Planck Data Out

Long awaited data from the Planck satellite was released today, papers available here. The accompanying press release leads with results about the timing of the first stars, 500 million years or so after the big bang, with little mention of the very early universe. This is also the main topic of BBC News coverage.

This paper reports a bound on r of .08-.09, exactly what Shaun Hotchkiss was predicting earlier this week here. This appears to be pretty much the end of the line for hopes that Planck would see primordial gravitational waves, with the paper seemingly pointing to other experiments being necessary to get below r=.05 (see page 35).

The BBC News story also characterizes these bounds as ruling out the simplest inflationary models, requiring they be supplemented by “exotic physics”.

What is clear from the Planck investigation is that the simplest models for how that super-rapid expansion worked are probably no longer tenable, suggesting some exotic physics will eventually be needed to explain it.

“We’re now being pushed into a parameter space we didn’t expect to be in,” said collaboration scientist Dr Andrew Jaffe from Imperial College, UK. “That’s OK. We like interesting physics; that’s why we’re physicists, so there’s no problem with that. It’s just we had this naïve expectation that the simplest answer would be right, and sometimes it just isn’t.”

For about as long as I can remember, string theorists and multiverse fans have been pointing to Planck data as the test of their ideas. For cosmic strings, the last Planck data release had a paper ruling them out. I don’t see a paper on this topic out or projected for the new data, it seems that this is now something not even worth looking for.

We’ve also been hearing for years that Planck will test supposed evidence of bubble collisions indicating other universes, see for instance this article about this paper, where the article states that

Data from the Planck telescope should resolve the question once and for all.

I don’t see anything in the new data even looking for this. Has it already been ruled out, without any publicity, or did the Planck people think it was something not worth even looking for?

This entry was posted in Uncategorized. Bookmark the permalink.

18 Responses to Planck Data Out

  1. Ben Gold says:

    The bubble-collision paper you reference is looking for the signature in temperature only. I think I’ve seen some calculation of polarization signatures for specific multiverse models, but I’m unaware of a general “bubble collision”-type search that uses polarization. So I think the answer to your question is “mostly ruled out without publicity”, but also that even defining a generic polarization signature for such models is hard, so there could be room for something clever.

  2. Peter Woit says:

    Thanks Ben,
    What about this one
    which claims a polarization signal?
    For the Planck data, is this in the category of already ruled out, Planck thinks not worth looking for, or results buried somewhere I didn’t see?

  3. Ben Gold says:

    OK, the punch line of that paper (which is looking basically at the E-mode generated by the temperature gradient) seems to be that small spots would be detected first in temperature, and “very large spots [angular radius > 25º] could be detected almost as easily in polarization as temperature”.

    There are some particulars of the signature that might make it stand out a bit more, but basically the upshot is that you’d be looking for large-scale E modes (and TE correlations), where there’s plenty of ordinary signal (unlike B modes) and where Planck has been having trouble this whole time.

    Side note – while it looks like Planck has released a lot of data, it notably does not include polarization power spectra, or in fact any polarized maps except for the LFI frequencies (where foregrounds are larger) and, for some reason, at 353 GHz. Not the 100/143 that I’d think to use for CMB analysis.

  4. Peter Woit says:

    Thanks Ben,
    OK, so for that one I guess it’s still wait and see…

  5. Jeff says:

    I just watched this on BBC and it was bizarre. After three minutes of talking about the earliest stars they cut to an expert who completely changed the subject (or they cut the segue) and babbled “Multiverse, multiverse, multiverse.”

  6. Bernhard says:

    This whole story resembles SUSY so much…

  7. David says:

    How does this reconcile with the earlier BICEP/Planck joint paper?

  8. Peter Woit says:

    They get a somewhat stronger bound on r from their data. As Ben Gold explains, they still haven’t finished the full analysis of their polarization data, presumably will do even better when that is done.

  9. It deserves to be highlighted more explicitly: the model of inflation strongly preferred by the Planck2015 data is — as was already the case after the Planck2013 analysis — the Starobinsky model.

  10. Anonymous says:

    Urs, there is no such preference. The data is not sufficient make such strong distinctions in any case, but with the new r distribution it is not true that the Starobinski model is preferred. More will be known in the coming years, no need to jump the gun.

  11. Peter Woit says:

    You really can’t claim strong support from experiment for your model when your model’s prediction is that experiment won’t see anything. All sorts of early universe models don’t have observable primordial gravitational waves.

    I just looked out the window, saw no large monster green turtles. This provides strong support for my theory that the universe is controlled by a large monster green turtle, one that lives on the other side of the planet….

