Is string theory a theory of quantum gravity?

[marcus]

http://arxiv.org/abs/1105.6359
Is string theory a theory of quantum gravity?
Steven B. Giddings
(Submitted on 31 May 2011)
Some problems in finding a complete quantum theory incorporating gravity are discussed. One is that of giving a consistent unitary description of high-energy scattering. Another is that of giving a consistent quantum description of cosmology, with appropriate observables. While string theory addresses some problems of quantum gravity, its ability to resolve these remains unclear. Answers may require new mechanisms and constructs, whether within string theory, or in another framework.
32 pages, 5 figures. Invited contribution for "Forty Years of String Theory: Reflecting on the Foundations," a special issue of Found. Phys., ed. by G 't Hooft, E. Verlinde, D. Dieks, S. de Haro.
===========================

EDIT: some interesting side information is that Foundations of Physics is devoting a special issue to Forty Years of String Theory. Gerard 't Hooft is the journal's chief editor. He has three other people help put the special issue on String together as co editors.
Sebastian de Haro's PhD advisor was Gerard 't Hooft. He got his PhD in 2001, I think. I'm not familiar with Dennis Dieks. Everybody knows who Erik Verlinde is. I'll put some links to some of these people's research:
de Haro http://arxiv.org/find/hep-th/1/au:+Haro_S/0/1/0/all/0/1
Dieks http://arxiv.org/find/grp_physics/1/au:+Dieks_D/0/1/0/all/0/1

 

[atyy]

His question on local observables is interesting. I think it's related to comments like only the S-matrix and on-shell quantities make sense in quantum gravity because of general covariance and background independence http://arxiv.org/abs/0908.0333 (p126). Or that in de Sitter space sharp observables seem unavailable to someone in the universe http://arxiv.org/abs/hep-th/0106109.

Motl wrote a great post in which he says "But because it seems clear that we can never measure things absolutely accurately in a de Sitter space anyway, it is questionable whether we really want and need a theory that can predict things absolutely accurately. Maybe we don't. If we don't want it, it still remains puzzling what it means to have a full theory that inherently predicts all probabilities inaccurately. http://motls.blogspot.com/2008/10/observables-in-quantum-gravity.html"

One strand of thinking, going back to DeWitt is that observables should somehow be "relational". Some recent papers are:
Rovelli http://arxiv.org/abs/gr-qc/0110035
Giddings, Marolf, Hartle http://arxiv.org/abs/hep-th/0512200

Do people agree this is a problem? Do Rovelli's and Giddings et al's sort of proposals point to a solution?

 

[Haelfix]

Before you can even formulate a theory, physicists first have to ask, what is it that I can measure and with which my theory can make predictions with? So for instance in particle physics we measure decay rates, scattering cross sections which are encoded into a mathematical object like the Smatrix. Only later do we puzzle out the details of the dynamics.

So when you formulate a theory of quantum gravity, what are the observables? Instant problem, DeWitt showed long ago that there was a major incompatibility between diffeomorphism invariance, dynamical gravity and local observables in the quantum sense. And in general, no such observables exist at all -ever- (this is why if your theory outputs say, a fundamental local area observable, you can instantly throw the theory out as it is at best an approximation). The only hope is to look for boundary observables, of which the Smatrix is an example (in asymptotically flat spaces) or correlator functions on the AdS boundary (DeSitter space apparently has no observables at all!). But again, no generalized object exists that subsumes the entire theory has ever been found, not even relational observables or quasi local ones (both of which are difficult to define quantum mechanically as they must necessarily ignore the fluctuations of the gravitational field itself).

This picture was further complicated when the black hole information paradox arose, and it was found out that there was a major clash between unitarity on one side, and the local formulation of the problem on the other. The only viable solution that has been found to date was to invoke what is known as black hole complementarity and holography.

