## The Bayesian Second Law of Thermodynamics

Entropy increases. Closed systems become increasingly disordered over time. So says the Second Law of Thermodynamics, one of my favorite notions in all of physics.

At least, entropy usually increases. If we define entropy by first defining “macrostates” — collections of individual states of the system that are macroscopically indistinguishable from each other — and then taking the logarithm of the number of microstates per macrostate, as portrayed in this blog’s header image, then we don’t expect entropy to always increase. According to Boltzmann, the increase of entropy is just really, really probable, since higher-entropy macrostates are much, much bigger than lower-entropy ones. But if we wait long enough — really long, much longer than the age of the universe — a macroscopic system will spontaneously fluctuate into a lower-entropy state. Cream and coffee will unmix, eggs will unbreak, maybe whole universes will come into being. But because the timescales are so long, this is just a matter of intellectual curiosity, not experimental science.

That’s what I was taught, anyway. But since I left grad school, physicists (and chemists, and biologists) have become increasingly interested in ultra-tiny systems, with only a few moving parts. Nanomachines, or the molecular components inside living cells. In systems like that, the occasional downward fluctuation in entropy is not only possible, it’s going to happen relatively frequently — with crucial consequences for how the real world works.

Accordingly, the last fifteen years or so has seen something of a revolution in non-equilibrium statistical mechanics — the study of statistical systems far from their happy resting states. Two of the most important results are the Crooks Fluctuation Theorem (by Gavin Crooks), which relates the probability of a process forward in time to the probability of its time-reverse, and the Jarzynski Equality (by Christopher Jarzynski), which relates the change in free energy between two states to the average amount of work done on a journey between them. (Professional statistical mechanics are so used to dealing with inequalities that when they finally do have an honest equation, they call it an “equality.”) There is a sense in which these relations underlie the good old Second Law; the Jarzynski equality can be derived from the Crooks Fluctuation Theorem, and the Second Law can be derived from the Jarzynski Equality. (Though the three relations were discovered in reverse chronological order from how they are used to derive each other.)

Still, there is a mystery lurking in how we think about entropy and the Second Law — a puzzle that, like many such puzzles, I never really thought about until we came up with a solution. Boltzmann’s definition of entropy (logarithm of number of microstates in a macrostate) is very conceptually clear, and good enough to be engraved on his tombstone. But it’s not the only definition of entropy, and it’s not even the one that people use most often.

Rather than referring to macrostates, we can think of entropy as characterizing something more subjective: our knowledge of the state of the system. That is, we might not know the exact position x and momentum p of every atom that makes up a fluid, but we might have some probability distribution ρ(x,p) that tells us the likelihood the system is in any particular state (to the best of our knowledge). Then the entropy associated with that distribution is given by a different, though equally famous, formula:

$S = - \int \rho \log \rho.$

That is, we take the probability distribution ρ, multiply it by its own logarithm, and integrate the result over all the possible states of the system, to get (minus) the entropy. A formula like this was introduced by Boltzmann himself, but these days is often associated with Josiah Willard Gibbs, unless you are into information theory, where it’s credited to Claude Shannon. Don’t worry if the symbols are totally opaque; the point is that low entropy means we know a lot about the specific state a system is in, and high entropy means we don’t know much at all.

In appropriate circumstances, the Boltzmann and Gibbs formulations of entropy and the Second Law are closely related to each other. But there’s a crucial difference: in a perfectly isolated system, the Boltzmann entropy tends to increase, but the Gibbs entropy stays exactly constant. In an open system — allowed to interact with the environment — the Gibbs entropy will go up, but it will only go up. It will never fluctuate down. (Entropy can decrease through heat loss, if you put your system in a refrigerator or something, but you know what I mean.) The Gibbs entropy is about our knowledge of the system, and as the system is randomly buffeted by its environment we know less and less about its specific state. So what, from the Gibbs point of view, can we possibly mean by “entropy rarely, but occasionally, will fluctuate downward”?

I won’t hold you in suspense. Since the Gibbs/Shannon entropy is a feature of our knowledge of the system, the way it can fluctuate downward is for us to look at the system and notice that it is in a relatively unlikely state — thereby gaining knowledge.

But this operation of “looking at the system” doesn’t have a ready implementation in how we usually formulate statistical mechanics. Until now! My collaborators Tony Bartolotta, Stefan Leichenauer, Jason Pollack, and I have written a paper formulating statistical mechanics with explicit knowledge updating via measurement outcomes. (Some extra figures, animations, and codes are available at this web page.)

