Every atom is made from a tiny atomic nucleus, made of neutrons and protons (which are very similar), and loosely surrounded by electrons. Most of an atom is empty space, so it can, under extreme circumstances, be crushed — but only if every electron and proton convert to a neutron (which remains behind) and a neutrino (which heads off into outer space.) When a giant star runs out of fuel, the pressure from its furnace turns off, and it collapses inward under its own weight, creating just those extraordinary conditions in which the matter can be crushed. Thus: a star’s interior, with a mass one to several times the Sun’s mass, is all turned into a several-mile(kilometer)-wide ball of neutrons — the number of neutrons approaching a 1 with 57 zeroes after it.
If the star is big but not too big, the neutron ball stiffens and holds its shape, and the star explodes outward, blowing itself to pieces in a what is called a core-collapse supernova. The ball of neutrons remains behind; this is what we call a neutron star. It’s a ball of the densest material that we know can exist in the universe — a pure atomic nucleus many miles(kilometers) across. It has a very hard surface; if you tried to go inside a neutron star, your experience would be a lot worse than running into a closed door at a hundred miles per hour.
If the star is very big indeed, the neutron ball that forms may immediately (or soon) collapse under its own weight, forming a black hole. A supernova may or may not result in this case; the star might just disappear. A black hole is very, very different from a neutron star. Black holes are what’s left when matter collapses irretrievably upon itself under the pull of gravity, shrinking down endlessly. While a neutron star has a surface that you could smash your head on, a black hole has no surface — it has an edge that is simply a point of no return, called a horizon. In Einstein’s theory, you can just go right through, as if passing through an open door. You won’t even notice the moment you go in. [Note: this is true in Einstein’s theory. But there is a big controversy as to whether the combination of Einstein’s theory with quantum physics changes the horizon into something novel and dangerous to those who enter; this is known as the firewall controversy, and would take us too far afield into speculation.] But once you pass through that door, you can never return.
Black holes can form in other ways too, but not those that we’re observing with the LIGO/VIRGO detectors.
Why are their mergers the best sources for gravitational waves?
One of the easiest and most obvious ways to make gravitational waves is to have two objects orbiting each other. If you put your two fists in a pool of water and move them around each other, you’ll get a pattern of water waves spiraling outward; this is in rough (very rough!) analogy to what happens with two orbiting objects, although, since the objects are moving in space, the waves aren’t in a material like water. They are waves in space itself.
To get powerful gravitational waves, you want objects each with a very big mass that are orbiting around each other at very high speed. To get the fast motion, you need the force of gravity between the two objects to be strong; and to get gravity to be as strong as possible, you need the two objects to be as close as possible (since, as Isaac Newton already knew, gravity between two objects grows stronger when the distance between them shrinks.) But if the objects are large, they can’t get too close; they will bump into each other and merge long before their orbit can become fast enough. So to get a really fast orbit, you need two relatively small objects, each with a relatively big mass — what scientists refer to as compact objects. Neutron stars and black holes are the most compact objects we know about. Fortunately, they do indeed often travel in orbiting pairs, and do sometimes, for a very brief period before they merge, orbit rapidly enough to produce gravitational waves that LIGO and VIRGO can observe.
Why do we find these objects in pairs in the first place?
Stars very often travel in pairs… they are called binary stars. They can start their lives in pairs, forming together in large gas clouds, or even if they begin solitary, they can end up pairing up if they live in large densely packed communities of stars where it is common for multiple stars to pass nearby. Perhaps surprisingly, their pairing can survive the collapse and explosion of either star, leaving two black holes, two neutron stars, or one of each in orbit around one another.
What happens when these objects merge?
Not surprisingly, there are three classes of mergers which can be detected: two black holes merging, two neutron stars merging, and a neutron star merging with a black hole. The first class was observed in 2015 (and announced in 2016), the second was announced yesterday, and it’s a matter of time before the third class is observed. The two objects may orbit each other for billions of years, very slowly radiating gravitational waves (an effect observed in the 70’s, leading to a Nobel Prize) and gradually coming closer and closer together. Only in the last day of their lives do their orbits really start to speed up. And just before these objects merge, they begin to orbit each other once per second, then ten times per second, then a hundred times per second. Visualize that if you can: objects a few dozen miles (kilometers) across, a few miles (kilometers) apart, each with the mass of the Sun or greater, orbiting each other 100 times each second. It’s truly mind-boggling — a spinning dumbbell beyond the imagination of even the greatest minds of the 19th century. I don’t know any scientist who isn’t awed by this vision. It all sounds like science fiction. But it’s not.