The punchline is that something must be a little bit nonlocal in a quantum theory of gravity, but not so nonlocal that it messes with cherished symmetry principles that must hold to fantastic accuracy (like Lorentz invariance). String theory seemed to provide an example of this (in the sense that Giddings describes in his paper), and AdS/CFT seemed to provide an example of holography at work. Of course the question then is, how exactly does this work in microscope detail? We obviously don't live or perform experiments off at infinity somewhere, and we'd like to ask questions about a free falling observer falling into say a horizon. So what are the approximate observables in question, how do they behave and in what sense do they break down?

The problem is that this has so far been elusive, even in the most well defined and concrete example known: AdS/CFT, and worse there seems to be persistent technical problems in asking that question. Therefore the Giddings paper poses a natural question, and that is whether or not the duality is completely exact... Are we still tied to a formulation of physical law that is still too localized?

 

[mitchell porter]

Giddings gave a talk at the KITP http://online.itp.ucsb.edu/online/qcdscat11/giddings/" .

I'm very interested in what he has to say but need some time to consider it. I can see that he has two chief propositions. First, "ultraplanckian scattering in the strong-gravity regime manifests a unitarity crisis" (page 14). Second, demonstrating that the AdS/CFT correspondence really is an equivalence faces identifiable technical problems (page 22). His suggestion that maybe it's not an equivalence, and that the boundary theory only approximates the bulk theory, is iconoclastic and I don't believe it, but it's the sort of informed skepticism which can spur others towards the complete and proper solution of a problem. (I had my eye on http://arxiv.org/abs/1102.2910" as an approach to the problem of local observables in the bulk.)

 

[Fra]

I agree that the key question of observables are one of the key pivot points that seems to be revealing of when people say what they think.

1) In the first papter Giddings says

"...local observables aren’t gauge invariant. A proposed resolution of this – whose semiclassical form is used in current studies of inflationary cosmology – is that local observables are approximately recovered relationally"

This is one possible resolution, and smells very much Rovelli and structural realism. This sounds good, but the problem of Rovellis view that I recall from this RQM paper is that the relations themselves aren't relational, their communication are described by QM (which is take as is).

It symbolizes a strong realist attutide towards the notion of "symmetry".

2) The other option, that I personally find much more reasonable and rational is that it's rather the other way around: global-observables ang symmetries are approximately recovered asymptotically as the observer complexity goes to infinity and/or as the system complexity goes to zero.

This symbolizes the view of observer symmetries as emergent within a particular observer complex, rather than been imposed as realist constraints. The symmetries can still be unique as unique limiting symmetries when you observer a subsystem.

I think the understanding of how symmetries as inferred or just as realist elements is the key problem here.

Is is scientifically rational, to accept a symmetry principle as constraining the action in a measurment theory if this symmetry is not itself inferred from experiment? The point would be that in actual physics this symmetry IS inferred; BUT this inference is only valid for small subsystems where the asymptotic obsevables can in fact at least approximately be realized.

Anything beyond that is IMHO, clear speculation, or subscription to unjustified realism views of symmetry principles.

This is why I don't think Rovelli's reasoning (as far as I understood the last versions I've read) is the resolution.

Edit: In addition one can characterize
(2) is less precise int he predictions than (1). (1) is deductive, but the choice of deduction is not rationally inferred. (2) is not deductive, bur rather softly inductive, but the induction is rationally inferred and thus apparently more defendable.

/Fredrik

 

[unusualname]

EDIT: some interesting side information is that Foundations of Physics is devoting a special issue to Forty Years of String Theory. Gerard 't Hooft is the journal's chief editor. He has three other people help put the special issue on String together as co editors.
Sebastian de Haro's PhD advisor was Gerard 't Hooft. He got his PhD in 2001, I think. I'm not familiar with Dennis Dieks. Everybody knows who Erik Verlinde is. I'll put some links to some of these people's research:
de Haro http://arxiv.org/find/hep-th/1/au:+Haro_S/0/1/0/all/0/1
Dieks http://arxiv.org/find/grp_physics/1/au:+Dieks_D/0/1/0/all/0/1

String Theory's been doing 40 year specials for a few years now, here's a Physics World article from 2007 which includes contributions from some of the famous string theory contributors and skeptics

http://physicsworld.com/cws/article/print/30940

't Hooft suggests that the debate shouldn't occur outside "professional circles" (hmm, maybe that's feasible for a few years but not for decades I don't think)

 

[marcus]

... a Physics World article from 2007 which includes contributions from some of the famous string theory contributors and skeptics

http://physicsworld.com/cws/article/print/30940
...