The Bayesian Second Law of Thermodynamics
Anthony Bartolotta, Sean M. Carroll, Stefan Leichenauer, and Jason Pollack

We derive a generalization of the Second Law of Thermodynamics that uses Bayesian updates to explicitly incorporate the effects of a measurement of a system at some point in its evolution. By allowing an experimenter’s knowledge to be updated by the measurement process, this formulation resolves a tension between the fact that the entropy of a statistical system can sometimes fluctuate downward and the information-theoretic idea that knowledge of a stochastically-evolving system degrades over time. The Bayesian Second Law can be written as ΔH(ρm,ρ)+⟨Q⟩F|m≥0, where ΔH(ρm,ρ) is the change in the cross entropy between the original phase-space probability distribution ρ and the measurement-updated distribution ρm, and ⟨Q⟩F|m is the expectation value of a generalized heat flow out of the system. We also derive refined versions of the Second Law that bound the entropy increase from below by a non-negative number, as well as Bayesian versions of the Jarzynski equality. We demonstrate the formalism using simple analytical and numerical examples.

The crucial word “Bayesian” here refers to Bayes’s Theorem, a central result in probability theory. Continue reading

Posted in arxiv, Science, Time | 45 Comments

## Hypnotized by Quantum Mechanics

It remains embarrassing that physicists haven’t settled on the best way of formulating quantum mechanics (or some improved successor to it). I’m partial to Many-Worlds, but there are other smart people out there who go in for alternative formulations: hidden variables, dynamical collapse, epistemic interpretations, or something else. And let no one say that I won’t let alternative voices be heard! (Unless you want to talk about propellantless space drives, which are just crap.)

So let me point you to this guest post by Anton Garrett that Peter Coles just posted at his blog:

Hidden Variables: Just a Little Shy?

It’s quite a nice explanation of how the state of play looks to someone who is sympathetic to a hidden-variables view. (Fans of Bell’s Theorem should remember that what Bell did was to show that such variables must be nonlocal, not that they are totally ruled out.)

As a dialogue, it shares a feature that has been common to that format since the days of Plato: there are two characters, and the character that sympathizes with the author is the one who gets all the good lines. In this case the interlocutors are a modern physicist Neo, and a smart recently-resurrected nineteenth-century physicist Nino. Trained in the miraculous successes of the Newtonian paradigm, Nino is very disappointed that physicists of the present era are so willing to simply accept a theory that can’t do better than predicting probabilistic outcomes for experiments. More in sorrow than in anger, he urges us to do better!

My own takeaway from this is that it’s not a good idea to take advice from nineteenth-century physicists. Of course we should try to do better, since we should alway try that. But we should also feel free to abandon features of our best previous theories when new data and ideas come along.

A nice feature of the dialogue between Nino and Neo is the way in which it illuminates the fact that much of one’s attitude toward formulations of quantum mechanics is driven by which basic assumptions about the world we are most happy to abandon, and which we prefer to cling to at any cost. That’s true for any of us — such is the case when there is legitimate ambiguity about the best way to move forward in science. It’s a feature, not a bug. The hope is that eventually we will be driven, by better data and theories, toward a common conclusion.

What I like about Many-Worlds is that it is perfectly realistic, deterministic, and ontologically minimal, and of course it fits the data perfectly. Equally importantly, it is a robust and flexible framework: you give me your favorite Hamiltonian, and we instantly know what the many-worlds formulation of the theory looks like. You don’t have to think anew and invent new variables for each physical situation, whether it’s a harmonic oscillator or quantum gravity.

Of course, one gives something up: in Many-Worlds, while the underlying theory is deterministic, the experiences of individual observers are not predictable. (In that sense, I would say, it’s a nice compromise between our preferences and our experience.) It’s neither manifestly local nor Lorentz-invariant; those properties should emerge in appropriate situations, as often happens in physics. Of course there are all those worlds, but that doesn’t bother me in the slightest. For Many-Worlds, it’s the technical problems that bother me, not the philosophical ones — deriving classicality, recovering the Born Rule, and so on. One tends to think that technical problems can be solved by hard work, while metaphysical ones might prove intractable, which is why I come down the way I do on this particular question.

But the hidden-variables possibility is still definitely alive and well. And the general program of “trying to invent a better theory than quantum mechanics which would make all these distasteful philosophical implications go away” is certainly a worthwhile one. If anyone wants to suggest their favorite defenses of epistemic or dynamical-collapse approaches, feel free to leave them in comments.

