About every three years KEK issues a hype-filled press release announcing that Jun Nishimura and collaborators have used a supercomputer to get evidence for string theory. Back in 2008, the announcement was of a numerical simulation on a supercomputer of a supersymmetric QM system that supposedly showed that superstring theory explained the properties of black holes (press release here, preprint here, blogging here). In 2011, the claim was of a numerical simulation on a supercomputer that used superstring theory to understand the birth of our universe (press release here, preprint here, blogging here). Both of these papers were published in PRL.
The 2014 press release is now out (see here), based on this preprint from last December. The latest claim is that the authors have solved the black hole information paradox, have shown that we live in a hologram, as well as showing that string theory provides a self-consistent quantization of gravity, all by doing a numerical simulation of a QM system. Even better, they have made the quantum gravity problem just as well-understood and tractable as QCD:
In short, we feel that problems involving quantum gravity have become as tractable as problems involving the strong interaction. The latter can be studied by simulating a gauge theory on a four-dimensional (4D) lattice, and such a method has recently been used to reproduce the mass spectrum of hadrons (28) and the nuclear force (29). We can now apply essentially the same method to study quantum gravity, which has been thought to be far more difficult.
This latest version of the KEK-hype has gotten a lot more attention than the previous two versions. Based on the preprint, late last year for some reason Nature covered this with a story about how Simulations back up theory that Universe is a hologram and this got a lot of media attention (see here for example).
The paper has now been published, and this time it’s not in PRL, but in Science magazine (submission there was a month after the preprint came out, could it be that PRL wouldn’t have it?). Science is giving it a high profile, including together with it a piece by Juan Maldacena, which claims the paper as “further evidence of the internal consistency of string theory”. Science provides the following one-line summary of the Maldacena piece:
A numerical test shows that string theory can provide a self-consistent quantization of gravity.
One obvious problem with this is that even if you take the most optimistic view of it all, what is being described is quantum gravity in 10d space-time. The Japanese authors deal with this problem with a footnote:
Theoretical consistency requires that superstring theory should be defined in ten-dimensional space-time. In order to realize our four-dimensional space-time, the size of the extra six dimensions can be chosen to be very small without spoiling the consistency.
Remarkably, Maldacena has another answer: the multiverse, which now he seems to take as accepted fact.
Of course, the 10-dimensional space under consideration here is not the same as the four-dimensional region of the multiverse where we live. However, one could expect that such holographic descriptions might also be possible for a region like ours.
Absurd hype about string theory is a continuing problem, and it’s not one that can be blamed on journalists, with this latest example getting help from HEP lab press releases, a highly reputable journal, and an IAS faculty member.
From what I can tell, these aren’t simulations (in the usual sense). There is no dynamics, nothing is evolving. They simply used a computer to compute a partition function. The integral is too hard to compute analytically so they discretized it and evaluated it numerically. From what I can tell, that’s the only reason a computer was involved. As there is no Hawking radiation or dynamics in the calculation, I don’t think it gives any direct insight into the information problem.
…
As for the result, maybe I can try to summarize what they did and someone more knowledgeable can correct me. On the “gravity side” they considered a 10 dimensional black hole. It is a solution of Einstein gravity with some extra higher curvature terms in the action. They computed this black hole’s Hawking temperature and entropy in the usual way, and then they integrated dE=TdS to find its internal energy as a function of temperature, E(T).
…
On the “gauge theory side,” they considered a specific gauge theory action and computed its partition function, Z, in the usual way. Then they took a derivative with respect to inverse temperature to get E(T) = \partial_\beta Z. They showed that this matches the E(T) of the black hole. It’s another consistency check of AdS/CFT.
rfp, they are simulating a quantum state numerically. If they did everything right, the partition function is all the dynamics they need.
That said, I have another issue: if you read this paper, they achieve something conventional lattice theorists are _very_ far away from doing (simulating a supersymmetric theory, with Fermions and everything) using a tower of assumptions:
the field theory is reduced to quantum mechanics, the regulator is a momentum cutoff, fermionic complex phases are ignored, a complex partition function is artifically truncated, etc.
Peter, perhaps, given your expertise, you could comment on whether these assumptions are plausible from a numerical point of view.
I am disturbed that the only calculations made with this method are the ones which “test the theory”. No exact solution or toy model exists which shows that the approximations are under control.
I am sure Science has conducted refereeing, but on the other hand Maldacena is not a lattice theorist, and no other research group working in numerical field theory is using this approach and nd some of them are working on supersymmetric problems.
This sounds very dubious and not different from what Peter was complaining about. There’s no control in their approximations, and they’ve simply skipped over all the hard parts that keep the lattice gauge theory people sweating. That amounts to cheating and then making crazy claims afterwards.
