Page 1 of 1

[0801.1491] High resolution CMB power spectrum from the com

Posted: January 11 2008
by Alessandro Melchiorri
Hi all,

The new results from ACBAR are impressive.
However I have some doubts on the following sentence:
"We see evidence for weak gravitational lensing of the CMB at >3-sigma significance by comparing the likelihood for the best-fit lensed/unlensed models to the ACBAR+WMAP3 data.".
How you can claim any evidence for weak lensing by just comparing the best fits ?
The "true" model can be 2 sigma away from the best fit in both cases...
I don't understand...

cheers
Alessandro

Re: [0801.1491]

Posted: January 12 2008
by Antony Lewis
I agree. What I would do is introduce a new parameter A_L that scales the lensing potential, and plot the posterior constraint on A_L. So A_L=0 correspond to unlensed, A_L=1 is the expected lensed result. The ratio of the marginalized posterior curve at 1 and 0 gives you the posterior odds for lensing. But you can also look at the curve, and if it peaks at A_L=10 then you should be a bit worried. [I did this with old WMAP3+ data, the peak was away from zero, but consistent with both zero and one - great ACBAR is now giving something interesting!]