Wednesday, August 29, 2018 ... Deutsch/Español/Related posts from blogosphere

Team Stanford launches Operation Barbarossa against quintessence

The disagreement between Team Stanford – which defends its paradigm with a large landscape of de Sitter solutions of string theory – and Team Vafa – which suggests that de Sitter spaces could be banned due to general stringy "swampland" principles (and which proposes quintessence as an alternative) – has been seemingly confined to short enough exchanges in the questions-and-answers periods of various talks.

The arguments couldn't have been properly analyzed and compared in such a limited context. In science, it is better to write them down. You may look at these arguments and equations for hours – and so can your antagonists – which usually increases the quality of the analyses. Team Stanford clearly believes that the de Sitter vacua are here to stay, the criticisms are wrong, and quintessence has fatal problems. But can they back these opinions by convincing arguments?



Today, in the list of new hep-th preprints, we received an avalanche of papers that say something about the deSitter-vs-quintessence controversy in string theory. Using the [numbers] from the daily ordering of papers, we talk about the following papers:

[3] De Sitter vs Quintessence in String Theory (by Cicoli+4, 49 pages)

[4] A comment on effective field theories of flux vacua (by Kachru+Trivedi, 22 pages)

[15] dS Supergravity from 10d (by Kallosh+Wrase, 18 pages)

[16] de Sitter Vacua with a Nilpotent Superfield (by Kallosh+3, 6 pages)

[18] The landscape, the swampland and the era of precision cosmology (by Akrami+3, 43 pages)
I have omitted Tadashi Takayanagi's paper(s) although one of them also talks about de Sitter spaces.

First, concerning the affiliations: I include all of the collaborations into "Team Stanford" because they defend de Sitter solutions in string theory. But the first paper is really international (Bologna-Boulder-India-Cambridge), the second paper is Stanford-Bombay, the third paper is Stanford-Vienna, the fourth paper is Stanford-Brown-Leuven (Belgium), and the last paper is Stanford-Leiden (the Netherlands).



Well, you may hopefully see that Stanford is overrepresented in these papers. Moreover, it seems to play the role of the "headquarters" of this campaign. And the first paper among the five which is the only Stanford-free is arguably the least combative one, too. ;-) I think it's fair to say that the stringy landscape picture of cosmology is the greatest source of pride for Stanford's theoretical physicists in recent 15 years. At some human level, we could understand why they could be anxious if someone were basically saying that those 15 years revolved around a mistake or some sloppiness. But the pride doesn't imply that those papers were right and safe, of course.

Now, the number of papers – five – is rather large and the salvos had to be at least partially coordinated. Can the colleagues be expected to swallow a reasonably high percentage of the content? Wasn't the number of papers chosen to be high to simply intimidate the opposition? To replace the quality of the arguments with the quantity of papers? I am not saying that. I am just asking. The high number of papers leads me to similar feelings as the proposed large number of de Sitter vacua. Less is sometimes more.

Let's talk about the separate papers. The middle paper, one by Kallosh and Wrase, claims that the anti-D3-branes in the KKLT "uplifting" procedure may be replaced by anti D5, D6, D7, or D9-branes, too. That seems like a bold statement to me. If this were the case, why wouldn't have KKLT noticed these four new possible dimensions right away? Fifteen years ago, I was surely asking the question why anti-D3-branes were used and not some branes of other dimensions and I was surely given a – not so convincing – answer implying that it had to be anti-D3-branes. If one says that 4 possible dimensionalities of the antibranes are just as OK, and one does so 15 years after the game-changing paper is released, it doesn't exactly help both of these papers to be trustworthy.

I would probably choose to disbelieve the new Kallosh-Wrase paper. One general problem with this paper (but, to some extent, with many other papers and perhaps with Team Stanford's papers in general) is that it seems to be a supergravity paper, not a full-blown stringy paper. And I think it's fair to describe both Kallosh and Wrase as supergravity experts, not string theory experts. Shouldn't a full-blown string theory expert validate claims that D-branes may be used in a certain new way? My answer is that he or she should.

At their supergravity level of analysis, many things are possible and they may change the dimensionality of the uplifting antibrane. Great. But have they actually demonstrated that string theory allows such solutions, especially the new ones? I don't think that they have made the full-blown string analysis. Whatever is intrinsically stringy is treated in a sloppy way. For example, search for an "open string" in the Kallosh-Wrase paper. You will get three hits – and all of them just say that they have ignored the open string moduli.

