[For your reference if you can’t follow this post: My History of the Universe, and a primer to help you understand what’s going on today.]
I’m still updating this post as more information comes in and as I understand more of what’s in the BICEP2 paper and data. Talking to and listening to experts, I’d describe the mood as cautiously optimistic; some people are worried about certain weird features of the data, while others seem less concerned about them… typical when a new discovery is claimed. I’m disturbed that the media is declaring victory before the scientific community is ready to. That didn’t happen with the Higgs discovery, where the media was, wisely, far more patient.
The Main Data
Here’s BICEP2’s data! The black dots at the bottom of this figure, showing evidence of B-mode polarization both at small scales (“Multipole” >> 100, where it is due to gravitational lensing of E-mode polarization) and at large scales (“Multipole” << 100, where it is potentially due to gravitational waves from a period of cosmic inflation preceding the Hot Big Bang.) All the other dots on the figure are from other experiments, including the original BICEP, which only put upper bounds on how big the B-mode polarization could be. So all the rest of the points are previous non-detections.
Note: for some reason, they do not show the detection of B-modes at small scales, due to lensing, by the South Pole Telescope (SPT) and POLARBEAR.
NOTE: DESPITE WHAT MANY IN THE MEDIA ARE SAYING, THIS IS NOT THE FIRST INDIRECT DISCOVERY OF GRAVITATIONAL WAVES (AND THEREFORE A TRIUMPH OF EINSTEIN’S THEORY OF RELATIVITY.) (The first indirect discovery of gravitational waves was decades ago and won the 1993 Nobel Prize. [Some are arguing that this detection is more direct; ok… I agree, it is. Not as direct as LIGO would be though.]) IT WOULD POTENTIALLY REPRESENT A TRIUMPH FOR THE THEORY OF INFLATION, WHICH USES EINSTEIN’S THEORY, BUT REALLY IS A SUCCESS FOR 1970s-80s PHYSICISTS — PEOPLE LIKE STAROBINSKY, GUTH, LINDE, STEINHARDT… NOT EINSTEIN.
The claim that BICEP2 makes is that their measurement is 5.2 standard deviations (or “sigma”s) inconsistent with zero B-mode polarization on the large scales (small Multipoles). That’s normally enough to be considered a discovery, but there are some details that need to be understood to be sure that there are no subtleties with that number. Note that this is not a 5.2 sigma detection of inflationary gravitational waves! For that, they need enough data to show their observed data agrees in detail with the predictions of inflation. The 5.2 sigmas refers to the level of the detection of B-mode polarization that is not merely due to lensing.
They can only disfavor the possibility that their measurement is caused by dust or by synchroton radiation at the 2.3 sigma level, however. This may be something to watch.
A Point of Concern
One thing you can worry about is that the points at large multipoles are systematically higher than expected from lensing. Why is that? Could it suggest an effect that is being neglected that could also affect small multipoles where they’re making their big claim of discovery? The more I look at this, the more it bothers me; see the figure below.
The effects of “gravitational waves” (dashed lines) should be very small around Multipole of 200, but in fact (comparing the solid lensing prediction with the black dots data) they seem to be as large as they are around Multipole of 80. One might argue that this actually disfavors, at least somewhat, the interpretation in terms of gravitational waves. However, this may be too hasty as there may be other aspects of the data, not shown on this plot, that support the standard interpretation. I’ll be looking into this in coming days. [And I’ve just notice that David Spergel is also concerned about this — he also points out this anomaly shows in a poor fit in Figure 9 of the paper, and that there are also problems, at *low* multipoles, in Figure 7. Definitely things to worry about here…]
[[However, this point was addressed by the BICEP2 folks in their presentation. Their view is that (1) the high data points are not very statistically significantly high, and (2) with new data that they haven’t released from their third-generation experiment, they don’t see the same effect. So this is presumably what gives them confidence that the excess is a temporary, statistical fluke that will go away when they have more data.]]
How It Compares with Planck Data
Just to clarify what the orange regions are, and emphasize a point: in the figure below is Planck’s data (nothing about BICEP there) [thanks to Oliver DeWolfe for digging this up.] If you compare the blue region of the figure below — Planck data interpreted in inflationary models in which n_s is a constant as the universe inflates — with BICEP2 data, which prefers r around 0.15-0.3, you would conclude that inflationary models with n_s = constant are disfavored. But models where n_s varies a little bit, which fill the orange region, are much more consistent with BICEP2. Conclusion: if you take Planck and BICEP2 at face value, n_s is probably not a constant — which might mean yet another discovery!
But it’s a little early, still, to be sure about that. For one thing, the true value of r is likely to be lower than what BICEP2 says right now — because of a well-known statistical bias. Discoveries tend to be on the high side, just for statistical reasons: if an experiment has a statistical fluke on the low side, they will discover an effect later, when early discoveries tend to involve a statistical fluke on the high side. So the value of r might well be 0.1 – 0.15, despite what BICEP2 says now.
Be More Cautious than the Media
As always, I have to caution you that although I’m fairly impressed, and reasonably optimistic about this measurement, it is a measurement by only one experiment. Until this measurement/discovery is confirmed by another experiment, you should consider it provisional. Although this is too large a signal to be likely to be due to a pure statistical fluke, it could still be due to a mistake or problem, or due to something other than gravitational waves from inflation. The history of science is littered with examples; remember the 2011 measurement by OPERA that showed neutrinos moving faster than the speed of light was far too large to be a statistical fluke. Fortunately there will be other experiments coming and so we’ll have a chance for various experiments to either agree or disagree with each other in the very near future.
What It Means if it’s True
If this measurement is correct, and if indeed it reflects gravitational waves from inflation in the most conventional way, then it would tell us that inflation occurred with a dark energy per unit volume (i.e. dark energy density) that is comparable to the energy scales associated for decades with the energy and distance scale at which all the known non-gravitational forces would naively have about the same strength — the so-called “unification of coupling constants”, sometimes extended to “grand unification” in which the various forces actually turn out to be manifestations of just a single force. This would be very remarkable, though not necessarily evidence for unification. There are other ways to get the same scale, which is about 100 times lower in energy (100,000,000 times lower in energy per unit volume) than the scale of quantum gravity (the Planck scale, which, roughly, tells you the energy density required to make the smallest possible black hole.)