Wednesday, June 5, 2013


This is a continuation of posts about the Cosmological Perturbations post-Planck conference being held in Helsinki this week. You can see my introduction post here and my first post from the conference here. If you're a member of the general public and want to understand more, please ask. If you're a cosmologist and want to add anything to what I've written, please add a comment.

The Curvaton

Yesterday I tried to introduce what the curvaton is. We had a few talks yesterday that related to this particular framework for generating the initial perturbations in the density of the universe (the curvaton is named the "curvaton" because, in this framework it is responsible for perturbations in the curvature of space-time). Prior to the Planck release, the curvaton had become quite a popular model because it would be capable of producing a distribution of density perturbations that was almost, but not quite Gaussian. This is quite a technical sounding term, non-Gaussian. To hopefully simplify it a little, I'll say that a Gaussian distribution is the familiar bell-shaped curve of a normal distribution. There is no a priori reason to expect that the distribution of the primordial density perturbations has to be Gaussian, but in many aspects of physics (and statistics in general) a Gaussian distribution does turn out to be the default.

However, even before Planck, we knew that the distribution of primordial density perturbations was close to Gaussian. Planck was going to be capable of measuring this distribution even more accurately and thus would be sensitive to even more subtle deviations from a Gaussian distribution. Ordinary inflation predicts a deviation from a perfect Gaussian distribution that would have been too small to detect with Planck. And, the WMAP satellite's measurements of the CMB provided a small amount of evidence that the perturbations did deviate from being Gaussian. This would have been fascinating to discover, and if WMAP's best-fit distribution had been true, Planck would have detected it beyond "all reasonable doubt".

Unfortunately, as we all know now, Planck found that WMAP's evidence was (probably) a statistical fluke (they do happen). So where, does that leave the curvaton?

David Wands gave a talk addressing precisely this question. I suppose the spirit of this talk (and a few others so far) could be summed up by one sentence on one of David's slides: "absence of evidence is not evidence of absence". This sounds like a cop-out, and of course to a certain degree it is. I'm certain David would have preferred to have been giving his talk in the context of a definitive detection of a slightly non-Gaussian distribution of density perturbations. He could tell us which specific curvaton models are favoured, which are ruled out, what needs done to tell the curvaton apart from other mechanisms that can generate non-Gaussianity, etc. On the other hand, the quoted sentence is also true. While the curvaton could generate detectably non-Gaussian perturbations, it could also generate perturbations that Planck wouldn't have distinguished from Gaussian ones.

However, the situation now is that there is no (strong) observational evidence that prefers a curvaton type mechanism to simple inflation. It is customary in this sort of situation to appeal to Occam's Razor and say that, in the absence of evidence that distinguishes between them, the simpler model should be preferred. In the case of the curvaton, I think this is probably going to be the community's consensus, for now (though if you're in the community and you disagree, speak up!).

The hemispherical power asymmetry

Having said that, David Lyth spoke yesterday about one of the infamous WMAP anomalies that Planck confirmed. David's choice of anomaly was the "hemispherical power asymmetry". This anomaly comes from the fact that the amplitude of the fluctuations in the temperature of the CMB along one particular line of sight seems to be systematically slightly larger than the amplitude in the opposite direction. I say "systematically" because this larger amplitude seems to persist when you average the CMB over a range of angular scales. Obviously, for any single angular scale there will be a direction of maximum asymmetry, but it wouldn't be expected that this direction of maximum asymmetry would be the same for other angular scales. The anomaly is then a combination of the fact the magnitude of this asymmetry is unlikely in the standard cosmology and the fact that all angular scales seem to have the same maximal direction of asymmetry.

I want to pause for just a moment to stress something for the people outside of the cosmology community who like to dwell on anomalies like this to claim that cosmology needs to be over-turned. This asymmetry is small (of the order of a few percent). The thing is though that Planck (and WMAP before it) has measured the CMB so incredibly accurately that even very small effects can now be noticed with quite strong statistical significance. Therefore, even if it turns out that this hemispherical asymmetry is more than a statistical fluke, this doesn't mean that the universe is very asymmetric. The universe would be almost symmetric, with a small perturbation away from perfect symmetry. It is certainly conceivably possible that some other, very different, model, will replace the current model (many cosmologists desperately hope for this); however whatever this model is it will still describe an almost perfectly symmetric universe, because that's not a theoretical prejudice, that's observed fact!

Back to David Lyth's talk. David started by making a somewhat over the top proclamation (mentioned in a comment in an earlier post about the conference) that the detection of this asymmetry was as important as the detection of the fluctuations in the CMB themselves (by COBE). I would probably back David up that if the asymmetry is not a statistical fluke and is primordial in origin, that it does rank as highly in importance; however, it is not unlikely enough to rule out the possibility that it is a fluke, yet. However, that wasn't the main point of David's talk. He's a theorist so he wanted to explain where the asymmetry might have come from (and in the process try to make a prediction for how to check whether this explanation is true).

Here, fans of the curvaton might have had their interest piqued, because David's explanation needs the curvaton to work. The method he described was originally proposed by Adrienne Erickcek, Mark Kamionkowski and Sean Carroll (EKC). Thankfully, Sean is also a blogger and has written a blogpost about this method. You should check it out.

I will try to give my own description later, but David did have a clear consistency relationship that would be satisfied if the curvaton and EKC method was responsible for the asymmetry...

The final post of the conference now appears here...

Twitter: @just_shaun


  1. Just wanted to thank you for devoting time to covering the conference like this. It's really great to read a personal description of the event as it unfolds - just like sports commentary/highlights! All very exciting.
    Great job, keep it up!

    1. Cool, thanks dude. We had the conference dinner tonight, so no chance to write anything from today. But the conference ends at about midday tomorrow, so I'll be free to write lots from then. I think I will then drip-feed what I write on Friday in daily posts until I've run out of stuff.

      Thanks again!