How do we know this isn’t science fiction?
We know, if we believe Einstein’s theory of gravity (and I’ll give you a very good reason to believe in it in just a moment.) Einstein’s theory predicts that such a rapidly spinning, large-mass dumbbell formed by two orbiting compact objects will produce a telltale pattern of ripples in space itself — gravitational waves. That pattern is both complicated and precisely predicted. In the case of black holes, the predictions go right up to and past the moment of merger, to the ringing of the larger black hole that forms in the merger. In the case of neutron stars, the instants just before, during and after the merger are more complex and we can’t yet be confident we understand them, but during tens of seconds before the merger Einstein’s theory is very precise about what to expect. The theory further predicts how those ripples will cross the vast distances from where they were created to the location of the Earth, and how they will appear in the LIGO/VIRGO network of three gravitational wave detectors. The prediction of what to expect at LIGO/VIRGO thus involves not just one prediction but many: the theory is used to predict the existence and properties of black holes and of neutron stars, the detailed features of their mergers, the precise patterns of the resulting gravitational waves, and how those gravitational waves cross space. That LIGO/VIRGO have detected the telltale patterns of these gravitational waves. That these wave patterns agree with Einstein’s theory in every detail is the strongest evidence ever obtained that there is nothing wrong with Einstein’s theory when used in these combined contexts. That then in turn gives us confidence that our interpretation of the LIGO/VIRGO results is correct, confirming that black holes and neutron stars really exist and really merge. (Notice the reasoning is slightly circular… but that’s how scientific knowledge proceeds, as a set of detailed consistency checks that gradually and eventually become so tightly interconnected as to be almost impossible to unwind. Scientific reasoning is not deductive; it is inductive. We do it not because it is logically ironclad but because it works so incredibly well — as witnessed by the computer, and its screen, that I’m using to write this, and the wired and wireless internet and computer disk that will be used to transmit and store it.)
THE SIGNIFICANCE(S) OF YESTERDAY’S ANNOUNCEMENT OF A NEUTRON STAR MERGER
What makes it difficult to explain the significance of yesterday’s announcement is that it consists of many important results piled up together, rather than a simple takeaway that can be reduced to a single soundbite. (That was also true of the black hole mergers announcement back in 2016, which is why I wrote a long post about it.)
So here is a list of important things we learned. No one of them, by itself, is earth-shattering, but each one is profound, and taken together they form a major event in scientific history.
First confirmed observation of a merger of two neutron stars: We’ve known these mergers must occur, but there’s nothing like being sure. And since these things are too far away and too small to see in a telescope, the only way to be sure these mergers occur, and to learn more details about them, is with gravitational waves. We expect to see many more of these mergers in coming years as gravitational wave astronomy increases in its sensitivity, and we will learn more and more about them.
New information about the properties of neutron stars: Neutron stars were proposed almost a hundred years ago and were confirmed to exist in the 60’s and 70’s. But their precise details aren’t known; we believe they are like a giant atomic nucleus, but they’re so vastly larger than ordinary atomic nuclei that can’t be sure we understand all of their internal properties, and there are debates in the scientific community that can’t be easily answered… until, perhaps, now.
From the detailed pattern of the gravitational waves of this one neutron star merger, scientists already learn two things. First, we confirm that Einstein’s theory correctly predicts the basic pattern of gravitational waves from orbiting neutron stars, as it does for orbiting and merging black holes. Unlike black holes, however, there are more questions about what happens to neutron stars when they merge. The question of what happened to this pair after they merged is still out — did the form a neutron star, an unstable neutron star that, slowing its spin, eventually collapsed into a black hole, or a black hole straightaway?
But something important was already learned about the internal properties of neutron stars. The stresses of being whipped around at such incredible speeds would tear you and I apart, and would even tear the Earth apart. We know neutron stars are much tougher than ordinary rock, but how much more? If they were too flimsy, they’d have broken apart at some point during LIGO/VIRGO’s observations, and the simple pattern of gravitational waves that was expected would have suddenly become much more complicated. That didn’t happen until perhaps just before the merger. So scientists can use the simplicity of the pattern of gravitational waves to infer some new things about how stiff and strong neutron stars are. More mergers will improve our understanding. Again, there is no other simple way to obtain this information.