The more stringy a given concept or structure is, the more it is ignored in this paper. Again, I think that this criticism applies to most of the Team Stanford papers in general. But the whole point of the Vafa Team is to carefully study the fine, characteristically stringy features, phenomena, and constraints that are completely invisible at the level of supergravity – i.e. at the level of effective field theory. I have doubts about every particular, precise enough "swampland statement" made by Vafa or any disciple (including our "weak gravity conjecture" group). On the other hand, I have no doubts that it is extremely important to appreciate that string theory is not just supergravity and most of the particular low-energy supergravity-based effective field theories have no consistent quantum gravity or stringy completion.

Kallosh and Wrase – and, as I said, much of the Team Stanford – seem to use string theory as the "ultimate justification of the 'anything goes' paradigm in supergravity". You may do anything you want in supergravity, add any string-inspired object, fluxes, branes, whatever you like, and then you use the term "string theory" as if it were the ultimate and universal justification of the validity of all such constructions. For them, string theory is just a knife that always unties your hands. Like with Elon Musk's promises, anything goes with string theory.

OK, I am sure that this is just a wrong usage or interpretation of "string theory". String theory offers some new tools, new objects, new transitions, phenomena, and relationships between the objects. But string theory also – and maybe primarily – brings us new constraints, new bans, new universal, and particular predictions. For me, string theory may have produced new ingredients and possibilities but it's still primarily a theory that has a greater predictive power than the effective quantum field theory. It's clearly a sloppy, skewed way to use string theory if someone only uses string theory as the "source of many new objects and possibilities" – and not as a "book full of new constraints, universal laws and principles, and previously impossible predictions for particular situations".

(There has been a community of "extremely applied" string theorists – whom I would surely call non-string theorists – who have used the term "string theory" as an excuse for really non-standard pieces of physics including the Lorentz symmetry violation and the violation of the equivalence principle. I believe that string theory is, on the contrary, a solid framework that bans or at least greatly discourages such experiments.)

Because we are discussing the question whether the carefully and accurately studied string/M-theory allows de Sitter vacua, the KKLT construction, and similar things, another supergravity-level sloppy analysis just cannot possibly be relevant for the big question defining the Team Stanford vs Team Vafa controversy. To resolve this controversy, one simply needs a higher stringy precision of the arguments. The paper by Kallosh and Wrase doesn't have it and it's questionable whether they could make such an analysis in any other paper.

OK, let's now look at the fourth paper among the five about a "nilpotent superfield". The new paper is a response to a 2017 paper Towards de Sitter from 10D by Moritz, Retolaza, and Westphal. OK, those authors have claimed that the KKLT didn't work because during the uplift, there was a stronger backreaction than previously thought and the compactification remains AdS and doesn't become dS. In the new paper, they claim that the nilpotent superfield as a SUSY breaking tool isn't compatible with the nonlinearly realized SUSY. But that doesn't really matter because even if one allows it, they do get a de Sitter, not anti de Sitter.

I would believe that one of these groups must admit defeat soon enough because the claims and arguments look rather straightforward.

Now, let's turn our attention to the new Kachru-Trivedi paper. It's written as a "positive paper" on effective field theories of the KKLT-style flux vacua. I haven't read the paper in its entirety but the abstract and the general organization of the paper does suggest that they're reviewing the thoughts that have been around from the KKLT. It seems to me that concerning the validity and existence of the effective field theories for the stringy situations, they always rely on field-theory-based, e.g. Wilsonian arguments. I am not persuaded that this is good enough. String theory may invalidate the effective field theories by making sure that an energy-\(E\) effective theory isn't a local quantum field theory at all.

What really bothers me is the superficial approach of Kachru and Trivedi to the arguments given by the Vafa Team:
A recent paper [46 Obied Ooguri Spodyneiko Vafa], motivated largely by no-go theorems with limited applicability to a partial set of classical ingredients, made a provocative conjecture implying that quantum gravity does not support de Sitter solutions. [Footnote about two previous papers saying similar things.] Our analysis – and more importantly, effective field theory applied to the full set of ingredients available in string theory – is in stark conflict with this conjecture. This leads us to believe that the conjecture is false.
Do Kachru and Trivedi consider this non-technical, judgmental paragraph to be enough to deal with the proposed alternative picture? OK, let's rephrase what they are saying:
We may repeat what we said 15 years ago. We may pay no attention whatsoever to the detailed arguments given by the Vafa Team. We don't need to be impartially interested in the validity of the proposed new principles, inequalities, and no-go theorems. We just don't want to learn any and we prefer to believe that no such new insights exist. Instead, it's enough to dismiss all these papers with a simple slogan, with slurs such as "provocative" that make the Vafa Team look limited while we look unlimited, repeat that everything we have ever claimed to be true must be true, and that's enough to "prove" that we are right and they are wrong.
I am sorry but it doesn't seem enough to me. The claim that the Vafa Team's statements are limited to a "partial set of classical ingredients" while Team Stanford is better because effective field theory is "applied to all ingredients available in string theory" seems utterly demagogic to me. KKLT and followers have used lots of ingredients from string theory but there's no proof that they are "all" ingredients of string theory. New ingredients kept on emerging and we still can't prove that we know "all of them" because we don't have a universal definition of string theory. Moreover, the high number of such ingredients makes it more likely, and not less likely, that one of them breaks down and invalidates the KKLT construction as a whole. So they have many, not "all", ingredients of string theory and this fact makes their construction more vulnerable, not less so!