Posted in Philosophy, Science | 100 Comments

## Spacetime, Storified

I had some spare minutes the other day, and had been thinking about the fate of spacetime in a quantum universe, so I took to the internet to let my feelings be heard. Only a few minutes, though, so I took advantage of Twitter rather than do a proper blog post. But through the magic of Storify, I can turn the former into the latter!

Obviously the infamous 140-character limit of Twitter doesn’t allow the level of precision and subtlety one would always like to achieve when talking about difficult topics. But restrictions lead to creativity, and the results can actually be a bit more accessible than unfettered prose might have been.

Anyway, spacetime isn’t fundamental, it’s just a useful approximation in certain regimes. Someday we hope to know what it’s an approximation to.

Posted in Science | 16 Comments

## Guest Post: Aidan Chatwin-Davies on Recovering One Qubit from a Black Hole

The question of how information escapes from evaporating black holes has puzzled physicists for almost forty years now, and while we’ve learned a lot we still don’t seem close to an answer. Increasingly, people who care about such things have been taking more seriously the intricacies of quantum information theory, and learning how to apply that general formalism to the specific issues of black hole information.

Now two students and I have offered a small contribution to this effort. Aidan Chatwin-Davies is a grad student here at Caltech, while Adam Jermyn was an undergraduate who has now gone on to do graduate work at Cambridge. Aidan came up with a simple method for getting out one “quantum bit” (qubit) of information from a black hole, using a strategy similar to “quantum teleportation.” Here’s our paper that just appeared on arxiv:

How to Recover a Qubit That Has Fallen Into a Black Hole
Aidan Chatwin-Davies, Adam S. Jermyn, Sean M. Carroll

We demonstrate an algorithm for the retrieval of a qubit, encoded in spin angular momentum, that has been dropped into a no-firewall unitary black hole. Retrieval is achieved analogously to quantum teleportation by collecting Hawking radiation and performing measurements on the black hole. Importantly, these methods only require the ability to perform measurements from outside the event horizon and to collect the Hawking radiation emitted after the state of interest is dropped into the black hole.

It’s a very specific — i.e. not very general — method: you have to have done measurements on the black hole ahead of time, and then drop in one qubit, and we show how to get it back out. Sadly it doesn’t work for two qubits (or more), so there’s no obvious way to generalize the procedure. But maybe the imagination of some clever person will be inspired by this particular thought experiment to come up with a way to get out two qubits, and we’ll be off.

I’m happy to host this guest post by Aidan, explaining the general method behind our madness.

If you were to ask someone on the bus which of Stephen Hawking’s contributions to physics he or she thought was most notable, the answer that you would almost certainly get is his prediction that a black hole should glow as if it were an object with some temperature. This glow is made up of thermal radiation which, unsurprisingly, we call Hawking radiation. As the black hole radiates, its mass slowly decreases and the black hole decreases in size. So, if you waited long enough and were careful not to enlarge the black hole by throwing stuff back in, then eventually it would completely evaporate away, leaving behind nothing but a bunch of Hawking radiation.

At a first glance, this phenomenon of black hole evaporation challenges a central notion in quantum theory, which is that it should not be possible to destroy information. Suppose, for example, that you were to toss a book, or a handful of atoms in a particular quantum state into the black hole. As the black hole evaporates into a collection of thermal Hawking particles, what happens to the information that was contained in that book or in the state of (what were formerly) your atoms? One possibility is that the information actually is destroyed, but then we would have to contend with some pretty ugly foundational consequences for quantum theory. Instead, it could be that the information is preserved in the state of the leftover Hawking radiation, albeit highly scrambled and difficult to distinguish from a thermal state. Besides being very pleasing on philosophical grounds, we also have evidence for the latter possibility from the AdS/CFT correspondence. Moreover, if the process of converting a black hole to Hawking radiation conserves information, then a stunning result of Hayden and Preskill says that for sufficiently old black holes, any information that you toss in comes back out almost a fast as possible!

Even so, exactly how information leaks out of a black hole and how one would go about converting a bunch of Hawking radiation to a useful state is quite mysterious. On that note, what we did in a recent piece of work was to propose a protocol whereby, under very modest and special circumstances, you can toss one qubit (a single unit of quantum information) into a black hole and then recover its state, and hence the information that it carried.