Thanks for the link, I gather it's to an article by [science journalist*] Matthew Chalmers that contains quotes gathered from various notables?
"In its near 40-year history, string theory has gone from a theory of hadrons to a theory of everything to, possibly, a theory of nothing. Indeed, modern string theory is not even a theory of strings but one of higher-dimensional objects called branes. Matthew Chalmers attempts to disentangle..."

I'm curious to see what happens with the special issue of Foundations of Physics (FP). It could be a while before it appears, or before the other invited articles are posted on arxiv. Presumably they will be peer-reviewed professional level, leaning more to the philosophy of science side---perhaps more balanced critical assessment than pro/con opinion.

*He describes himself as a science journalist and freelance writer, but he has a PhD in particle physics. Well qualified!

 

[unusualname]

Thanks for the link, I gather it's to an article by Matthew Chalmers that contains quotes assembled from various notables?

Yes, don't know if it's well known here, apologies if so. I wonder how many of the views expressed have changed (at all) in the last four years? I wonder if 't Hooft, Verlinde et al will get an illustrious group of contributors. Dieks is editor of a collection on "The Ontology of Spacetime" from a few years ago (amazon link)

 

[marcus]

Yes, don't know if it's well known here, apologies if so. I wonder how many of the views expressed have changed (at all) in the last four years? I wonder if 't Hooft, Verlinde et al will get an illustrious group of contributors. Dieks is editor of a collection on "The Ontology of Spacetime" from a few years ago (amazon link)

I checked http://www.scienceblogs.de/lindaunobel/2009/07/a-non-debate.php to see who Matthew Chalmers is. Got a very good (quick first) impression. Physics PhD who takes science journalism seriously. Wish there were more like that.

I checked your amazon link to Ontology of Spacetime (Dieks ed) and saw it was a heavy-reading scholarly book with chapters by John Earman, Carlo Rovelli, Harvey Brown...
Dieks seems to be a professor somewhere (Utrecht?) with a specialty in Phil of Phys? These are all good signs!

It indicates to me (at first sight) that the special issue will not be part of the "debate" about String where often the happytalk, tit-for-tat, defensiveness, point-scoring, and spin drown out careful critical accounting.

It may indeed get very little circulation! That's not a bad thing. It is more important that a balanced realistic assessment be made than that it be widely circulated---as long as it gets to the intended audience.

I'm actually a bit impatient now to see what comes out of this.

 

[marcus]

Atyy and Fra, you put the spotlight on a fascinating bunch of questions! I think the balanced way Giddings confronts them in this paper is a good sign.

One strand of thinking, going back to DeWitt is that observables should somehow be "relational". Some recent papers are:
Rovelli http://arxiv.org/abs/gr-qc/0110035
Giddings, Marolf, Hartle http://arxiv.org/abs/hep-th/0512200
...

I agree that the key question of observables are one of the key pivot points that seems to be revealing of when people say what they think.

1) In the first papter Giddings says

"...local observables aren’t gauge invariant. A proposed resolution of this – whose semiclassical form is used in current studies of inflationary cosmology – is that local observables are approximately recovered relationally"

This is one possible resolution, and smells very much Rovelli and structural realism...

It symbolizes a strong realist attutide towards the notion of "symmetry".

2) The other option, that I personally find much more reasonable and rational is that it's rather the other way around: global-observables ang symmetries are approximately recovered asymptotically as the observer complexity goes to infinity and/or as the system complexity goes to zero.

This symbolizes the view of observer symmetries as emergent within a particular...