  12. The Starobinsky model sits right in the center of the parameter range preferred by the Planck data, the famous figure 1 in arXiv:1303.5082, which is now confimed as figure 22 of arXiv:1502.01589.

    Quoting Jörg Paul Rachen, member of the Planck collaboration, from his IMAPP talk just yesterday:

    The Starobinsky model is the model with the highest Bayesian evidence as it is right in the center of the likelihood peak and at the same time has the lowest number of free parameters.

  13. Anonymous says:

    The central value of r is nonzero (.03-05), but with (in)famously low significance. There is no statistically significant distinction among the many models in the allowed range, including Starobinski, but the latter is not at the central value. There’s more data on the way, so anyway there’s no need to guess right now.

  14. I am not guessing, I am quoting public statements made by experts with an identifiable name attached to them.

  15. Eduardo Lira says:

    Be it that SUSY doesn’t show up at the LHC or that Planck sees only dust, it all leads over and over again to the ultimate objection critics make to ST: that it can’t be falsified.
    However, doesn’t the history of physics record other circumstances where falsifiability was also absent and sometimes so for many years, but only to turn out different in the longer term? Consider for example the question about the chemical composition of stars that scientists asked themselves in the first half of the 19th century. You couldn’t then (nor can today!) bridge the immense distances to the stars, so it seemed impossible to perform analysis of any kind to find out what they are made of. Among others, French philosopher Auguste Comte properly (or so it seemed then) presented the case as an example of a class of facts that were to stay forever hidden from human knowledge.
    Any theory aimed at describing what stars are composed of that had been proposed those days wouldn’t have been falsifiable, and should then have been classified as “not scientific”. But of course later discovery of the absorption spectra phenomenon ultimately led astronomers and physicists to determine the precise composition of stars, never mind if you can’t go measure it in situ.
    Back to ST, can we discard the possibility of coming someday across the likes of spectra absorption lines that allow us to reveal the precise shape of Planck-scale space-time, and the inner structure (if there is any) of elementary particles? If you ask people like George Ellis and Joe Silk -to mention a recent reference in this blog- the answer would be that we’ll never find out. In their own words, “the higher dimensions are wound so tightly that they are too small to observe at energies accessible through collisions in any practicable future particle detector” (Nature, Dec. 16, 2014).
    But what if you replace the idea of “too small tightly wound higher dimensions, too small to observe” with “too large distances to the stars, too far to analyze”, and also the one of “any practicable future particle detector” with “any practicable future means to reach a star”?
    Could a future discovery nobody anticipates now take ST out of the not-even-wrong limbo, as it would have been the case for any “not scientific” (i.e. unfalsifiable) theory which happened to be proposed 150 years ago to try to predict the composition of stars, and which became falsifiable against all odds when spectra absorption was discovered and explained? As improbable as it may look today (and as it seemed in 1840 that we would stumble upon a means of finding out what stars are made of), if such a breakthrough comes to occur (whether this happens in the near future or 200 years from now is fundamentally irrelevant), confrontation of ST with the newly acquired way to probe reality will allow physicists to determine whether it is plainly wrong or just right, the latter if only in the broad sense of pointing in the right general direction for further construction and refinement.
    A much less important consequence of this last hypothetical outcome is of course that the category where Auguste Comte belongs in history (whatever it may be) would be joined by numerous new members.

  16. Peter Woit says:

    Eduardo Lira,
    Please, this is off-topic, has nothing to do with yesterday’s Planck results. I will leave it though, just so I can point to a relevant FAQ entry (the problem with string theory is not that its predictions are unobservable, but that there aren’t any)

    I will however delete further off-topic comments that have nothing new and interesting to add.

  17. West says:

    @Urs: Did you happen to catch how much better the marginal likelihood (hate using “evidence” for this) is for the Starobinsky model than its competitors?

  18. West says:

    Am actually able to answer my own question now that the Planck collaboration’s paper about constraints on inflationary models is out.

    Table 6 (p.18) is a nice summary of the model comparisons using only Planck data and as Urs suggested yesterday, the Starobinsky model comes out on top. Mind you not by much, and if one includes the BKP cross-correlation results, some other models do just as well at fitting the data. The results lend credence to models with small r values but no single one has truly distinguished itself from the pack.

    So in the end, we have a 95% upper limit on r at 0.08, no evidence for spectral running and an uncertainty interval (68%) for n_s at the sub-1% of the peak posterior value. Impressive work all around.

Comments are closed.