More precisely, the protocol describes how to recover a single qubit that is encoded in the spin angular momentum of a particle, i.e., a spin qubit. Spin is a property that any given particle possesses, just like mass or electric charge. For particles that have spin equal to 1/2 (like those that we consider in our protocol), at least classically, you can think of spin as a little arrow which points up or down and says whether the particle is spinning clockwise or counterclockwise about a line drawn through the arrow. In this classical picture, whether the arrow points up or down constitutes one classical bit of information. According to quantum mechanics, however, spin can actually exist in a superposition of being part up and part down; these proportions constitute one qubit of quantum information.

So, how does one throw a spin qubit into a black hole and get it back out again? Suppose that Alice is sitting outside of a black hole, the properties of which she is monitoring. From the outside, a black hole is characterized by only three properties: its total mass, total charge, and total spin. This latter property is essentially just a much bigger version of the spin of an individual particle and will be important for the protocol.

Next, suppose that Alice accidentally drops a spin qubit into the black hole. First, she doesn’t panic. Instead, she patiently waits and collects one particle of Hawking radiation from the black hole. Crucially, when a Hawking particle is produced by the black hole, a bizarro version of the same particle is also produced, but just behind the black hole’s horizon (boundary) so that it falls into the black hole. This bizarro ingoing particle is the same as the outgoing Hawking particle, but with opposite properties. In particular, its spin state will always be flipped relative to the outgoing Hawking particle. (The outgoing Hawking particle and the ingoing particle are entangled, for those in the know.)

The picture so far is that Alice, who is outside of the black hole, collects a single particle of Hawking radiation whilst the spin qubit that she dropped and the ingoing bizarro Hawking particle fall into the black hole. When the dropped particle and the bizarro particle fall into the black hole, their spins combine with the spin of the black hole—but remember! The bizarro particle’s spin was highly correlated with the spin of the outgoing Hawking particle. As such, the new combined total spin of the black hole becomes highly correlated with the spin of the outgoing Hawking particle, which Alice now holds. So, Alice measures the black hole’s new total spin state. Then, essentially, she can exploit the correlations between her held Hawking particle and the black hole to transfer the old spin state of the particle that she dropped into the hole to the Hawking particle that she now holds. Alice’s lost qubit is thus restored. Furthermore, Alice didn’t even need to know the precise state that her initial particle was in to begin with; the qubit is recovered regardless!

That’s the protocol in a nutshell. If the words “quantum teleportation” mean anything to you, then you can think of the protocol as a variation on the quantum teleportation protocol where the transmitting party is the black hole and measurement is performed in the total angular momentum basis instead of the Bell basis. Of course, this is far from a resolution of the information problem for black holes. However, it is certainly a neat trick which shows, in a special set of circumstances, how to “bounce” a qubit of quantum information off of a black hole.

Posted in arxiv, Guest Post, Science | 23 Comments

## Why is the Universe So Damn Big?

I love reading io9, it’s such a fun mixture of science fiction, entertainment, and pure science. So I was happy to respond when their writer George Dvorsky emailed to ask an innocent-sounding question: “Why is the scale of the universe so freakishly large?”

You can find the fruits of George’s labors at this io9 post. But my own answer went on at sufficient length that I might as well put it up here as well. Of course, as with any “Why?” question, we need to keep in mind that the answer might simply be “Because that’s the way it is.”

Whenever we seem surprised or confused about some aspect of the universe, it’s because we have some pre-existing expectation for what it “should” be like, or what a “natural” universe might be. But the universe doesn’t have a purpose, and there’s nothing more natural than Nature itself — so what we’re really trying to do is figure out what our expectations should be.

The universe is big on human scales, but that doesn’t mean very much. It’s not surprising that humans are small compared to the universe, but big compared to atoms. That feature does have an obvious anthropic explanation — complex structures can only form on in-between scales, not at the very largest or very smallest sizes. Given that living organisms are going to be complex, it’s no surprise that we find ourselves at an in-between size compared to the universe and compared to elementary particles.

What is arguably more interesting is that the universe is so big compared to particle-physics scales. The Planck length, from quantum gravity, is 10^{-33} centimeters, and the size of an atom is roughly 10^{-8} centimeters. The difference between these two numbers is already puzzling — that’s related to the “hierarchy problem” of particle physics. (The size of atoms is fixed by the length scale set by electroweak interactions, while the Planck length is set by Newton’s constant; the two distances are extremely different, and we’re not sure why.) But the scale of the universe is roughly 10^29 centimeters across, which is enormous by any scale of microphysics. It’s perfectly reasonable to ask why.