First visual observation of an event that produces both immense gravitational waves and bright electromagnetic waves: Black hole mergers aren’t expected to create a brilliant light display, because, as I mentioned above, they’re more like open doors to an invisible playground than they are like rocks, so they merge rather quietly, without a big bright and hot smash-up. But neutron stars are big balls of stuff, and so the smash-up can indeed create lots of heat and light of all sorts, just as you might naively expect. By “light” I mean not just visible light but all forms of electromagnetic waves, at all wavelengths (and therefore at all frequencies.) Scientists divide up the range of electromagnetic waves into categories. These categories are radio waves, microwaves, infrared light, visible light, ultraviolet light, X-rays, and gamma rays, listed from lowest frequency and largest wavelength to highest frequency and smallest wavelength. (Note that these categories and the dividing lines between them are completely arbitrary, but the divisions are useful for various scientific purposes. The only fundamental difference between yellow light, a radio wave, and a gamma ray is the wavelength and frequency; otherwise they’re exactly the same type of thing, a wave in the electric and magnetic fields.)
So if and when two neutron stars merge, we expect both gravitational waves and electromagnetic waves, the latter of many different frequencies created by many different effects that can arise when two huge balls of neutrons collide. But just because we expect them doesn’t mean they’re easy to see. These mergers are pretty rare — perhaps one every hundred thousand years in each big galaxy like our own — so the ones we find using LIGO/VIRGO will generally be very far away. If the light show is too dim, none of our telescopes will be able to see it.
But this light show was plenty bright. Gamma ray detectors out in space detected it instantly, confirming that the gravitational waves from the two neutron stars led to a collision and merger that produced very high frequency light. Already, that’s a first. It’s as though one had seen lightning for years but never heard thunder; or as though one had observed the waves from hurricanes for years but never observed one in the sky. Seeing both allows us a whole new set of perspectives; one plus one is often much more than two.
Over time — hours and days — effects were seen in visible light, ultraviolet light, infrared light, X-rays and radio waves. Some were seen earlier than others, which itself is a story, but each one contributes to our understanding of what these mergers are actually like.
Confirmation of the best guess concerning the origin of “short” gamma ray bursts: For many years, bursts of gamma rays have been observed in the sky. Among them, there seems to be a class of bursts that are shorter than most, typically lasting just a couple of seconds. They come from all across the sky, indicating that they come from distant intergalactic space, presumably from distant galaxies. Among other explanations, the most popular hypothesis concerning these short gamma-ray bursts has been that they come from merging neutron stars. The only way to confirm this hypothesis is with the observation of the gravitational waves from such a merger. That test has now been passed; it appears that the hypothesis is correct. That in turn means that we have, for the first time, both a good explanation of these short gamma ray bursts and, because we know how often we observe these bursts, a good estimate as to how often neutron stars merge in the universe.
First distance measurement to a source using both a gravitational wave measure and a redshift in electromagnetic waves, allowing a new calibration of the distance scale of the universe and of its expansion rate: The pattern over time of the gravitational waves from a merger of two black holes or neutron stars is complex enough to reveal many things about the merging objects, including a rough estimate of their masses and the orientation of the spinning pair relative to the Earth. The overall strength of the waves, combined with the knowledge of the masses, reveals how far the pair is from the Earth. That by itself is nice, but the real win comes when the discovery of the object using visible light, or in fact any light with frequency below gamma-rays, can be made. In this case, the galaxy that contains the neutron stars can be determined.
Once we know the host galaxy, we can do something really important. We can, by looking at the starlight, determine how rapidly the galaxy is moving away from us. For distant galaxies, the speed at which the galaxy recedes should be related to its distance because the universe is expanding.
How rapidly the universe is expanding has been recently measured with remarkable precision, but the problem is that there are two different methods for making the measurement, and they disagree. This disagreement is one of the most important problems for our understanding of the universe. Maybe one of the measurement methods is flawed, or maybe — and this would be much more interesting — the universe simply doesn’t behave the way we think it does.
