What experiments compete with BICEP 2, and when are their results expected?

Question

The recent results of the BICEP 2 experiment published on March 17th 2014, has generated a lot of media attention, with the general consensus being that "this is a major discovery" perhaps leading to a Nobel Prize for some.

But for those that are still skeptical, what are the competitors to this experiment, and when are their results expected to be published?

This review article assesses many of the competing experiments: http://www.arxiv.org/abs/1309.5381 — DavePhD, Mar 18 '14 at 01:30

Art Brown · Answer 1 · 2015-01-31T19:36:18.107

4

The Planck experiment looks to be "in tension" (e.g. it conflicts) with BICEP2's results: Planck has previously reported an upper bound $r<0.11$ for the parameter that BICEP2 reports at 0.20 (see for example, http://arxiv.org/abs/1303.5082). Planck is expected to publish new results this ~~summer~~ fall.

21 Sep 2014 update: Planck has now published their galactic dust measurement results.

The situation to date:

BICEP2 assessed (via indirect means) that polarized emissions from galactic dust ("foregrounds") constituted at most a small fraction of the signal they observed at 150 GHZ in the multipole region $l=30-150$, so that all (or at least most) of the signal was cosmological in origin and therefore Nobel-worthy evidence for inflation. (The larger the foreground, the lower the value of the tensor-to-scalar ratio $r$; BICEP2's reported $r=0.20$ corresponds to no foreground.)
However, Planck's new results put the level of those foreground signals substantially higher than BICEP2's estimates. (Planck's result is not a direct measurement at BICEP2's 150GHz frequency, where Planck cannot match BICEP2's sensitivity, but a calibrated extrapolation from 353 GHz data, where the dust signal is much stronger.) In fact, it is possible that Planck's foreground accounts for all of BICEP2's observations, which would mean that BICEP2 detected no cosmological signal at all.
Currently, Planck and BICEP2 are jointly analyzing their combined data sets to determine how much, if any, of BICEP2's signal can be attributed to cosmological inflation. The results of their joint analysis are expected by year end 2014.

31 Jan 2015 update:

The joint Planck-BICEP2 analysis is now complete, with the paper available here. The result: BICEP's claim of detecting primordial B-mode polarization at a level of $r = 0.16-0.20$ is withdrawn, and replaced by a 95%-confidence upper bound $r<0.12$, which is almost exactly the same as Planck's upper bound $r<0.11$ before BICEP2 reported last year.