Part of the answer is that “typical” configurations of stuff, given the laws of physics as we know them, tend to be very close to empty space. (“Typical” means “high entropy” in this context.) That’s a feature of general relativity, which says that space is dynamical, and can expand and contract. So you give me any particular configuration of matter in space, and I can find a lot more configurations where the same collection of matter is spread out over a much larger volume of space. So if we were to “pick a random collection of stuff” obeying the laws of physics, it would be mostly empty space. Which our universe is, kind of.

Two big problems with that. First, even empty space has a natural length scale, which is set by the cosmological constant (energy of the vacuum). In 1998 we discovered that the cosmological constant is not quite zero, although it’s very small. The length scale that it sets (roughly, the distance over which the curvature of space due to the cosmological constant becomes appreciable) is indeed the size of the universe today — about 10^26 centimeters. (Note that the cosmological constant itself is inversely proportional to this length scale — so the question “Why is the cosmological-constant length scale so large?” is the same as “Why is the cosmological constant so small?”)

This raises two big questions. The first is the “coincidence problem”: the universe is expanding, but the length scale associated with the cosmological constant is a constant, so why are they approximately equal today? The second is simply the “cosmological constant problem”: why is the cosmological constant scale so enormously larger than the Planck scale, or event than the atomic scale? It’s safe to say that right now there are no widely-accepted answers to either of these questions.

So roughly: the answer to “Why is the universe so big?” is “Because the cosmological constant is so small.” And the answer to “Why is the cosmological constant so small?” is “Nobody knows.”

But there’s yet another wrinkle. Typical configurations of stuff tend to look like empty space. But our universe, while relatively empty, isn’t *that* empty. It has over a hundred billion galaxies, with a hundred billion stars each, and over 10^50 atoms per star. Worse, there are maybe 10^88 particles (mostly photons and neutrinos) within the observable universe. That’s a lot of particles! A much more natural state of the universe would be enormously emptier than that. Indeed, as space expands the density of particles dilutes away — we’re headed toward a much more natural state, which will be much emptier than the universe we see today.

So, given what we know about physics, the real question is “Why are there so many particles in the observable universe?” That’s one angle on the question “Why is the entropy of the observable universe so small?” And of course the density of particles was much higher, and the entropy much lower, at early times. These questions are also ones to which we have no good answers at the moment.

Posted in Science | 68 Comments

## Yoichiro Nambu

It was very sad to hear yesterday that Yoichiro Nambu has died. He was aged 94, so it was after a very long and full life.

Nambu was one of the greatest theoretical physicists of the 20th century, although not one with a high public profile. Among his contributions:

• Being the first to really understand spontaneous symmetry breaking in quantum field theory, work for which he won a (very belated) Nobel Prize in 2008. We now understand the pion as a (pseudo-) “Nambu-Goldstone boson.”
• Suggesting that quarks might come in three colors, and those colors might be charges for an SU(3) gauge symmetry, giving rise to force-carrying particles called gluons.
• Proposing the first relativistic string theory, based on what is now called the Nambu-Goto action.

So — not too shabby.

But despite his outsized accomplishments, Nambu was quiet, proper, it’s even fair to say “shy.” He was one of those physicists who talked very little, and was often difficult to understand when he does talk, but if you put in the effort to follow him you would invariably be rewarded. One of his colleagues at the University of Chicago, Bruce Winstein, was charmed by the fact that Nambu was an experimentalist at heart; at home, apparently, he kept a little lab, where he would tinker with electronics to take a break from solving equations.

Any young person in science might want to read this profile of Nambu by his former student Madhusree Mukerjee. In it, Nambu tells of when he first came to the US from Japan, to be a postdoctoral researcher at the Institute for Advanced Study in Princeton. “Everyone seemed smarter than I,” Nambu recalls. “I could not accomplish what I wanted to and had a nervous breakdown.”

If Yoichiro Nambu can have a nervous breakdown because he didn’t feel smart enough, what hope is there for the rest of us?

Here are a few paragraphs I wrote about Nambu and spontaneous symmetry breaking in The Particle at the End of the Universe. Continue reading

Posted in Science | 11 Comments

## Infinite Monkey Cage

The Infinite Monkey Cage is a British science/entertainment show put on by the dynamic duo of physicist Brian Cox and comedian Robin Ince. It exists as a radio program, a podcast, and an occasional live show. There are laughs, a bit of education, and some guests for the hosts to spar with. The popular-science ecosystem is a lot different in the UK than it is here in the US; scientists and science communicators can generally have a much higher profile, and a show like this can really take off.

