The Future of Theoretical Cosmology

I’m back from an extraordinarily hectic yet unusually rewarding April Meeting of the American Physical Society in Dallas. The APS has two big meetings each year, the April meetings for very large- and small-scale types (particle physics, nuclear physics, gravitation, astrophysics), and the March meeting for medium-scale types (condensed matter, atomic physics, biophysics). The March meeting is a crucially important event for its constituency, while the April meeting suffers from too much competition and far less customer loyalty, and is correspondingly a much smaller conference (perhaps 1,000-1,500 attendees, as opposed to 6,000 at a typical March meeting). That’s a subject for another post, for those of you out there with an unhealthy interest in APS politics.

(For other reports from the meeting, see Jennifer Ouellette’s Cocktail Party Physics or the mysterious and anonymous Charm &c. Common refrain: “It’s 2006! Why isn’t there decent wireless in this hotel??!!”)

There’s a rule to the effect that any person can give no more than one invited talk at an APS meeting, but such rules are made to be broken and I sneaked in there with two talks. One was a general overview of the accelerating universe and its associated problems, at a special session on Research Talks Aimed at Undergraduates. Having a session devoted to undergrads was a splendid idea, although I suspect that the median age of attendees at my talk was something like 45. That’s because, when asked to pitch a talk to an audience of level of expertise x, most physicists will end up pitching it at a level of expertise x+3. So various people with Ph.D.’s concluded that their best chance of understanding a talk outside their specialty was to attend a session for undergraduates. Perhaps they were right. Before my talk they got to hear nice presentations by Florencia Canelli on particle physics and the top quark, and Paul Chaikin on packing ellipsoids. (Okay, “packing ellipsoids” doesn’t sound like the sexiest topic, but it was filled with fascinating tidbits of information. Did you know that both prolate and oblate ellipsoids pack more efficiently than spheres? That ordered crystalline packings are generally found to be more efficient than random packings, but nobody can prove it in general? That M&M’s are extremely reliable ellipsoids, to better than 0.1%? That the method by which the Mars Corporation makes their M&M’s so regular is a closely guarded secret?)

My other talk was at a joint double session on the past, present, and future of cosmology, co-sponsored by the Division of Astrophysics and the Forum for the History of Physics. Six talks naturally needed to be given: one each on the past/present/future of observational/theoretical cosmology, and organizer Virginia Trimble invited me to speak on The Future of Theoretical Cosmology. The observational session conflicted with my talk to the “undergrads,” but I got to hear the talks on the past and present of theory by Helge Kragh and David Spergel, respectively.

Of course nobody has any idea what the future of theoretical cosmology will be like, given that we know neither what future experiments will tell us, nor what ideas future theorists will come up with. So I defined “the future” to be “100 years from now,” by which time I figured (1) I won’t be around, or (2) if I am around it will be because we will all be living in pods and communicating via the Matrix, and nobody will be all that interested in what I said about the future of cosmology a century earlier.

interactive dark sector

With those caveats in mind, I did try to make some prognostications about how we will be thinking about three kinds of cosmological issues: composition questions, origins questions, and evolution questions. You can peek at my slides in html or pdf, although I confess that many were cannibalized from other talks. The abbreviated version:

  • Composition Questions. We have an inventory of the universe consisting of approximately 4% ordinary matter, 22% dark matter, and 74% dark energy. But each of these components is mysterious: we don’t know what the dark matter or dark energy really are, nor why there is more matter than antimatter. My claim was that we will have completely understood these questions in 100 years. In each case, there is an active experimental program aimed at providing us with clues, so I’m optimistic that the matter will be closed long before then.
  • Origins Questions. Where did the universe come from, and why do we find it in this particular configuration? Inflation, which received an important boost from the recent WMAP results, is a crucial ingredient in our current picture, but I stressed that there is a lot that we don’t yet understand. In particular, we need to understand the pre-inflationary universe to know whether inflation really provides a robust theory of initial conditions. Thinking about inflation naturally leads us to the multiverse, and I argued that untestable predictions of a theory are perfectly legitimate science, so long as the theory makes other testable predictions. We don’t yet have a theory of quantum gravity that does that, and I prevaricated about whether one hundred years would be sufficient time to establish one. (Naive extrapolation predicts that we won’t be doing Planck-scale experiments until two hundred years from now.)
  • Evolution Questions. Given the initial conditions, we already understand the evolution of small perturbations up to the point where they become large (“nonlinear”). That’s when numerical simulations become crucial, and here I was a little more bold. The very idea of a computer simulation is only about 50 years old, so there’s every reason to expect that the way in which computers are used will look completely different 100 years from now. Quantum computers will be commonplace, and enable parallel processing of enormous power. More interestingly, the types of computation that we’ll be doing will be dramatically different; I suggested that the computers will not only be running simulations to test theories against observations, but will be coming up with theories themselves. Such a prospect is a natural outgrowth of the idea of genetic algorithms, so I don’t think it’s as crazy as it sounds.

The next day I managed to catch no fewer than three sessions filled with provocative talks — one on ultra-high-energy cosmic rays, one on cosmology and gravitational physics, and one on precision cosmology. And I would tell you all about them if I hadn’t lost the keys to my special time-stretching machine that allows me to put aside my day job for arbitrarily long periods so that I can blog at leisure. Probably the most intriguing suggestions were those by Shamit Kachru from SLAC, who argued that considerations from string theory (and in particular the constraint that scalar fields cannot evolve by amounts greater than the Planck scale) imply that gravitational waves produced by inflation will never be strong enough to be observable in the CMB, and those by David Saltzberg from UCLA, who listed an amazing variety of upcoming experiments to detect high-energy astrophysical neutrinos, including listening for sound waves (!) produced when a neutrino interacts with ocean water off the Bahamas. If I decide to become an experimentalist, that’s the one I’m joining.