What gravitational waves do is give us a third method: the gravitational waves directly provide the distance to the galaxy, and the electromagnetic waves directly provide the speed of recession. There is no other way to make this type of joint measurement directly for distant galaxies. The method is not accurate enough to be useful in just one merger, but once dozens of mergers have been observed, the average result will provide important new information about the universe’s expansion. When combined with the other methods, it may help resolve this all-important puzzle.
Best test so far of Einstein’s prediction that the speed of light and the speed of gravitational waves are identical: Since gamma rays from the merger and the peak of the gravitational waves arrived within two seconds of one another after traveling 130 million years — that is, about 5 thousand million million seconds — we can say that the speed of light and the speed of gravitational waves are both equal to the cosmic speed limit to within one part in 2 thousand million million. Such a precise test requires the combination of gravitational wave and gamma ray observations.
Efficient production of heavy elements confirmed: It’s long been said that we are star-stuff, or stardust, and it’s been clear for a long time that it’s true. But there’s been a puzzle when one looks into the details. While it’s known that all the chemical elements from hydrogen up to iron are formed inside of stars, and can be blasted into space in supernova explosions to drift around and eventually form planets, moons, and humans, it hasn’t been quite as clear how the other elements with heavier atoms — atoms such as iodine, cesium, gold, lead, bismuth, uranium and so on — predominantly formed. Yes they can be formed in supernovas, but not so easily; and there seem to be more atoms of heavy elements around the universe than supernovas can explain. There are many supernovas in the history of the universe, but the efficiency for producing heavy chemical elements is just too low.
It was proposed some time ago that the mergers of neutron stars might be a suitable place to produce these heavy elements. Even those these mergers are rare, they might be much more efficient, because the nuclei of heavy elements contain lots of neutrons and, not surprisingly, a collision of two neutron stars would produce lots of neutrons in its debris, suitable perhaps for making these nuclei. A key indication that this is going on would be the following: if a neutron star merger could be identified using gravitational waves, and if its location could be determined using telescopes, then one would observe a pattern of light that would be characteristic of what is now called a “kilonova” explosion. Warning: I don’t yet know much about kilonovas and I may be leaving out important details. A kilonova is powered by the process of forming heavy elements; most of the nuclei produced are initially radioactive — i.e., unstable — and they break down by emitting high energy particles, including the particles of light (called photons) which are in the gamma ray and X-ray categories. The resulting characteristic glow would be expected to have a pattern of a certain type: it would be initially bright but would dim rapidly in visible light, with a long afterglow in infrared light. The reasons for this are complex, so let me set them aside for now. The important point is that this pattern was observed, confirming that a kilonova of this type occurred, and thus that, in this neutron star merger, enormous amounts of heavy elements were indeed produced. So we now have a lot of evidence, for the first time, that almost all the heavy chemical elements on and around our planet were formed in neutron star mergers. Again, we could not know this if we did not know that this was a neutron star merger, and that information comes only from the gravitational wave observation.
MISCELLANEOUS QUESTIONS
Did the merger of these two neutron stars result in a new black hole, a larger neutron star, or an unstable rapidly spinning neutron star that later collapsed into a black hole?
We don’t yet know, and maybe we won’t know. Some scientists involved appear to be leaning toward the possibility that a black hole was formed, but others seem to say the jury is out. I’m not sure what additional information can be obtained over time about this.
If the two neutron stars formed a black hole, why was there a kilonova? Why wasn’t everything sucked into the black hole?
Black holes aren’t vacuum cleaners; they pull things in via gravity just the same way that the Earth and Sun do, and don’t suck things in some unusual way. The only crucial thing about a black hole is that once you go in you can’t come out. But just as when trying to avoid hitting the Earth or Sun, you can avoid falling in if you orbit fast enough or if you’re flung outward before you reach the edge.
The point in a neutron star merger is that the forces at the moment of merger are so intense that one or both neutron stars are partially ripped apart. The material that is thrown outward in all directions, at an immense speed, somehow creates the bright, hot flash of gamma rays and eventually the kilonova glow from the newly formed atomic nuclei. Those details I don’t yet understand, but I know they have been carefully studied both with approximate equations and in computer simulations such as this one and this one. However, the accuracy of the simulations can only be confirmed through the detailed studies of a merger, such as the one just announced. It seems, from the data we’ve seen, that the simulations did a fairly good job. I’m sure they will be improved once they are compared with the recent data.