So it was a great honor for me to appear as one of the guests when the show breezed through LA back in March. It was a terrific event, as you might guess from the other guests: comedian Joe Rogan, TV writer David X. Cohen, and one Eric Idle, who used to play in the Rutles. And now selected bits of the program can be listened to at home, courtesy of this handy podcast link, or directly on iTunes.

Be sure to check out the other stops on the IMC tour of the US, which included visits to NYC, Chicago, and San Francisco, featuring many friends-of-the-blog along the way.

These guys, of course, are heavy hitters, so you never know who is going to show up at one of these things. Their relationship with Eric Idle goes back quite a ways, and he actually composed and performed a theme song for the show (below). Naturally, since he was on stage in LA, they asked him to do a live version, which was a big hit. And there in the band, performing on ukulele for just that one song, was Jeff Lynne, of the Electric Light Orchestra. Maybe a bit under-utilized in this context, but why not get the best when you can?

Posted in Entertainment, Science and the Media | 12 Comments

## Why Is There Dark Matter?

Years ago I read an article by Martin Rees, in which he surveyed the options for what the dark matter of the universe might be. I forget the exact wording, but near the end he said something like “There are so many candidates, it would be quite surprising to find ourselves living in a universe without dark matter.”

I was reminded of this when I saw a Quantum Diaries post by Alex Millar, entitled “Why Dark Matter Exists.” Why do we live in a universe with five times as much dark matter as ordinary matter, anyway? As it turns out, the post was more about explaining all of the wonderful evidence we have that there is so much dark matter. That’s a very respectable question, one that I’ve covered now and again. The less-respectable (but still interesting to me) question is, Why is the universe like that? Is the existence of dark matter indeed unsurprising, or is it an unusual feature that we should take as an important clue as to the nature of our world?

With respect to dark matter, we’re contemplating a measure over all the various ways the universe could be, including both the laws of physics (which tell us what particles there can be) and the initial conditions (which set the stage for the later evolution). Clearly finding the “right” such measure is pretty much hopeless! But we can try to set up some reasonable considerations, and see where that leads us. Continue reading

Posted in Science | 53 Comments

## Algebra of the Infrared

In my senior year of college, when I was beginning to think seriously about graduate school, a magical article appeared in the New York Times magazine. Called “A Theory of Everything,” by KC Cole, it conveyed the immense excitement that had built in the theoretical physics community behind an idea that had suddenly exploded in popularity after burbling beneath the surface for a number of years: a little thing called “superstring theory.” The human-interest hook for the story was simple — work on string theory was being led by a brilliant 36-year-old genius, a guy named Ed Witten. It was enough to cement Princeton as the place I most wanted to go to for graduate school. (In the end, they didn’t let me in.)

Nearly thirty years later, Witten is still going strong. As evidence, check out this paper that recently appeared on the arxiv, with co-authors Davide Gaiotto and Greg Moore:

Algebra of the Infrared: String Field Theoretic Structures in Massive N=(2,2) Field Theory In Two Dimensions
Davide Gaiotto, Gregory W. Moore, Edward Witten

We introduce a “web-based formalism” for describing the category of half-supersymmetric boundary conditions in 1+1 dimensional massive field theories with N=(2,2) supersymmetry and unbroken U(1)R symmetry. We show that the category can be completely constructed from data available in the far infrared, namely, the vacua, the central charges of soliton sectors, and the spaces of soliton states on ℝ, together with certain “interaction and boundary emission amplitudes”. These amplitudes are shown to satisfy a system of algebraic constraints related to the theory of A∞ and L∞ algebras. The web-based formalism also gives a method of finding the BPS states for the theory on a half-line and on an interval. We investigate half-supersymmetric interfaces between theories and show that they have, in a certain sense, an associative “operator product.” We derive a categorification of wall-crossing formulae. The example of Landau-Ginzburg theories is described in depth drawing on ideas from Morse theory, and its interpretation in terms of supersymmetric quantum mechanics. In this context we show that the web-based category is equivalent to a version of the Fukaya-Seidel A∞-category associated to a holomorphic Lefschetz fibration, and we describe unusual local operators that appear in massive Landau-Ginzburg theories. We indicate potential applications to the theory of surface defects in theories of class S and to the gauge-theoretic approach to knot homology.

