Thursday, April 4, 2013

The universe as seen by Planck - Day Three (two rumours)

The conference dinner here is about to start (has already started), so I don't have time for a proper post. However, there were some very interesting rumours/revelations today so I'll write them down super-quickly. In increasing order of potential interest (note this post might be a bit technical, I'll explain all of this before the end of the weekend):

The feature at l=1700

A senior Planck figure gave a talk today on the features in the Planck angular power spectrum. Much of his talk was devoted to the apparent feature at \(l\simeq 1700\). In the 15 months worth of data that Planck has used to generate the cosmological results shown in their released papers, the statistical significance of this feature (when any feature is looked for) was \(\sim 3\sigma\). This was with a look elsewhere effect that took into account the possibility of the feature occurring at another \(l\) value.

What he let slip was that, when they analyse this same feature with the full temperature data set, the significance of the feature drops to \(\sim 2\sigma\).

Of course, not too much should be read into this because the additional data isn't quite as well understood as that first 15 months; however, its the same telescope looking at the same sky and foregrounds, so there shouldn't be too many complications. Note that this feature is out of the resolution range of Planck's polarisation capabilities, so the new temperature data is the only additional data we will get in the next data release.

Planck's data analysed on the SPT sky

One of the curiosities of the Planck release was that it seems to give cosmological results that are slightly discrepant with what the South Pole Telescope was giving. If Planck disagrees with BAO or supernovae, or galaxy clusters this is all interesting, but potentially the result of Planck and/or one of those other analyses getting it wrong. However, SPT is another CMB experiment, the fact that Planck and SPT are a bit discrepant is very confusing.

Perhaps SPT made a mistake and the CMB they measured is not the correct CMB?

The obvious way to test this is to analyse the Planck data on the same part of the sky that SPT measured. I overheard a conversation between lead figures in Planck, WMAP and SPT and it seems this is exactly what SPT have done (in unpublished work).

The result is striking.

They found a cosmology that agrees with SPT.

If true, this means that it isn't just Planck and SPT that are slightly discrepant, but different regions of Planck's sky.

What this means cosmologically is unsure. I'll speculate a bit tomorrow.

Power asymmetry

There was quite a bit of excitement over a plot that showed power asymmetry in different directions of the sky. I was going to write about it, but upon reflection, the excitement seems confusing. I'll try to explain the excitement and background before the end of the week.

[The final summary is now available here]


  1. Thanks Shaun. These posts are great. Nice to be able to keep up with what's being talked about in Noordwijk. Looking forward to your next post!

    1. Thanks Daniel. I'll probably write something tomorrow afternoon (Dutch time) so that I can enjoy a relaxed weekend.

  2. Great posts! I love all these rumours, almost like being there in person. I'm looking forward to hearing more about the first two tomorrow.

    However, regarding the hemispherical asymmetry (and many of the other anomalies that we already knew from WMAP data), I think I agree with your lack of excitement. Is there any discussion at the conference about why Planck only chose to apply the statistical tests of anomalies previously proposed by those scientists who are now members of the Planck team, and ignored alternative tests used by the WMAP team or various unaffiliated individuals (which, at least with WMAP data, tended to suggest the anomalies were much less anomalous than claimed)?

    1. It was actually a non-Planck scientist who got crazy-super-excited about the asymmetry plot. (I think) The significance Planck quote relates to the probability of the direction of asymmetry remaining the same over all multipole ranges (even though it actually doesn't because it oscillates between the one direction and the direction of the dipole). The person in the audience had done an estimate of the magnitude of the asymmetry and was convinced it would be \(>6\sigma\).

      His arguments seemed reasonable and nobody in Planck could give a calculated significance (they haven't done it). I was going to write all this down, but then I noticed in the Planck paper that they also show a plot of the magnitude of the asymmetry along the dipole direction and it is comparable. Also, the points from multipole to multipole oscillate by more than the questioner's estimate of the uncertainty, so something must be wrong there.

