Friday, September 19, 2014

A pro-BICEP2 paper

Update Sep 22nd: a Planck paper on polarization is out, suggesting dust could explain the BICEP2 signal – or just 1/2 of it – but lacking the resolution to settle anything. A joint Planck-BICEP2 paper should be out in November but it seems predetermined that they only want to impose an upper bound on \(r\) so it won't be too strong or interesting, either.
It's generally expected that the Planck collaboration should present their new results on the CMB polarization data within days, weeks, or a month. Will they be capable of confirming the BICEP2 discovery – or refute it by convincing data?

Ten days ago, Planck published a paper on dust modelling:
Planck intermediate results. XXIX. All-sky dust modelling with Planck, IRAS, and WISE observations
I am not able to decide whether this paper has anything to say about the discovery of the primordial gravitational waves. It could be relevant but note that the paper doesn't discuss the polarization of the radiation at all.

Perhaps more interestingly, Wesley Colley and Richard Gott released their preprint
Genus Topology and Cross-Correlation of BICEP2 and Planck 353 GHz B-Modes: Further Evidence Favoring Gravity Wave Detection
that seems to claim that the data are powerful enough to confirm some influence of the dust yet defend the notion that the primordial gravitational waves have to represent a big part of the BICEP2 observation, too.

What did they do? Well, they took some new publicly available maps by Planck – those at the frequency 353 GHz (wavelength 849 microns). Recall that the claimed BICEP2 discovery appeared at the frequency 150 GHz (wavelength 2 millimeters).

They assume, hopefully for good reasons, that the dust's contribution to the data should be pretty much the same for these two frequencies, up to an overall normalization. Planck sees a lot of radiation at 353 GHz – if all of it were due to dust, the amount of dust would be enough to account for the whole BICEP2 signal.

However, if this were the case, the signals in the BICEP2 patch of the sky at these two frequencies would have to be almost perfectly correlated with each other. Instead, Colley and Gott see the correlation coefficient to be\[

15\% \pm 4\%

\] (does someone understand why the \(\rm\LaTeX\) percent sign has a tilde connecting the upper circle with the slash?) which is "significantly" (four-sigma) different from zero but it is still decidedly smaller than 100 percent. The fact that this correlation is much smaller than 100% implies that most of the BICEP2 signal is uncorrelated to what is classified as dust by the Planck maps or, almost equivalently, that most of the observations at 353 GHz in the BICEP2 region is due to noise, not dust.

When they quantify all this logic, they conclude that about one-half of the BICEP2 signal is due to dust and the remaining one-half has to be due to the primordial gravitational waves; that's why their preferred \(r\), the tensor-to-scalar ratio, drops from BICEP2's most ambitious \(r=0.2\) to \(r=0.11\pm 0.04\), a value very nicely compatible with chaotic inflation. The values "one-half" aren't known quite accurately but with the error margins they seem to work with, they still seem to see that the value \(r=0\) – i.e. non-discovery of the primordial gravitational waves – may be excluded at the 2.5-sigma level.

"Engineers with a diploma" vs "The Big Bang Theory"


  1. Re LaTex. Look at their % sign and compare it to the one I just typed. LaTex has a little connector between the first zero and the slash, which is the correct, traditional way of setting the type.

  2. Interesting, I've never heard that this would be the right way. Why isn't it mentioned e.g. at

    Or is it?

  3. How did a post about dust and gravity waves becomes focussed on Latex percent symbol?

  4. There are a couple of mentions on the talk page, prefixed with "I speculate" and "I imagine".

    "Even tho i speculate that the very well established 20th-century form that looks kind of like O7O has its horizontal bar only bcz it's supposed to imitate a cursive connection between the upper o
    and the slash, i'm not willing that we present an evolution scenario
    that fails to mention, let alone account for, such a striking departure
    from the stages it deems worthy of mention"

    "I've seen quite often fonts that connect the top “o” with the top of
    the bar with a little downwards-curved line, for example Century
    Schoolbook L and cmr10 (Computer Modern, used by TeX, and thus present
    in lots of scientific texts). This should be probably mentioned (I
    imagine it's a leftover of the per cento abreviation)."

  5. I suspect you need very specialized knowledge to say something sensible about the actual paper. Percent signs are a bit more generally known ;)

  6. On the special session of ICHEP-2014, devoted to the BICEP2 results(July 8th, Valencia, Spain) the speaker of the Planck collaboration said, that they want to publish the paper on the dust contribution in the BICEP2's field of view "within a month". Summer passed away, but no paper was published. One quoted in the post seems to be unrelated (I am not the astrophysicist, so I can be totally wrong) since it even do not cite BICEP2 paper, and BICEP2 field of view is not marked on the sky maps presented in the paper.

    Also the Planck spokesperson said, that they are working on the combination of BICEP2 and Planck results.

    Maybe they decided to publish the combined result, since without a proper combination with BICEP2 data, the data on dust could be misleading...

  7. Antonio (AKA "Un físico")Sep 19, 2014, 8:21:00 PM

    Hi nefmax, I also read O'Brient's slides and I agree with you: something is misleading.
    Bicep2 used only one frequency, obtaining an r different from zero with 7 sigmas. But the transfer functions (of dust and noise) calculated from their simulations should have been wrong. And now, with Planck, the r is different from zero with only 2.5 sigmas.
    My opinion is that Bicep2 manipulated their simulations to gain worldwide popularity.
    Well let's wait and see ... but may be Bicep2 will require a convincing apology (in case they want us to still trust in that research team).