I want to quote some passages from the Giddings paper primarily to gauge the tone and the extent of fairness/openness. I see him as a string community bellwether.

 

[Fra]

I like Haelfix summary, it describes the somewhat mutually conflicting traits.

The punchline is that something must be a little bit nonlocal in a quantum theory of gravity, but not so nonlocal that it messes with cherished symmetry principles that must hold to fantastic accuracy (like Lorentz invariance).
...
Of course the question then is, how exactly does this work in microscope detail? We obviously don't live or perform experiments off at infinity somewhere, and we'd like to ask questions about a free falling observer falling into say a horizon. So what are the approximate observables in question, how do they behave and in what sense do they break down?

I'm tempted suggest that if we acknowledge the actual observing context(which people almost never do for some reason), this riddle can be seen at least conceptually in a better light (that might guide us to creating a predictive theory):

No real observer is actually pointlike, meaning that all actual observers are "slightly" de-localised. If we look at the different scales here, obviously if we are talking about interacting galaxies, the degree of delocalisation is quite large, relative to interacting molecules in a gas.

I think this suggests that a realistic observation IS "a little bit" non-local, to the extent that the observer is non-local.

A non-local observer (say a laborariry embracing a collider) can easily infer lorentz symmetry because realtive to a atomic scale collision domain, the observing lab context is completely delocalized. We are here an effective "external observer"

Yes this same lab, can not in the same way embrace space, we are an "internal observer", and relative to say our galaxy VERY localized. The assumetry and relativity in localization here seems very clear.

Now, if we could only find a general description of the inference process, whereby we actually interact with atomic scale systems in lab; and see how this scales/transforms are we turn our observational googles from into matter, to out into space, and how we are forced to revise the abstractions in terms of "ensembles" and statistics, I think we would see more clearly.

It seems clear, that when you see the assymmetry above, we have no rational reason to expect that the mathematical "abstractions" that we KNOW work well for subsystems such as atomic scale stuff in laboratory where the observer is dominating and providing a fixed reference, will hold when the asymmetry turns around and the observer is suddenly localized, instead of the system.

It seems to me that the idea to picture hilbert spaces of the entire universe is pretty much madness. The state spaces must be constrained by the observing context, which in the example is the human lab.

For me the great lack of understanding of this, stems from the fact that we so far have not really analyzed the process where information is inferred, and stored in the observer, and how this information induces an action in the observer. This is because we just mentally picture information encoded in "ensembles" which are infinite information sinks. This is fine as long as the delocalized approxiation holds, but must fail when it doesn't.

I think the Giddings paper above, doesn't emphasise this, but I still think he fairly acknowledges some of these problems as deep and serious, rather than avoiding them.

This does apply directly to the case of diff invariance. EXactly how is diff invariance inferred and stored? These questions are non-existent in classical GR since it's a realist theory. This is why I am convinced that the understanding of the constructing principles of GR, really must be reanalyzed in the light of measurement theory. The old Einstein style arguments just doesn't hold IMO.

I've personally always had the vie that the CORE of GR, is a very basic principle: observer invariance of the laws of physics (then you apply that to special cases; the class of inertial and also non-inertial obesrver to get SR and GR). The principle here is very simple.

But the "observer invariance" of Einsteine here is used in a classical realist sense.

One core point where there seems to be debate is wether this does hold in an inferential theories? The point is that there is no inferential justification of "observer invarnace" beyond the point that the laws of physics would then lack "classical reality". And this latter point not longer holds rational like it did maybe 100 years ago.

/Fredrik

 

[atyy]

Giddings gave a talk at the KITP http://online.itp.ucsb.edu/online/qcdscat11/giddings/" .