13 Comments

13 thoughts on “The Future of Theoretical Cosmology”

  1. I got a chance to see Sean’s talk while at the APS meeting (standing room only), and it was very interesting and good fun.

    The bit about quantum computers and genetic algorithms was a little unexpected. However, the general point made sense. Given the youth of computer science (half the age of cosmology), we have no idea what will be computationally possible in 100 years.

    At one point, I was with a bunch of US students who got a chance to meet Gerard ‘t Hooft, and we got to talking about his current work. He’s convinced that quantum computing will never work, and not for engineering reasons, but rather for fundamental physics reasons. (“Reason” is not the best term. Perhaps “suspicion” is better.) He was kind of cryptic on the topic, but his pessimism sounded like the inverse of Sean’s optimism. Quantum computers were too good to be true (simulate the universe with a subset of the universe?!), and we would hit some wall where decoherence or even a new feature of QM always blocked our ability to extract new information from the quantum computer. ‘t Hooft was looking into how to construct a classical, non-local (to avoid Bell’s theorem), statistical mechanics-like theory that would appear like quantum mechanics. An interesting, crazy idea which I look forward to seeing land in the arXiv some day if he ever makes it work.

    Also, between Sean’s talk and Nima Arkani-Hamed’s talk, I got a better understanding of what The Landscape means as far as physics goes. Sean pointed out it was an untestable feature of a theory, but not a theory itself. A theory with a landscape isn’t necessarily Not Science, as long as it has other testable features. Then I saw Nima’s talk, which closed the loop, and convinced me that the landscape really is a philosophy that says our definition of “naturalness” might be too strict, and we are tossing out theoretical ideas for aesthetic reasons which could make useful predictions. His idea for split-supersymmetry needs a flexible definition of “natural”, but it also makes very concrete predictions.

  2. One proposed addition to your list: Does the Equivalence Principle have a parity violation? Weak interactions (e.g., the Weak Interaction) violate parity conservation. Gravitation is the weakest interaction.

    Either way, half of contemporary gravitation theory is dead wrong.

    Gravitation theory can be written parity-even or parity-odd; spacetime curvature vs. spacetime torsion. Classical gravitation has Green’s function Newton and metric Einstein vs. affine Weitzenböck and teleparallel Cartan. String theory has otherwise and heterotic subsets. Though their maths are wildly different, their testable empirical predictions are exactly identical…

    … with one macroscopic disjoint exception: Do identical chemical composition local left and right hands vacuum freefall identically? Parity-even spacetime is blind to geometric parity (chirality simultaneously in all directions). Parity-odd spacetime would manifest as a background pseudoscalar field. The left foot of spacetime would be energetically differently fit by a sock or left shoe vs. a right shoe.

    String theory could be marvelously pruned. Does a single crystal solid sphere of space group P3(1)21 quartz (right-handed screw axes) vacuum freefall identically to an otherwise macrscopically identical single crystal solid sphere of space group P3(2)21 quartz (left-handed screw axes)? Both will fall along minimum action paths. In parity-odd spacetime those local paths will be diastereotopic and measurably non-parallel.

    A chiral background pseudoscalar field diverges Big Bang evolution of matter and antimatter. It sources biological homochirality (exclusively left-handed chiral protein amino acids and right-handed chiral sugars). At a 10^(-10) difference/average or smaller level it is consistent with 420+ years of physical observations.

    The world has lots of Eötvös balances with 10^(-13) difference/average sensitivity. The proper challenge of geometric gravitation is test mass geometry. More than 2100 tonnes of single crystal quartz are grown annually. Somebody should look.

  3. Sean,

    Enjoyed your talks at the meeting, you lend a humor to the subject that isn’t found in many other talks and tends to keep the audience much more engaged. Thanks for putting the slides online, perhaps I will lift a few of your graphics for my presentations (don’t worry, I’ll give credit where it’s due).

    Best Regards,
    Adam Drake

  4. Pingback: Political Apathy

  5. 1000-1500 attendants sounds like a nice numbers…. The last AAS meeting had over 3000, and it felt like too much to me. It was the first meeting where I really noticed that there were people I knew were there, but I never ran into. It was too much to take in all at once.

  6. I like the idea of computers coming up with theories. I would go even further. In the future, about a few hundreds years from now, computers will be much more powerful than the human brain. Machines will have replaced humans by that time so they would probably do physics, if they are interested in it.

  7. Oh… the sound of neutrinos surfing the oceanic waves! “The Music of The Neutrinos” might gain an edge over “The Music of The Primes”. Great closing story on the future of theoretical cosmology!

  8. Torbjörn Larsson

    “untestable predictions of a theory are perfectly legitimate science, so long as the theory makes other testable predictions.”

    Yes, I finally got that it’s ok to add unfalsifiable hypotheses some time ago, which is why I stopped arguing against the string landscape (since string theory is already theoretically tested against older physics) and also tentatively embraced the parsimonious anthropic view. So I will take the opportunity to retract those earlier arguments here, for what it is worth. 🙂

  9. Talking about the multiverse, anthropic things, etc, I hope everyone saw this rather cool paper:

    http://arxiv.org/abs/physics/0604134

    This, along with Polchinski’s latest paper, have caused me to seriously reconsider my previous strong opposition to multiverses and such things. I urge people to read these two papers, forget that they ever heard the name Susskind, and think again about their position on these issues.

  10. Pingback: Odyssey and Oracle

  11. Pingback: Charm &c. » Blog Archive » Dallas redux

Comments are closed.

Scroll to Top