Wednesday, March 20, 2013

The Planck guessing game

At 10 am CET on Thursday morning, the Planck mission will hold a press conference and announce the first cosmology results based on data from their satellite, which has now been in orbit for nearly 1406 days, according to the little clock on their website. (I think the conference information will be available live here, though the website's not as clear as it could be.)

Planck is an incredible instrument, which has been measuring the pattern of cosmic microwave background (CMB) temperature anisotropies with great precision. And the CMB itself is an incredible treasure trove of information about the history of the universe, telling us not only about how it began, but what it consists of, and what might happen to it in the future. When the COBE and WMAP satellites first published detailed data from measurements of the CMB, the result was basically a revolution in cosmology and our understanding of the universe we live in. Planck will provide a great improvement in sensitivity over WMAP, which in turn was a great improvement on everything that came before it.

Another feature of the Planck mission has been the great secrecy with which they have guarded their results. The members of the mission themselves have known most of their results for some time now. Apparently on the morning of March 21st they will release a cache of something like 20 to 30 scientific papers detailing their findings, but so far nobody outside the Planck team itself has much of an idea what will be in them.

So let's have a little guessing game. What do you think they will announce? Dramatic new results, or a mere confirmation of WMAP results and nothing else? I'll list below some of the things they might announce and how likely I think they are (I have no inside information about what they actually have seen). Add your own suggestions via the comments box!
Tensors: Planck is much more sensitive to a primordial tensor perturbation spectrum than the best current limits. If they did see a non-zero tensor-to-scalar ratio, indicative of primordial gravitational waves, this would be pretty big news, because it is a clear smoking gun signal for the theory of inflation. Of course there are other bits of evidence that make us think that inflation probably did happen, but this really would nail it.

Unfortunately, I think it is unlikely that they will see any tensor signal – not least because many (and some would argue the most natural) inflation models predict it should be too small for Planck's sensitivity.

Number of relativistic species: CMB measurements can place constraints on the number of relativistic species in the early universe, usually parameterised as the effective number of neutrino species. I wrote about this a bit here. The current best fit value is $N_{\rm eff}=3.28\pm0.40$ according to an analysis of the latest WMAP, ACT and SPT data combined with measurements of baryon acoustic oscillations and the Hubble parameter (though some other people find a slightly larger number).

I would be very surprised indeed if Planck did not confirm the basic compatibility of the data with the Standard Model value $N_{\rm eff}=3.04$. It will help to resolve the slight differences between the ACT and SPT results and the error bars will probably shrink, but I wouldn't bet on any dramatic results.

Non-Gaussianity: One thing that all theorists would love to hear is that Planck has found strong evidence for non-zero non-Gaussianity of the primordial perturbations. At a stroke this would rule out a large class of models of inflation (and there are far too many models of inflation to choose between), meaning we would have to somehow incorporate non-minimal kinetic terms, multiple scalar fields or complicated violations of slow-roll dynamics during inflation. Not that there is a shortage of these sorts of models either …

Current WMAP and large-scale structure data sort of weakly favour a positive value of the non-Gaussianity parameter $f_{\rm NL}^{\rm local}$ that is larger than the sensitivity claimed for Planck before its launch. So if it lives up to that sensitivity billing we might be in luck. On the other hand, my guess (based on not very much) is it's more likely that they will report a detection of the orthogonal form, $f_{\rm NL}^{\rm ortho}$, which is more difficult – but not impossible – to explain from inflationary models. Let's see.

Neutrino mass: The CMB power spectrum is sensitive to the total mass of all neutrino species, $\Sigma m_\nu$, through a number of different effects. Massive neutrinos form (hot) dark matter, contributing to the total mass density of the universe and affecting the distance scale to the last-scattering surface. They also increase the sound horizon distance at decoupling and increase the early ISW effect by altering the epoch of matter-radiation equality.

