Monthly Archives: April 2013

Planck results: Boring or Anomalous?

The Planck Collaboration released a 29-paper avalanche on the cosmology community just over a week ago, comprising their first set of CMB results (all of the papers are listed here). Along with a lot of people in the field (I imagine…), I’ve been letting the implications of this data release sink in for a few days and, on the face of it, the results are a little boring: no detection of non-Gaussianity, no extra neutrino species, “robust support for the standard, six parameter LCDM model of cosmology” — standard, standard, standard. OK, so the best-fit parameters are a bit different to the WMAP 9-year ones, but that hardly seems like front-page news. A few other anomalies were flagged up in the Planck papers too, but none of the cosmologists I’ve spoken to so far have seemed particularly excited about them. Raised eyebrows, murmurs of “that’s interesting”, but no suggestion that we’re looking at anything earth-shattering.

But hey; they’re worth a look, right? These are the “anomalies” that have piqued my interest, and why:

CMB/SZ σ8 discrepancy: The value of σ8 measured from the CMB is in tension with the value measured from the SZ cluster sample. The Planck Collaboration “attribute this tension to uncertainties in cluster physics”, but an interesting note by Macaulay, Wehus and Eriksen suggests that the SZ value is actually pretty consistent with a number of independent redshift space distortion (RSD) measurements from galaxy redshift surveys (see figure, right). This is interesting: If Planck are right about the reason for the CMB/SZ discrepancy, then perhaps there is a potentially important systematic bias in the RSD analyses that has been left unaccounted for. And if they’re wrong, and the value of σ8 from multiple probes of large-scale structure is genuinely different from the CMB value, then maybe we’re looking at some interesting physics (e.g. a spatial variation in the power spectrum amplitude, or an interesting dark energy equation of state).

f * sigma_8 from Macaulay et al.

f * sigma_8 from Macaulay et al. The dashed line is the theory curve for the Planck CMB best-fit parameters, and is clearly discrepant with the RSD data.

Hemispherical power asymmetry: The Planck results confirm the existence of a hemispherical power asymmetry that was previously seen by WMAP. This is the result that the amplitude of the angular power spectrum appears to have a dipolar variation across the sky. A letter by Liang Dai and others considers a number of physical models for this asymmetry, including a spatial variation in the initial scalar spectral index, an isocurvature mode, and an inhomogeneous reionisation optical depth. If some of these explanations turn out to be true, we might need to seriously revise our picture of what the Universe looks like on the largest scales (which would be cool).

H0 discrepancy: The best-fit value of the local Hubble rate found with Planck is significantly lower than the Riess et al. measurement from local distance measures. I’ve been noticing for a while that H0 estimates from WMAP+BAO analyses were tending to come in lower than the Riess value, but the new value from Planck has smaller error bars, and so the discrepancy is more significant. Again, the simple answer is that someone has a systematic error that they need to deal with, but maybe there could be some neat physics here too. Valerio Marra and others point out that one should expect a local measurement of H0 to differ from the background value as a result of cosmic variance, although they expect a smaller deviation than the observed one. Myself and Tim Clifton have previously suggested that you’d expect such a discrepancy if backreaction effects were important too.

Bulk flow: The official Planck analysis failed to detect a bulk flow using the KSZ effect from galaxy clusters, which is ostensibly bad news for previous claims of an anomalously large bulk flow detection with WMAP. A subsequent paper by Atrio-Barandela claims that the statistical errors have been overestimated in the Planck analysis, and that a significant detection would be found if they were corrected. A large bulk flow seems to be quite difficult to explain within the standard LambdaCDM framework, although I think this result needs more work to be convincing.