HEALPix: E and B mode polarization power spectra

Use of Healpix, camb, CLASS, cosmomc, compilers, etc.
Post Reply
Josha van Houdt
Posts: 2
Joined: August 18 2015
Affiliation: Leiden University

HEALPix: E and B mode polarization power spectra

Post by Josha van Houdt » August 19 2015

Hello,

Yesterday I attempted to extract E and B mode maps from the Planck Q and U stokes maps. I tried this in HEALPy first, later we did the same with the IDL implementation of HEALPix.

I'm using the latest version of healpy, 1.8.6. I'm using the 2015 datarelease of Planck, with the smica pipeline (COM_CMB_IQU-smica_1024_R2.02_full.fits)
As a test, we wanted to make a quick plot of the E and B power spectra. To this end, we first read in the maps such that 'IQU' below is a tuple of the three relevant maps.

Code: Select all

Cls = hp.anafast(IQU,pol=True)
Plotting this, we saw the following:

EBpowspec.png

Comparing this to CAMB output, the EE spectrum (cyan) looked quite fine in shape and amplitude; the BB spectrum however seems to have the expected shape of a lensing-induced B mode spectrum, but a couple of orders of magnitude too big!

What am I doing wrong, or what am I missing?

Thanks in advance!!

Antony Lewis
Posts: 1547
Joined: September 23 2004
Affiliation: University of Sussex
Contact:

Re: HEALPix: E and B mode polarization power spectra

Post by Antony Lewis » August 20 2015

Is that not just noise?

Josha van Houdt
Posts: 2
Joined: August 18 2015
Affiliation: Leiden University

HEALPix: E and B mode polarization power spectra

Post by Josha van Houdt » August 20 2015

That's what I thought as well, but then I couldn't really make sense of why it looks like the lensed B spectrum and falls off at the high end?

bin hu
Posts: 23
Joined: September 18 2008
Affiliation: leiden univ

HEALPix: E and B mode polarization power spectra

Post by bin hu » August 20 2015

One question, is the Planck sensitivity good enough to detect directly the lensing B mode? Not indirectly computed from the leakage of E mode?

Antony Lewis
Posts: 1547
Joined: September 23 2004
Affiliation: University of Sussex
Contact:

Re: HEALPix: E and B mode polarization power spectra

Post by Antony Lewis » August 21 2015

On the shape, I assume you haven't deconvolved with the 10 arcminute effective beam or the pixel window function - see

http://wiki.cosmos.esa.int/planckpla201 ... n_products

bin hu
Posts: 23
Joined: September 18 2008
Affiliation: leiden univ

HEALPix: E and B mode polarization power spectra

Post by bin hu » August 24 2015

Hi Antony,

If my understanding is correct, NILC should be deconvolved with 5 arcmin beam. I tried ``COM_CMB_IQU-nilc_1024_R2.02_full.fits'', the bb spectrum size and shape are still pretty much as Josha got from ``COM_CMB_IQU-smica_1024_R2.02_full.fits''.

Image

Post Reply