WMAP claim a current upper bound of $$\Sigma m_\nu<0.44\;{\rm eV}$$ at 95% confidence from the CMB and baryon acoustic oscillations and the Hubble parameter value. But a more recent SPT analysis suggests that WMAP and SPT data alone give weak indications of a non-zero value, so it is possible that Planck could place a lower bound on $\Sigma m_\nu$. This would be cool from an observational point of view, but it's not really "new" physics, since we know that neutrinos have mass.

Running of the spectral index: Purely based on extrapolating from WMAP results, I expect Planck will find some evidence for non-zero running of the spectral index. But given the difficulty in explaining such a value in most inflationary models, I also expect the community will continue to ignore this, especially since the vanilla model with no running will probably still provide an acceptable fit to the data.

Anything else? Speculate away … we'll find out on Thursday!


  1. My guesses are mostly in line with yours. I have two comments.

    - Regarding tensors, I don't think that tomorrow Planck will add much sensitivity. Primordial tensors only affect the largest scales of the temperature anisotropy. WMAP has measured them quite well. Planck will gain extra information about the over-all normalisation by measuring the small scales and will thus gain a little extra information about the power on large scales, but only a little. Where Planck really will increase the sensitivity regarding tensors is in its polarisation measurements; however those aren't being released tomorrow.

    - Regarding neutrino masses. With just measurements of temperature anisotropies, Planck won't be all that sensitive to neutrino masses. If it does have evidence it will partially come also from the abundance of galaxy clusters that Planck will detect using the SZ effect (of which Planck should have found many). The growth rate of structure is quite sensitive to neutrino masses, so the abundance of clusters is too.

    1. I didn't realise the polarisation results weren't going to be announced tomorrow. In that case tensors are even less likely.

      I expect Planck will be at least as sensitive to neutrino masses from temperature anisotropies alone as WMAP+SPT together. Probably more, and then as you say they will have SZ measurements. WMAP+SPT alone favours a value for $\Sigma m_\nu$ around 1 eV (a bit more than 1$\sigma$ away from zero). But obviously adding the BAO and H0 constraints in will help a lot too, and they're definitely going to do this.

  2. The interesting thing for me is the value of Omega+lambda. This is quite near 1 and constraints have been improving. Barrow and Shaw predict 1.0056. If the error bars get substantially smaller AND 1.0056 is still allowed, or even favoured, then surely it is just a matter of time until the Nobel Prize is awarded to Barrow and Shaw.

    I don't have as much faith in retrodictions (i.e. explaining known facts, even if the explanation looks logical) as predictions, especially predictions which can be falsified and which not everyone is predicting.

    1. Planck on its own will be unable to constrain $\Omega_k$ to that accuracy without other strong assumptions about the equation of state of dark energy. To get tight enough constraints requires independent measurements of distance scales at redshifts of a few. We're not going to get that tomorrow.

    2. Barrow and Shaw also predict w=-1 exactly, so it would be interesting to see if their value 1.0056 for Omega+lambda can be ruled out if one assumes w=-1. It doesn't have to be using just the Planck data; if one can rule out 1.0056 assuming w=-1 exactly and using some other data, then their model is ruled out. It's a very testable prediction. (I note that Barrow also has another paper where he claims that large-scale magnetic fields are primordial and this favours (how strongly, I'm not sure) a spatially open universe, so his two predictions are not very compatible. But that, in itself, is not a problem.)

    3. In case you are interested Phillip, the Planck constraints, when combined with all the other data available, are $\Omega_k=-0.0005^{+0.0065}_{-0.0066}$. Which leaves us none the wiser about Barrow and Shaw's claim.

      When I said measurements at a redshift of a few were required, I really meant redshifts of $z$~$3$. The BAO data, which are the best we currently have, are at $z<0.8$.

    4. How many sigma are these constraints?

      In these units, they predict -0.0056. So, this is still allowed, though it is at one side of the allowed range.

      Ruling out 0 would also be very interesting.