edited Jan 31 '15 at 19:36

answered Mar 18 '14 at 00:44

Art Brown

5,933

1

BICEP2's response to the apparent conflict is "Our measurements don’t disagree. Constraints on the gravitational-wave background level r reported from Planck and previous experiments are not from measurements of B-mode polarization. Instead, they come from the CMB temperature measurements which show surprisingly low power at the largest scales, implying little room for an additional contribution from tensors in the context of the simplest models. (continued below) – DavePhD Mar 18 '14 at 14:41
(continued from above) B-mode measurements like ours aim to directly measure the inflationary gravitational-wave pattern itself at the degree angular scales where it should peak. The tension between the high level of B-mode polarization we see and the apparent low power at large scales may be a statistical fluke, but many possible extensions to the simplest model could also relieve this apparent tension." source: http://bicepkeck.org/faq.html – DavePhD Mar 18 '14 at 14:42
@DavePhD: Interesting. This portion of the conclusion of the BICEP2 "Detection" paper acknowledges the "tension": – Art Brown Mar 18 '14 at 16:33
"Subtracting the various dust models and re-deriving the r constraint still results in high significance of detection. For the model which is perhaps the most likely to be close to reality (DDM2 cross) the maximum likelihood value shifts to r = 0.16 +0.06 -0.05 with r = 0 disfavored at 5.9σ. – Art Brown Mar 18 '14 at 16:34
These high values of r are in apparent tension with previous indirect limits based on temperature measurements and we have discussed some possible resolutions including modifications of the initial scalar perturbation spectrum such as running. However we emphasize that we do not claim to know what the resolution is." – Art Brown Mar 18 '14 at 16:35
This "most likely" value of r is barely consistent with Planck's upper bound. (Why didn't they report it as the headline number?) – Art Brown Mar 18 '14 at 16:36
BICEP and Planck measured two different things; polarisation and amplitude, respectively. There can be no tension between such measurements. Particular models might not be able to reproduce both measurements, but that is a different matter. – innisfree Mar 18 '14 at 16:57
@innisfree: Both experiments reported their results in terms of the same parameter (r), and those results disagree. – Art Brown Mar 18 '14 at 17:26
But that is just a parameter/artefact of lambdaCDM model. If I come up with a model that predicts electron mass equals higgs mass, would you say those measurements were in conflict? One can always make up such silly models for any 2 measurements. Not that lambdaCDM is silly. – innisfree Mar 18 '14 at 17:30
@innisfree: I don't get it: your model would be disproved by the measurements. Are you saying that Planck and/or BICEP2 are misinterpreting their data? That makes sense, but the fact remains that they are now claiming results for r that don't appear to be consistent. That's the tension that needs to be resolved. – Art Brown Mar 18 '14 at 17:36
BICEP show that a small modification to the lambdaCDM, a running spectral index, can reproduce Planck and BICEP2 results. It's the final figure in the paper. It is not an experimental tension. It's a matter of finding a model that can reproduce all measurements. – innisfree Mar 18 '14 at 17:40
@innisfree: The quote above, from the BICEP paper, shows that BICEP acknowledges tension with other experiments and doesn't currently have a resolution. – Art Brown Mar 18 '14 at 17:47
@ArtBrown BICEP2 says r = 0.16 +0.06 -0.05, and the uncertainties are only a 68% confidence interval, it seems there is at 16% chance that r<0.11 as Plank said. Plus BICEP2 gives a specific line of reasoning that Plank could be consistent with r<0.26. – DavePhD Mar 18 '14 at 18:30
This comment thread shows that most people on this board are theoretically inclined. After correcting for dust contributions to the B mode the r of BICEP2 is given as ).16 +0.06 -> -.05. The difference between the limit of Planck of 0.1 is less than two sigma. It is ridiculous to talk of discrepancy. There r=0 is off from the BICEP2 results by 5.9 sigma. People should read up on statistical significance in physics. – anna v May 22 '14 at 03:54
@annav: As I excerpted above in comments 3-5, the BICEP2 paper itself acknowledges the tension. (After all, it claims a signal at $5.9 \sigma$ confidence, while Planck sees nothing.) By the way, BICEP2's handling of the dust foreground is under intense pressure; see here: http://pcts.princeton.edu/PCTS/SpecialEventSimplicity2014/SpecialEventSimplicity2014.html – Art Brown May 22 '14 at 04:53
What you d not realize is that giving limits is a way of displaying very large errors. Remember the firs Brazil plots when looking for the Higgs? When one reaches the point of quoting sigmas over five then the signal is indisputable, the interpretation can be disputed only, i.e. if most of it is dust and not the small bit attributed by BICEP2. I think they have been extremely meticulous in chasing after instrument systematics. Now Planck does not see a signal means that Planck may not have as good an experimental aparatus, considering it must have been transported to outer space. – anna v May 22 '14 at 05:40
@annav: BICEP2 claimed $5.9 \sigma$ for a cosmological signal, i.e. after subtracting out the dust. That number is under pressure. All reviewers I have read agree with you that BICEP2 has seen a B-mode signal "in the sky". The question is whether they understand the foreground contributions. (In fact, they have no error bars on their foregrounds!) Again, BICEP2 themselves acknowledge the tension. – Art Brown May 22 '14 at 05:50
5.9 sigma from 0, not from r=.1 which is the limit of Planck. I have the paper before me because I am preparing a general scientific audience lecture on this. All I am saying is that their number agrees statistically with the Planck limit. – anna v May 22 '14 at 06:03
Look at the higgs brazil plot, before the establishement of Higgs by invariant mass plots. http://physics.stackexchange.com/questions/13170/particle-physics-plots . The higgs is lost in the two sigma band at 125 – anna v May 22 '14 at 06:11
@annav: I agree that BICEP2's result including foreground model DDM2 is (just) consistent with Planck's upper bound, although BICEP2 still considers it to be in tension with Planck (section 12). The recent criticism of the BICEP2's result is that the DDM2 foreground model's B-mode levels are too low, so the r-range should move lower (improving agreement with Planck), and the 5.9 confidence should drop. (See the link in my previous comment.) In fact, BICEP2 is now waiting for a measured dust map from Planck (due this fall) to replace the DDM2 model. The message is "patience". – Art Brown May 22 '14 at 07:23
Thanks. There was general interest in the subject and as I was the closest one to cosmology I was asked if I could explain to an audience of solid state physicsists and chemist and biologists. On th subject we do not disagree after all. I think that we in the elementary particle field rely more on "5 sigma a signal" 2 sigma can be ignored. I have seen in our data 4 sigma disappear with a newer analysis on the same data, but not 5. Of course the use of dust background is crucial to disentangling the sources. – anna v May 22 '14 at 08:36
@annav: You have now seen a "5 signal signal" go away... – Art Brown Jan 31 '15 at 19:39
@ArtBrown The polarization signal has not gone away in the BICEP2 data. The interpretation/split has become uncertain, how much is due to dust and how much to gravity. It is evident that a new experiment to define dust in the region of BICEP2is needed. Planck floods the combined data but it has limited accuracy. – anna v Jan 31 '15 at 19:54
@annav: No. Planck measured dust by looking at multiple frequencies, which BICEP2 couldn't do, and the two organizations have now agreed that the signal BICEP2 observed is just dust. – Art Brown Jan 31 '15 at 19:54
they have agreed to limits using the combined data, but the data from Planck are such that the combination can only give limits. So we have a very good determination oftensor to scalar polarization at the one degree from BICEP2, and a blanket from Planck that covers everything with limits. Planck does it with fitting the CMB in the whole region, Bicep by physically measuring tensor to scalar in small increments. Unless an equally good to BICEP2 accuracies dust measurement can be obtained , it is an impasse. – anna v Jan 31 '15 at 20:12