> Please note that the paper is “Groeneboom and Eriksen”, not “Eriksen et al.”…

Oops, my mix-up! (literally)

> But do you think the current experiments are deep/wide enough to really get a good handle on this effect?

At Fisher matrix level the SDSS photometric samples should be able to detect g* if it’s really 0.15 … at back of the envelope level since there are Fourier modes spanning a wide range of directions you need ~1/g*^2~44 linear modes to measure it. The factors of order unity are unfortunately not so kind: the Fisher matrix has a factor of 2/45 in it (because the variations in the power spectrum as a function of angle are 10^4 modes. BUT: No promises until the systematics tests are all in

Chris

A few comments to your posts:

1) The problem with a cross-correlation analysis (between, say, V and W) for this particular analysis is that it’s very difficult to handle the signal covariance matrix on a cut sky. The only reason it’s sparse in our analysis is that we’re using the Gibbs sampling algorithm, which essentially “fills in” the cut region. I don’t see many alternatives to this, really, if one wants to go to high l’s. However, the Gibbs sampler is an exact likelihood approach, and as such, intrinsically an auto-correlation method; it’s not straightforward to get rid of these auto-correlations, while still have a correct likelihood. Of course, the problem becomes smaller the more independent bands you have, but it’s always going to be there to some extent. But of course, in principle it’s of course possible that one may construct some “pseudo-Cl” approach for this particular model, but right now, I don’t see how.. For the moment, I think the best approach is simply to analyse realistic WMAP5 noise simulations, and see if something similar pops up.

2) Personally, I don’t think asymmetric beams is relevant for this result. The model signature has a substantial (as in several degrees, I’d say) correlation length along the plane normal to the preferred axis, and even though the WMAP beams are somewhat asymmetric, they’re not *that* asymmetric.. Correlated noise is definitely my biggest concern here.

and

3) Please note that the paper is “Groeneboom and Eriksen”, not “Eriksen et al.”…

Finally, I’m really looking forward to see what comes out of the LSS analyses! But do you think the current experiments are deep/wide enough to really get a good handle on this effect? Or do we need to wait for the next generation surveys?

Thanks!

That said, of the possible systematics that could produce an asymmetry in the power spectrum, the first one on my list would be beam ellipticity because WMAP does not hit each pixel at a uniform distribution of angles of attack. (Same will be true, more so, for Planck.) The cross-power analysis won’t solve this problem, ultimately one needs to simulate it using the known beam maps and see what happens.

Regarding the search in large scale structure: Anthony Pullen (here at Caltech) is working on it, so stay tuned. I’m sure there will also be a lot more poring over WMAP and soon Planck, and probably other LSS data sets shortly after that. I for one find the situation exciting. A few years ago I went to conferences where people presented “explanations” of the low-multipole anomalies that made no predictions that I could hope to see verified at many sigmas in my lifetime. Well, with this particular anomaly I hold out hope that in 5-10 years it will either have gone away or be seen at many sigma in both CMB and LSS …