It certainly does nothing to prove string theory, or even that such a theory exists. It is merely another check (if it can be called even that) on the CFT on AdS space. Everything else is conjecture.
On a somewhat related aside, Renate Loll just gave a talk at Perimeter advocating for greater utilization of numerical simulations in checking various solution spaces of competing theories of quantum gravity. Interesting and worth watching (in my opinion).
http://pirsa.org/14050138/
Dear Peter,
you go too fast. I am still recovering from a lose cable that made neutrinos exceed the speed of light. And of a false understanding of a snapshot of a map presented in a meeting that proved inflation at 2 10^16 GeV. I must admit that the former would have broken all bounds, while the latter extrapolates no more than a mere 29 orders of magnitude from its observations at 3K. Now you come with stories that we live in a 10d multiverse, string theory is proven, quantum gravity is established. Is anybody wondering which questions, if any, we will leave for posterity?
One obvious problem with this is that even if you take the most optimistic view of it all, what is being described is quantum gravity in 10d space-time.
^
if this is the most optimistic view of it all are there other less optimistic views and theoretical reasons not to be optimistic?
…and then it turns out they put in the wrong sign in a forgotten corner of badly reviewed FORTRAN crud.
Is anybody wondering which questions, if any, we will leave for posterity?
Don’t fear, if we go on like this, there’ll be lots of questions of physics we leave them, and actually we’ll add some more – like the important one: “How could this happen?”
(After looking at the papers) Much less than the hype. As many have pointed out, any semiclassical theory of gravity has to produce the general form for black hole entropy. The numerical calculation gets you, in addition, some dimensionless factors for that particular model. Not much new. It certainly doesn’t prove string theory, whatever that would mean.
Comparing quantum gravity theories through numerical simulations is not a bad idea IF the approximations are controlled. We do need to shake off another piece of hype, that simulation constitutes a “third branch” of science, along side Experiment and Theory. Actually, simulation (if done right) is just another part of Theory. It’s another set of approximations leading to approximate answers. If done without control over the approximations, simulation is just a poorly constructed “model,” or really, just Bad Theory.
Peter, something OT.
See this nice paper by Avi Loeb on how there were prejudices against new ideas(which later turned out to be correct) which hurt the subject
http://arxiv.org/abs/1405.2954
Can you (or someone else) think of similar examples in particle physics
Shantanu
All humans are biased, it is in our nature. For every case of someone rejecting a valid new idea because of prejudices against new ideas, there are many more cases of someone rejecting a valid old idea in favour of their new idea that later turns out to be wrong. Cherry picking cases studies to support your thesis is a form of bias & is a very unscientific way to conduct research.
Off topic – I know – sorry
Check out Nature for this week. I no longer admit to being a theoretical physicist.
jd, when you say “check out Nature for this week” I assume you refer to the article by Amanda Gefter
Theoretical physics: Complexity on the horizon
A concept developed for computer science could have a key role in fundamental physics — and point the way to a new understanding of space and time.
At the moment I do not have ready access to the article, can you tell us a bit about it?
Marcus
Yes, you have the correct article.
Let’s see if I can summarize this. The news article quickly goes over Susskind’s latest ideas on how to solve the “firewall problem” and how to go much further. Susskind is apparently wearing a T shirt these days that says “I heart Complexity,” with the heart being a Mandelbrot set, or as much as you can get on a T shirt. The complexity is referring to computational complexity. Black holes, worm holes, computational complexity, AdS/CFT, and more are all invoked in the picture. Basically, the firewall does not exist (it is not needed) because the information just outside the event horizon is too complex to grab and take inside the event horizon. Then you can not violate the no-cloning theorem. The article goes on to mention that there may be connections between the length of a wormhole and complexity and, therefore, also time. Furthermore, entanglement is somehow related to space. Susskind is quoted. “I do not know where all of this will lead. But I believe these complexity-geometry connections are the tip of an iceberg.” I wonder if the lookouts on the Titanic said something similar.
Oh, there was also, “Things fall because there is a tendency toward complexity.”
Jd: this is hilarious… Perhaps peter can do a post about
Fred kantor, a curious character (former phd student) who hangs around
The Columbia physics department advertising his theory
Known as “information dynamics”
It’s basically this article, the same ideas and the same level of “rigor”
Amanda Gefter’s story, like all of Nature News’ content, is available to read for free. Here’s the link:
Theoretical physics: Complexity on the horizon
David Castelvecchi: That link is disturbing! Looking forward to our host’s views on the ideas.
What’s wrong with the idea that computational complexity has a role in physics?