Page 1 of 1

[0809.3734] A measurement of large-scale peculiar velociti

Posted: October 01 2008
by Syksy Rasanen
(This is a companion to the longer paper 0809.3733, where the authors provide more details.)

The authors claim to have detected a coherent bulk flow over a scale of at least 300 h^{-1}Mpc. While this might not be completely surprising (see 0809.4041, for example), their method leaves me baffled.

They claim to make the detection using the kinetic SZ effect. They argue as follows, as far as I understand (which is not very far). Take the CMB sky and a cluster catalogue. Remove the cosmological contribution from the CMB sky. Calculate the dipole and monopole at the location of the clusters from this subtracted sky. Now you have the thermals and kinetic SZ components, and from the dipole you identify the kinetic SZ effect, which you can use to measure the peculiar velocity of the cluster.

If their claim is true, it's important as a detection of the kinetic SZ effect at a high significance, apart from the implications for bulk flows. But there are a number of things I don't understand. One of them is: how is it possible to subtract the cosmological CMB sky from the observed one? I understand doing simulations (given a cosmological model) to assess whether something is likely or unlikely to be present in a realisation of a certain cosmology, but that does not seem to be what they are doing. They write that they use a Wiener filter (page 3 of this paper, page 12 of the longer paper), which doesn't say a lot to me.

[0809.3734] A measurement of large-scale peculiar velociti

Posted: October 02 2008
by Fergus Simpson
Also Figure 1(f) seems to suggest we should be involved in a bulk flow of around 1500 kms^{-1}. If this were true, then presumably our own peculiar velocity would have to be quite contrived (large and aligned anti-parallel to the flow) in order to produce the CMB dipole of v < 400kms^{-1}.

[0809.3734] A measurement of large-scale peculiar velociti

Posted: October 02 2008
by Syksy Rasanen
Well, the error bar is quite large. Doubling that would get you down to 500-600 km/s. In the conclusions of the longer paper, they quote the value 600-1000 km/s.

[0809.3734] A measurement of large-scale peculiar velocitie

Posted: October 03 2008
by Fergus Simpson
True. But then if you treble them you can get down to v=0. :)

Even with a bulk flow of 800kms^{-1}, which is lower than any of their data points, and assuming our peculiar velocity matched that in magnitude,
the alignment is still a bit of a problem. You'd still find a ~95% chance (I think) of our dipole being larger than it is.

Re: [0809.3734] A measurement of large-scale peculiar veloc

Posted: October 03 2008
by Syksy Rasanen
Fergus Simpson wrote:True. But then if you treble them you can get down to v=0. :)
I don't see your point. You only need the real value to be at the edge of the error bars at our location to be compatible with the CMB dipole, whereas getting rid of the bulk flow would require all points being at the edge.

However, I agree that there seems to be a bit of a coincidence required, in that we happen to be in a location where the flow is much smaller than the mean bulk flow (or anticorrelated with the CMB dipole).

All of this is assuming the analysis would be correct, of course...