I cannot, in good conscience, say that I understand very much about this new paper. It’s a kind of mathematica/formal field theory that is pretty far outside my bailiwick. (This is why scientists roll their eyes when a movie “physicist” is able to invent a unified field theory, build a time machine, and construct nanobots that can cure cancer. Specialization is real, folks!)

But there are two things about the paper that I nevertheless can’t help but remarking on. One is that it’s 429 pages long. I mean, damn. That’s a book, not a paper. Scuttlebutt informs me that the authors had to negotiate specially with the arxiv administrators just to upload the beast. Most amusingly, they knew perfectly well that a 400+ page work might come across as a little intimidating, so they wrote a summary paper!

An Introduction To The Web-Based Formalism
Davide Gaiotto, Gregory W. Moore, Edward Witten

This paper summarizes our rather lengthy paper, “Algebra of the Infrared: String Field Theoretic Structures in Massive N=(2,2) Field Theory In Two Dimensions,” and is meant to be an informal, yet detailed, introduction and summary of that larger work.

This short, user-friendly introduction is a mere 45 pages — still longer than 95% of the papers in this field. After a one-paragraph introduction, the first words of the lighthearted summary paper are “Let X be a Kähler manifold, and W : X → C a holomorphic Morse function.” So maybe it’s not that informal.

The second remarkable thing is — hey look, there’s my name! Both of the papers cite one of my old works from when I was a grad student, with Simeon Hellerman and Mark Trodden. (A related paper was written near the same time by Gary Gibbons and Paul Townsend.)

Domain Wall Junctions are 1/4-BPS States
Sean M. Carroll, Simeon Hellerman, Mark Trodden

We study N=1 SUSY theories in four dimensions with multiple discrete vacua, which admit solitonic solutions describing segments of domain walls meeting at one-dimensional junctions. We show that there exist solutions preserving one quarter of the underlying supersymmetry — a single Hermitian supercharge. We derive a BPS bound for the masses of these solutions and construct a solution explicitly in a special case. The relevance to the confining phase of N=1 SUSY Yang-Mills and the M-theory/SYM relationship is discussed.

Simeon, who was a graduate student at UCSB at the time and is now faculty at the Kavli IPMU in Japan, was the driving force behind this paper. Mark and I had recently written a paper on different ways that topological defects could intersect and join together. Simeon, who is an expert in supersymmetry, noticed that there was a natural way to make something like that happen in supersymmetric theories: in particular, you could have domain walls (sheets that stretch through space, separating different possible vacuum states) could intersect at “junctions.” Even better, domain-wall junction configurations would break some of the supersymmetry but not all of it. Setups like that are known as BPS states, and are highly valued and useful to supersymmetry aficionados. In general, solutions to quantum field theories are very difficult to find and characterize with any precision, but the BPS property lets you invoke some of the magic of supersymmetry to prove results that would otherwise be intractable.

Admittedly, the above paragraph is likely to be just as opaque to the person on the street as the Gaiotto/Moore/Witten paper is to me. The point is that we were able to study the behavior of domain walls and how they come together using some simple but elegant techniques in field theory. Think of drawing some configuration of walls as a network of lines in a plane. (All of the configurations we studied were invariant along some “vertical” direction in space, as well as static in time, so all the action happens in a two-dimensional plane.) Then we were able to investigate the set of all possible ways such walls could come together to form allowed solutions. Here’s an example, using walls that separate four different possible vacuum states:

As far as I understand it (remember — not that far!), this is a very baby version of what Gaiotto, Moore, and Witten have done. Like us, they look at a large-distance limit, worrying about how defects come together rather than the detailed profiles of the individual configurations. That’s the “infrared” in their title. Unlike us, they go way farther, down a road known as “categorification” of the solutions. In particular, they use a famous property of BPS states: you can multiply them together to get other BPS states. That’s the “algebra” of their title. To mathematicians, algebras aren’t just ways of “solving for x” in equations that tortured you in high school; they are mathematical structures describing sets of vectors that can be multiplied by each other to produce other vectors. (Complex numbers are an algebra; so are ordinary three-dimensional vectors, using the cross product operation.)

At this point you’re allowed to ask: Why should I care? At least, why should I imagine putting in the work to read a 429-page opus about this stuff? For that matter, why did these smart guys put in the work to write such an opus?