I'm very interested in what he has to say but need some time to consider it. I can see that he has two chief propositions. First, "ultraplanckian scattering in the strong-gravity regime manifests a unitarity crisis" (page 14). Second, demonstrating that the AdS/CFT correspondence really is an equivalence faces identifiable technical problems (page 22). His suggestion that maybe it's not an equivalence, and that the boundary theory only approximates the bulk theory, is iconoclastic and I don't believe it, but it's the sort of informed skepticism which can spur others towards the complete and proper solution of a problem. (I had my eye on http://arxiv.org/abs/1102.2910" as an approach to the problem of local observables in the bulk.)

Thanks for the http://arxiv.org/abs/1102.2910" on a similar question, using a somewhat different approach.

Maybe related to what Fra is saying, and reviewed by http://arxiv.org/abs/quant-ph/0312059" . Is the emergent locality in the Kabat et al and Gary et al papers related to Zurek's locality?

 

[Fra]

Second, demonstrating that the AdS/CFT correspondence really is an equivalence faces identifiable technical problems (page 22). His suggestion that maybe it's not an equivalence, and that the boundary theory only approximates the bulk theory, is iconoclastic and I don't believe it, but it's the sort of informed skepticism which can spur others towards the complete and proper solution of a problem.

I like that. I suppose I may need to read more of his arguments. For me which is quite biased towards the inferential perspective the reasoning in that paper is a little backwards.

I mean, the statement is bold sounds good to me, but I see that more as a starting point. The "equivalence" can instead be seen as an "equilibrium" between two interacting theories, that might be inferrable from certain perspectives. This is exactly in line with the arguments I have. This way of understanding holography is IMO rational and understandable. This also merges the best with the old ideas that diffemorphism symmetry (and Einsteins equaoitns) are like an equations of state (ie. an equilibirum), and not eqvuivalences cast in stone.

Except, since the above paper argues towards that from a different angle I'm not sure how to comment on the logic.

I'll try to get around to reading more on those paper.

/Fredrik

 

[atyy]

There's a new paper from http://arxiv.org/abs/1106.3553" .

 

[marcus]

There's a new paper from http://arxiv.org/abs/1106.3553" about obstacles to reconstructing bulk locality,...

Thanks for calling attention to Gary and Giddings http://arxiv.org/pdf/1106.3553
Constraints on a fine-grained AdS/CFT correspondence (Abstract: "For a boundary CFT to give a good approximation to the bulk flat-space S-matrix, a number of conditions need to be satisfied: some of those are investigated here. In particular...")

I was delighted by their quote from Polchinski, which I hadn't seen before:
==G&G footnote page 1==
1 Quoting Polchinski[2], description of strongly coupled gauge theory phenomena is only AdS/CFT’s “hobby, ... its real job is to provide a non-perturbative construction of quantum gravity.”
==endquote==

This reminded me of the recent comment by someone known only as "Wolfgang" in a debate at Not Even Wrong.

== http://www.math.columbia.edu/~woit/wordpress/?p=3793#comment-94039 ==

Wolfgang says:
June 21, 2011 at 7:29 am
“My objection here is one of language, not about the science.”

You do have a point.

People speak of AdS/CFT as a duality between the string theory in the bulk and a field theory on the boundary. Usually, when we speak about a “duality”, we have two independently-defined theories, which we claim are equivalent. Here, as you say, the field theory, on the boundary, has an independent definition, but the bulk theory does not.

At best, it has a set of properties it should satisfy. In the IIB case:

* It should reduce, at low energies, to type IIB supergravity.
* It should have an S-duality symmetry, which acts on the axio-dilaton by fractional linear transformations
* It should have a certain set of BPS extended objects (branes), permuted by the S-duality symmetry.
* etc.

The “right” way to think about AdS/CFT is to take it to be the DEFINITION of the bulk theory. Then these properties become predictions about the field theory on the boundary.

It’s certainly true that it would be nicer if there were independent definitions for both sides of the duality. Then one would get testable predictions in both directions (not just in one direction).

But, in the absence of such an independent definition, it is still true that AdS/CFT provides a nonperturbative definition of string theory in AdS. And it is nontrivial that this definition is consistent with all of the properties that we expect string theory to possess
==endquote==