arxiv Find: Analysis of the Apparent Lack of Power in the CMB Anisotropy at Large Angular Scales

By Sean Carroll | March 5, 2007 1:21 pm

Here’s a paper that was mentioned in comments, about which I’m not qualified to say all that much: astro-ph/0702723, “Analysis of the apparent lack of power in the cosmic microwave background anisotropy at large angular scales,” by Amir Hajian.

We study the apparent lack of power on large angular scales in the WMAP data. We confirm that although there is no apparent lack of power at large angular scales for the full-sky maps, the lowest multipoles of the WMAP data happen to have the magnitudes and orientations, with respect to the Galactic plane, that are needed to make the large scale power in cut-sky maps surprisingly small. Our analysis shows that most of the large scale power of the observed CMB anisotropy maps comes from two regions around the Galactic plane (~9% of the sky). One of them is a cold spot within ~40 degrees of the Galactic center and the other one is a hot spot in the vicinity of the Gum Nebula. If the current full-sky map is correct, there is no clear deficit of power at large angular scales and the alignment of the l=2 and l=3 multipoles remains the primary intriguing feature in the full-sky maps. If the full-sky map is incorrect and a cut is required, then the apparent lack of power remains mysterious. Future missions such as Planck, with a wider frequency range and greater sensitivity, will permit a better modeling of the Galaxy and will shed further light on this issue.

There are two issues here, as I understand it. Here’s a map of the temperature fluctuations in the CMB, from WMAP:

CMB map

When you decompose this into contributions at different angular scales (spherical harmonics), you get this power spectrum:

WMAP power spectrum
The point on the far left, the quadrupole at l=2, seems to be low compared to the predictions of the standard cosmological concordance model. That’s one thing. The other thing is that, when you dig into the individual contributions that are grouped together to make this plot, the other low-l contributions seem to pick out a preferred direction on the sky, sometimes called the axis of evil.

So that’s intriguing, but it’s not completely clear whether it’s really significant, or just an accident. For one thing, the preferred direction seems to match up pretty well with the ecliptic (the plane in which the planets orbit the Sun), possibly indicating some systematic error rather than a cosmological effect. We don’t get an unvarnished view of the primordial microwave background; it comes to us through the galaxy, and through the material in the Solar System itself.

This paper seems to be claiming that the large-angle anomalies are, in fact, just a matter of foreground contamination. At least I think that’s what it’s saying; there are a lot of negatives (“although there is no apparent lack of power…”). Of course, the abstract concludes in the way that all good data-analysis abstracts should: we need more data! Happily it’s coming, in the form of the Planck satellite. One or more of our expert readers may chime in.

  • Tom Renbarger

    There is no apparent lack of tortured prose in the abstract.

    I think your summary is correct, but I haven’t read the paper in detail, I must admit.

  • nc

    This paper seems to be claiming that the large-angle anomalies are, in fact, just a matter of foreground contamination.

    Microwaves (the CBR) are the most penetrating radio waves there are, but the ionised plasma in stars stops them easily. A local galaxy such as our galaxy the Milky way has a lot of ionised material which will create an asymmetry to the side of us where the centre of the galaxy is located.

    I did cosmology over a decade ago, but in the course guide there was quite a bit about the various corrections applied to the COBE data before these fluctuation spectra were extracted. It’s amazing, because the CBM is the most perfect blackbody radiation spectrum ever seen, yet the accuracy is supposedly so good (i.e. most error bars are invisible in the WMAP data analysis above) that you can take the very small angular variations and find a lot of physics in them.

    It would be good to know exactly what has been subtracted from the data. Obviously, the Milky Way noise has been deleted, but what has been filled in to replace the parts of the WMAP data which are spoiled by non-CBR “noise” which happens to be at the same microwave frequency?

    In regard to the evil of preferred directions, there is a massive istropy for the absolute motion of the Milky Way with respect to the CBR; the galaxy is going toward Andromeda at 400 km/s which causes a whopping +/-3mK istropy in the raw data which is simple Doppler shift:…64M

  • nc

    That link only works if you retype the … in it in your browser (this blogger software automatically replaces three separate full stops with a nicer looking character consisting of a triad of close spaced dots for abbreviation, which of course doesn’t hyperlink).