It’s a reasonable question, but there’s also a reasonable answer. In theoretical physics there are a number of puzzles and unanswered questions that we are faced with, from “Why is the mass of the Higgs 125 GeV?” to “How does information escape from black holes?” Really these are all different sub-questions of the big one, “How does Nature work?” By construction, we don’t know the answer to these questions — if we did, we’d move onto other ones. But we don’t even know the right way to go about getting the answers. When Einstein started thinking about fitting gravity into the framework of special relativity, Riemannian geometry was absolutely the last thing on his mind. It’s hard to tell what paths you’ll have to walk down to get to the final answer.

So there are different techniques. Some people will try a direct approach: if you want to know how information comes out of a black hole, think as hard as you can about what happens when black holes radiate. If you want to know why the Higgs mass is what it is, think as hard as you can about the Higgs field and other possible fields we haven’t yet found.

But there’s also a more patient, foundational approach. Quantum field theory is hard; to be honest, we don’t understand it all that well. There’s little question that there’s a lot to be learned by studying the fundamental behavior of quantum field theories in highly idealized contexts, if only to better understand the space of things that can possibly happen with an eye to eventually applying them to the real world. That, I suspect, is the kind of motivation behind a massive undertaking like this. I don’t want to speak for the authors; maybe they just thought the math was cool and had fun learning about these highly unrealistic (but still extremely rich) toy models. But the ultimate goal is to learn some basic wisdom that we will someday put to use in answering that underlying question: How does Nature work?

As I said, it’s not really my bag. I don’t have nearly the patience nor that mathematical aptitude that is required to make real progress in this kind of way. I’d rather try to work out on general principles what could have happened near the Big Bang, or how our classical world emerges out of the quantum wave function.

But, let a thousand flowers bloom! Gaiotto, Moore, and Witten certainly know what they’re doing, and hardly need to look for my approval. It’s one strategy among many, and as a community we’re smart enough to probe in a number of different directions. Hopefully this approach will revolutionize our understanding of quantum field theory — and at my retirement party everyone will be asking me why I didn’t stick to working on domain-wall junctions.

Posted in arxiv, Science | 10 Comments

## So What Do You Do?

Kieran Healy has dusted off and re-posted some very good advice about attending academic conferences. It’s the advice you really need — who to go to dinner with, how not to embarrass yourself when introducing people to each other — rather than boring stuff like how to give a good talk. The spirit of the approach is captured by this quote:

As with teenagers, conference attendees secretly and falsely believe that other groups are having a much better time… Your conference strategies should therefore be geared towards counteracting the tendency to re-live your teenage years.

It’s surprising to realize how much “smart ways to behave at conferences” are really just “smart ways to behave in life.” (Though probably it shouldn’t be that surprising — academics aren’t the special flowers we like to think we are.) This bit of advice in particular struck me as useful:

[If] you worry someone will ask you what you work on, have something to say that’s three sentences long and takes fifteen seconds to get through. Write it down and practice it if you like.

I think every person should do that all the time. As you go through life, there will be multiple occasions on which people ask you “What do you do?” (If you’re in academia or an otherwise creative field, it will be “What are you working on?”) A high percentage of the time, questions like that elicit an awkwardly long pause, or something deflecting like “Oh, you know, lots of things.”

It makes sense. In your mind, “what you do” or “what you’re working on” is this incredibly rich, diverse, tightly interconnected set of things, and here is someone you don’t know asking you to instantly distill it down to a pithy phrase. Outrageous! And what’s worse, if you actually give a substantive answer, you’ll inevitably be leaving something out. You think, “Well I could mention this one thing that I’m mostly thinking about, but it’s not really representative of what I’m usually doing, so maybe I should mention this other thing…”

It’s not really a good look. Think about your own feelings when you ask someone what they do, and they respond with “Oh, I don’t know” or “Oh, lots of things.” Really? You don’t know what you do? Are you a spy whose memories are wiped at the conclusion of each mission? I’m sure you do many things, but perhaps picking out one would provide me with more useful information? At a conference in particular, it’s not the best first impression.

It’s important to come to terms with the fact that there are no perfect answers to the whatdoyoudo/whatareyouworkingon kinds of questions — and yet, we should have answers ready. Ones that are confident, short, and convey just a bit of the necessary flavor, so that more detail can emerge over the course of further conversation, which after all is the point of these well-meaning interrogations. This is especially true if you’re going to academic meetings, but it holds for life more generally. As adults, we should be better at these everyday skills than we were as teenagers.