  8. 2.5 sigma! This discovery is really dangling on a string.

  9. OK, this is just another, different way to quantify the signal. There are many ways to do so and they give different values. If the primordial gravitational waves ultimately become obviously visible, the original discovery simply has to be treated as a discovery.

  10. Aside from incoherent conspiracy theories and emotions, could you please tell me what the fuck you were talking about?

  11. Sure, I just mean the chances at this point of time, that this is right. Of course it is just one estimate among others but at least it is one estimate.

  12. This seems to be the typical kind of comment often found in anti-physics environments written by people who have no clue what they are talking about, are no experts, but feel entitled to issue strong negative opinions about others or science generally in a very pompous overreaching way anyway ...

    The BiCEP2 team did nothing intentionall wrong, and your attacks and accusations are really creeping up from a subterranian tabloid journal level.

  13. Because that's how you write it with a pencil and paper!!

    Just try it. The resolution is amazing ;-)

    U just have to pick up the pencil before the 2nd o.

  14. Almost as pathetic as your tea bagging "motls" blogspot...too bad your couldn't get your junk science published in a reputable journal and now your little panties are in a bundle as you cry away on your pathetic blog..... HAHAHAHa!!!!

  15. Antonio (AKA "Un físico")Sep 20, 2014, 11:11:00 AM

    Hi Lubos, I have no proofs for my above statement: "Bicep2 manipulated their simulations to gain worldwide popularity".
    But if you send me an email: I will show you all the bits and pieces that made me not trusting in that "r not 0 at 7 sigmas".

  16. Hi conspiracy theorist, I am not going to send you an e-mail because I am not interested in your psychopathological misinterpretations.

    The paper shows the full derivation of the 7-sigma figure, where it comes from etc. Everyone who has read at least the abstract knows what the 7-sigma figure means.

    "The observed B-mode power spectrum is well fit by a lensed-LCDM + tensor theoretical model with tensor-to-scalar ratio r=0.20+0.07−0.05, with r=0 disfavored at 7.0σ. Accounting for the contribution of foreground dust will shift this value downward by an amount which will be better constrained with upcoming data sets."

    It is as clear as sky and the only thing you may "add" is that you may show that you don't know how much 3+4 is, for example.

  17. Antonio (AKA "Un físico")Sep 20, 2014, 2:51:00 PM

    Hi Lubos. I do not understand your aggressiveness. I have read: 1403.3985, O'Brient's slides, and Martínez-González's slides.
    As you have already noticed dust is not well simulated in 1403.3985 and that 7 sigma is a wrong discovery.
    I am visiting your blog, Lubos, as a potential friend; but if you keep being so aggressive with me I will not stay in here for long.

  18. I agree on your estimate that Planck will only set limits. We now have to wait for the common paper to get a consistent picture of what is really what.

  19. " a Planck paper on polarization is out, suggesting dust could explain the BICEP2 signal – or just 1/2 of it – but lacking the resolution to settle anything."

    I find this very hard to take in, since I sided far more with the likelihood of the BICEP2 team having gone through all this to x10 ensure they weren't deluded victims of confirmation bias, bad data etc.

    And... Lubos has this tendency to be generally right most of the time ;)

  20. Hi Anna, I went further. I conjectured that even the joint paper is predetermined by design to just impose upper limit because that's how Planck approaches the problem and they will play the "stronger brother" in the joint venture.

    Of course it's plausible that there is just dust at the end but BICEP2 marginally disagreed with the previous limits so it's a bit of a dangerous mixture to combine experiments that possibly disagree in this way. All papers that combined BICEP2 and Planck ended up with the usual inconclusive, weakened, diluted, and complications-requiring things, and I am afraid that they won't have anything else to offer.

  21. Dear John, there is no "confirmation" here so one can't have "confirmation bias". If true, the detection of the primordial gravitational waves is the discovery of decades and perhaps a century, as some said.

    BICEP2 did a large amount of work to make their discovery result go away, but after an impressive and adequate amount of work, it survived, which is why they published it. It's exactly how things should be.

    If someone seems to be biased here, it's the Planck folks.

  22. Antonio (AKA "Un físico")Sep 22, 2014, 5:40:00 PM

    Despite the rudeness of the host: of Lubos; I would like to comment, in a future post, the paper 1409.5738v1 and the possibility of Bicep2 being a manipulation of science.
    Lubos, am I allowed to comment in your blog?. Are you going to delete my comments?.

  23. In their published paper BICEP2 address the concerns and also say that they are waiting for data from KECK before the end of the year ( I think) so all may not be lost.I think the angular resolution of Planck is much worse than of BICEP2 and that is why they end up giving limits and not talking of r, scalar to tensor with errors They do not have the resolution to do it, I suspect.

  24. Hi Anna, thanks for your comments. I wrote a new blog post about Planck and BICEP2, so let's move there.

    If I could speculate, I would guess that BICEP2 folks already had some better results from Keck, to be somewhat more certain about some things than they admitted.

    But the real reason why I remain a BICEP2 believer is the preferred-scale pattern, see the new blog post.