  • Garth

    This low-l deficiency has been around for some time now, COBE, BOOMERANG, WMAP years 1,2 & 3 and now we wait for Plank.

    While waiting can we ask what would be the consequence of the deficiency actually being real? i.e. not a statistical fluke or local contamination?

    Would it not be a good idea to be more questioning about the robustness of the standard LCDM model and already be actively searching for and evaluating alternatives?


  • Josh

    Wha?… Where are the videos of cute cats? Is this a blog or some sort of water-boarding like torture device?

  • Sean

    I believe you can find cute cats and a torturing device right here.

  • Cynthia

    Gotta say, this place isn’t called Cosmic Variance for nothing—minus the axis of evil, of course—minus the kitty torture, no doubt!;)

  • Chris Hirata

    nc: The issue with the Milky Way is not so much the microwaves that are absorbed (the optical depth is of order 10^-4 and is dominated by Thomson scattering) but those that are emitted. In addition to the radiation emitted by thermal plasma (“free-free”) there is synchrotron radiation and microwaves emitted by dust grains.

    What has been eliminated from WMAP: the maps available online have the isotropic intensity taken out (which is not even measured since WMAP can only see temperature differences), and have the dipole (which is mostly due to our motion) subtracted. WMAP has “raw” maps available that contain everything else in the sky (including Galactic emission) as well as several maps that have models of the Galaxy subtracted. These models are usually based on emission at other wavelengths, e.g. synchrotron radiation shows up very brightly in the radio; dust is brightest in the infrared; and the plasma that emits free-free radiation also emits spectral lines of hydrogen (6563 angstroms) that have been mapped with optical telescopes. There are also maps (e.g. the one by Tegmark et al) where the angular distribution of foregrounds is inferred entirely from microwave data, i.e. based on the temperature difference between maps at different frequencies.

    Some regions of sky are so contaminated that WMAP did not use them to calculate the power spectrum. You can read their papers if you want details, but estimating the power spectrum (C_l) using only part of the sky is a statistics problem and there are many estimators that one could write down that do the job (e.g. maximum likelihood assuming Gaussianity, although there are others). When people discuss alignment of the multipoles, “axis of evil”, etc. then in most cases you are forced to use the whole sky, and then you have to worry a lot more about how good your subtraction of the Galaxy is (particularly in the plane). This is why Amir is looking forward to Planck data, which will cover the millimeter wavelength range and hence get much better measurements of the dust in the plane …

  • Jason Dick

    This is in response to nc’s post #2 above.

    There are two methods that can be used to subtract foregrounds. One is to attempt to tease out what part of the measured CMB signal is foreground signal, and what part is primary CMB. This can be done for most forms of contamination through multi frequency analysis, and it basically affects your uncertainty in the temperature value for any given pixel.

    The second method that can be used is to mask out the contaminated regions of the sky and only do your analysis in those areas where the primary CMB is the strongest signal.

    The two methods are best combined, of course: some masking must be done, because many foregrounds are just going to be entirely too bright. But the question remains as to what precisely this does to our data.

    The basic effect is that it means the multipoles are no longer uncorrelated. In essence, the spherical harmonic transform is only orthogonal if the integration is performed over the whole sky. So if you calculate the spherical harmonic transform of a masked sky, some of your multipoles will be mixed into one another as a result of the mask. The goal, then, is to properly account for this induced correlation, and hopefully remove it.