      There hasn't been any discussion about the non-Planck team's lack-of-significance calculations, but there have been the occasional semi-heated discussion about whether we should care about the anomalies at all.

      They haven't been *that* important a thing to the conference.

  3. Thanks for all these posts Shaun they are very informative.

    Interesting what you say about the `SPT sky' - the appendix of the cosmological parameters paper is moderately impolite towards SPT, it doesn't seem to need too much reading between the lines to get a 'Planck message' that SPT doesn't do a correct matching onto low-l WMAP spectra.

    1. Thanks. Yeah, that's how I read the Planck appendix too. The SPT line seems to be that the discrepancy comes partially from WMAP having more power than Planck (which WMAP seems quite worried about - I think they think they might have made a mistake) and partially "extra smoothing of the high l peaks" in Planck compared to SPT. I think the interpretation is that this could be the result of lensing, which would result from extra matter (hence Planck's extra matter).

  4. From what I've read in these posts, physics conferences like this are a hell of a lot more tantalising than the equivalent big conferences in medical/biological sciences. We never have floods of highly-anticipated, unpublished data like this; I wish we did! Most of the data presented at conferences in my field are already published or are just extensions of published findings. I guess it comes from the scale and cost of the projects involved: work like Planck's has to be unveiled at large conferences do give a fully-rounded explanation because there's just so much data. Moreover, because the experiments involve massive multi-billion pound investments (like building a satellite!) they can release the data without fear of being scooped to publication. Molecular medicine/biology projects are relatively cheap in comparison and often overlap more severely so researchers are wary of presenting significant data that aren't published in case they alert others who are working on very similar things who then beat them to publication. The only big projects that get presented before publication are trials involving patented drugs so there's not risk of being scooped, but the outcomes of those trials aren't usually groundbreaking.

    I also think that physics generally does a bit better on the PR side of things than most other sciences - the LHC is the epitome of this - and holding this kind of high-profile, media-friendly conference is grist to the mill of popular physics. Medical sciences could take a leaf out of physics' book!

    Kudos for the reporting, Shaun - as an outsider I'm finding it all fascinating!

    1. To a certain degree, this conference was exceptional. Planck has been quite unique in the way it has published its results. Partially this is because of the lack of risk of getting scooped, but, also to a certain degree, they *did* get scooped by the two Earth based telescopes ACT and SPT.

      I agree that physics does a better job of PR, especially CERN. I still think we can/should do better. If Planck or the LHC had wanted to, with the right marketing I think they could have filled movie theatres with a pre-filmed documentary announcing the discoveries. The money made from this wouldn't be substantial, but the culture it created would be. Science would become an event for the public to attend, etc.

      Regarding the conference again though. I think physics is more collaborative than most (if not all?) other sciences/academic disciplines. It is not uncommon for two or more researchers to start discussing a problem at one of these conferences and to have those discussions grow into a full research project and paper.

  5. Shaun,

    do the new results from the Planck satellite shed any light on whether the cosmological constant decreases with time?


    1. Not much, unfortunately. Most of what Planck sees was set down at the last scattering surface, when the hydrogen plasma became hydrogen atoms. At this point of time, dark energy/the cosmological constant was utterly negligible in comparison to the energy density in matter and radiation.

      The cosmological constant does have an over-all normalisation effect on the CMB based on how much the universe has expanded since the CMB was formed. However various expansion histories can arrive at the same point today relative to the last scattering surface, so there is a lot of degeneracy.

      To really start testing whether the accelerated expansion is caused by a cosmological constant or something else we'll need more local observations. Planck does somewhat constrain the late/local-universe through the ISW effect and lensing effects, but not strongly.

      Baryon Acoustic Oscillations would probably provide the tightest constraints on the constancy of the cosmological constant so far, but even then there is, as yet, much room for manoeuvre. Although, the data still prefers a cosmological constant over any of the usually examined alternatives.

  6. Thank you. It is really a pity that so little can be deduced from the Planck data!

    All the best for your research!