And the very conclusion that "Vafa seems to disagree with something we wrote so he must be wrong" simply looks childish. This is not a rational way to argue. Vafa might say exactly the same thing – he doesn't – but none of these two stubborn propositions would imply a convincing argument in one way or another.

Finally, we have the first paper by Cicoli et al.; and the last, fifth paper by Akrami et al. These two papers explicitly claim to discuss the "de Sitter versus quintessence" controversy in string theory. The Akrami paper seems to have one proposed counterargument against quintessence that Akrami et al. are proud about and that they want to be carefully read by the reader. What is it? They pick the constant \(c\) from the Vafa Team inequality, claim that it should be equal to one (or being of order one, they're not sure), and then they claim that the cosmological observations rule out \(c\gt 1\) at the 3-sigma i.e. 99.7% level.

I am sorry but the right value of \(c\) isn't really known, at least not too reliably, so they can't determine the statistical significance well, either. The right \(c\) could be \(1/3\) and there would be no exclusion at all. It seems to me that this overemphasis on the \(c\sim 1\) "prediction" and its weak exclusion by the observational data is their strongest argument. If that's so, I find it extremely weak. Even if the calculation of the 3-sigma confidence level were solid, which it doesn't seem to be at all, it is still just a 3-sigma confidence. A few years ago, the LHC diphoton bump was "discovered" at four sigma and it was fake. A potential universal new principle of string theory is a different caliber. In my list of priorities, if I become sufficiently certain about a new universal principle of physics, it may beat even 5-sigma deviations from the predictions.

Finally, the first, Stanford-free paper is less arrogant than the Stanfordful papers. They prefer the de Sitter, KKLT-style models because they look concrete, there seems to be a calculational control, and it's apparently getting better with time. Quintessence is more "challenging" and requires more fine-tuning, we read. Well, Vafa et al. disagree with the second point, probably both points. At any rate, they're potentially subjective. You can't use your feelings that something is "challenging" – without any particular argument or quantification of the "challenge" – as a persuasive argument against an alternative theory.

So I am afraid that this Cicoli et al. paper is going to be too vague when arguing against the alternative picture based on the new general principles proposed by the Vafa Team. One problem, as I have mentioned, is that these two paradigms are very different from each other. They have completely different advantages, very different numbers of requires solutions or corners of the stringy configuration space, different importance of the precision needed to analyze things, and so on. Depending on one's philosophy, the prior probabilities assigned to these paradigms may be very different. The probability ratio may very well be more extreme than 300-to-1 in either direction – which makes some 3-sigma empirical arguments weaker than a weak tea.

At the end, it should be possible to resolve the controversy. But one simply needs to study the purely stringy effects in these compactifications (or would-be vacua) more accurately or more reliably than ever before. This increased control over the stringy effects or the increased reliability of the stringy arguments is probably necessary for any progress in resolving this open question. I haven't read the papers in their entirety but I am afraid it is obvious that they haven't really made any progress in resolving the actual disagreement. They are basically repeating the things that were done before and that's not a good path to progress.

And that's the memo.

Add to del.icio.us Digg this Add to reddit

snail feedback (0) :

(function(i,s,o,g,r,a,m){i['GoogleAnalyticsObject']=r;i[r]=i[r]||function(){ (i[r].q=i[r].q||[]).push(arguments)},i[r].l=1*new Date();a=s.createElement(o), m=s.getElementsByTagName(o)[0];a.async=1;a.src=g;m.parentNode.insertBefore(a,m) })(window,document,'script','//www.google-analytics.com/analytics.js','ga'); ga('create', 'UA-1828728-1', 'auto'); ga('send', 'pageview');