    One method of doing this is Gibbs sampling. With Gibbs sampling the basic idea is to simulate a bunch of maps with different power spectra and compare them to the CMB that is actually measured. In this way it is possible to marginalize over possible values of the data outside of the region in which the CMB is actually measured. This method is too computationally intensive to take out to very high multipole moment, but works excellently on large scales. You can see this method explained in the following paper:

    For smaller scale fluctuations the problem is computationally more difficult, but the induced correlations are also smaller so we can afford to be a bit more sloppy. There are a number of semi-analytic methods for subtracting out the effect of the masking on the data. For example, if you assume statistical isotropy (that is, that the various multipole directions on the primary CMB are truly uncorrelated from one another), then the problem of how to factor in the mask becomes computationally much simpler, and its effect can be integrated directly. Once this is done, and you have a basic mapping from the power spectrum of the masked map to the primary CMB power spectrum, you can iteratively correct the mapping by averaging over many realizations of the same primary CMB power spectrum. This method is highlighted in the MASTER algorithm, which is recommended for smaller area CMB experiments:

    I don’t want to give the impression that these are the only two methods, or that they’re even the best methods available. These are just two that I have been exposed to in my work under Lloyd Knox on the subject.

  • Garth

    and now we wait for Plank

    The joys of typos!
    Of course what I meant to say was

    and now we wait for Plonk!


  • Hard-ass experimentalist

    Only theorists would get excited about one point on a curve being two sigma off a model. Even if the lowest l mode is low, cosmic variance (Hello? Name of this blog?) means that little meaningful inference can be made based on it.

  • Jack

    Only theorists would have this weird idea that the universe might be interesting. Experimentalists know that the most boring possibility is *always* the right one.

  • B

    This is really interesting, thanks for mentioning the paper. (I hardly ever check astro-ph, just too much papers there).

    Regarding the alignments, see also

    Anomalous Alignments in the Cosmic Microwave Background

    Sorry for the self-ad :-)

    PS: I’ve submitted this comment already twice yesterday, but it never showed up. I hope it works this time…

  • spaceman

    When I look at the acoustic peaks in the CMB power spectrum (shown above) a question arises in my mind: how physically important are the low-l anomalies in light of the fact that the vast majority of the data points are in line with the theoretical curve in a Universe dominated by dark energy and dark matter? In other words, is it likely that these low-l anomalies are significant enough to completely overturn our coherent pciture?

    Perhaps we are expecting our data to fit the theoretical curve to perfection. We need to ask ourselves: is such an expectation realistic? Hardly ever in life do things turn out exactly the way we expect. The vast majority of astronomical data is consistent (the Universe is almost certainly 10-15 Gyr old, it’s not too curved, it has more matter and energy than is obviously detectable); hence the name “concordance cosmology.” Should cosmologists be so high and mighty as to expect that there would be no deviations from the theory at all?

  • spaceman

    This is a puzzling paper; it is hard, for me at least, to know what they are trying to convey. It seems this paper is suggesting the Gum Nebula and a spot near the Galactic center are responsible for the low-l power that exists in the maps. What’s interesting is that when these spots are masked and the least contaminated portions of the sky are left untouched, the power is suppressed. So, the genuine primordial signiture portion of the sky (away from the galaxy) is telling us that there is vanishing power. But then why are the authors even tentatively concluding that there is no deficit of large scale power? Shouldn’t the main result of their analysis be telling them precisely the opposite–that is, the power that exists is in a region where foregrounds are greatest and so any large scale power is mostly likely NOT cosmological?

  • Garth

    When you try to correct the low-l power by adjusting for foreground contamination the problem gets worse.

    The deficiency is real, the question is: is it significant?


  • George Musser

    Michael Longo at U Mich claims to see the axis of evil in the orientations of spiral galaxies cataloged by Sloan (astro-ph/0703325) — what do you think?


  • Pingback: Philosophia Naturalis #8 « {metadatta}()

  • Pingback: Test Post « metadatta()

  • Pingback: The Lopsided Universe | Cosmic Variance()


Discover's Newsletter

Sign up to get the latest science news delivered weekly right to your inbox!

Cosmic Variance

Random samplings from a universe of ideas.

About Sean Carroll

Sean Carroll is a Senior Research Associate in the Department of Physics at the California Institute of Technology. His research interests include theoretical aspects of cosmology, field theory, and gravitation. His most recent book is The Particle at the End of the Universe, about the Large Hadron Collider and the search for the Higgs boson. Here are some of his favorite blog posts, home page, and email: carroll [at] .


See More

Collapse bottom bar