[astro-ph/0506478] Evidence for Evolution or Bias in Host Extinctions of High Redshift Supernovae

Authors:  Pankaj Jain, John P. Ralston
Abstract:  High redshift supernova magnitudes of current interest are re-examined. The magnitudes conventionally reported include an additive parameter called the extinction coefficient. We find that the extinction coefficients of a popular ``gold'' set are well correlated with the deviation of magnitudes from standard Hubble diagrams. The statistical significance of the {\it extinction-acceleration correlation} has a random chance probability of less than one in a million. The sense of correlation tends to produce dimming. We discuss related features of the data and potential explanations for the phenomenon including evolution and bias. The hypothesis that extinction coefficients should be corrected empirically provides substantially improved fits to both accelerating and non-accelerating models, with the independent feature of eliminating any significant correlation of residuals.
[PDF]  [PS]  [BibTex]  [Bookmark]

Discussion related to specific recent arXiv papers
Post Reply
Syksy Rasanen
Posts: 119
Joined: March 02 2005
Affiliation: University of Helsinki

[astro-ph/0506478] Evidence for Evolution or Bias in Host Ex

Post by Syksy Rasanen » July 02 2005

The authors look at the correlation between the residuals (which I guess means observed luminosity distance (in log units) minus the distance in the model) and the extinction A_V (i.e. the amount the observed magnitude is corrected for because of absorption etc. on the way) in the 'gold' set of SNIa.

They find a correlation that is improbable at the level of 4.2\times 10^{-7} for a random sample. Introducing an extra parameter to phenomenologically correct for the correlation makes the nearby SN dimmer and so reduces the case for acceleration. The difference in \chi^2 between the 'concordance model' and a freely coasting empty universe drops from 13.5 to 6.5; the difference between the concordance model and an \Omega_m=1 universe is still \Delta\chi^2=138.6. Interestingly, this correction also reduces the difference between the best-fit assuming \Omega_{tot}=1 and leaving \Omega_{tot} unconstrained.

They also discuss a phenomenological fit for evolution of extinction with luminosity distance. As could be expected, this reduces the case for acceleration further, to the extent that it's not possible to differentiate between the models. The difference between the concordance model and the freely coasting case increases to \Delta\chi^2=4.2. More dramatically, the difference between the concordance model and an \Omega_m=1 universe drops to \Delta\chi^2=1.3. (They present no physical basis for the evolution.)

I am not familiar with these issues. I would imagined that such correlations have been checked. Is the paper correct? Has evolution of extinction (as opposed to evolution of the SNae) been ruled out?

Håvard Alnes
Posts: 4
Joined: January 20 2005
Affiliation: University of Oslo

[astro-ph/0506478] Evidence for Evolution or Bias in Host Ex

Post by Håvard Alnes » July 04 2005

I suppose they still would have problems explaining the position of the CMB peaks in a [tex]\Omega_m = 1[/tex] scenario?

Anze Slosar
Posts: 183
Joined: September 24 2004
Affiliation: Brookhaven National Laboratory
Contact:

Re: [astro-ph/0506478] Evidence for Evolution or Bias in Hos

Post by Anze Slosar » July 04 2005

Håvard Alnes wrote:I suppose they still would have problems explaining the position of the CMB peaks in a [tex]\Omega_m = 1[/tex] scenario?
Well, peaks in CMB alone are ok, as long as you have a stupid H (42!). Basicall, for \Omega_\Lambda you need CMB + something else (HST, LSS, etc.)

Sarah Bridle
Posts: 144
Joined: September 24 2004
Affiliation: University College London (UCL)
Contact:

[astro-ph/0506478] Evidence for Evolution or Bias in Host Ex

Post by Sarah Bridle » July 04 2005

It is difficult to reconcile other cosmological probes with [tex]\Omega_m=1[/tex]. But it is important to understand how seriously we should take supernova data as part of this evidence. Really just echoing Sysky here, but still:

They do find a correlation between residuals and extinction which I had not heard of before. Has anyone else ever reported this correlation? I see that it really only comes from ~10 data points, but still looks slightly worrying.

I would be interested to see 2d contour plots in [tex]\Omega_{\Lambda} [/tex]versus [tex]\Omega_m [/tex]with and without this correlation taken out (using their equation 6, [tex]\delta=0[/tex] cf [tex]\delta= [/tex]their best fit).

Similarly it would be interesting to see contours with and without supernovae with [tex]A_v>0.5[/tex] since these seem to be the only supernovae that drive the correlation. There are only a few supernovae here, so if it were these few supernovae that were driving any of the results then this would be a concern, no matter the interpretation.

I would also be interested to know what supernova experts think of this paper's proposed extinction evolution equation (eqn 7):
[tex] A_v(d_L) = (1+\delta) A_v + \delta_1 \log (d_L)[/tex]
as this seems quite radical (affecting all supernovae, instead of just those with high [tex]A_v[/tex]). Note that this paper does not provide any evidence to suggest this equation is right. It is just an extension of their finding: extinction corrections might not always be right. As Sysky says, no surprise that such a big change affects the conclusions on cosmology.

Syksy Rasanen
Posts: 119
Joined: March 02 2005
Affiliation: University of Helsinki

Re: [astro-ph/0506478] Evidence for Evolution or Bias in Hos

Post by Syksy Rasanen » July 04 2005

Anze Slosar wrote:Well, peaks in CMB alone are ok, as long as you have a stupid H (42!). Basicall, for \Omega_\Lambda you need CMB + something else (HST, LSS, etc.)
CMB+LSS is consistent with \Omega_m=1 (see astro-ph/0304237 or hep-ph/0503271). The things which are difficult to explain are

1) local measurements of the matter density, which give [tex]\Omega_m \lesssim 0.3[/tex] (some giving smaller values than 0.3),
2) the supernova data and
3) measurements of H_0 combined with the age of the universe. \Omega_m=1requires h=0.46, whereas the HST 2\sigma lower limit is 0.56 (though see the above papers on this issue).

I don't know about baryon oscillations. At what statistical significance can one say something about [tex]\Omega_m[/tex] from the new SDSS and 2dF data?

At any rate, echoing Sarah in turn, in order to claim 'concordance', one has to analyse each piece of evidence separately to see if they really agree. If one assumes data A when analysing data B, all one can show is that they are not inconsistent, not that they point to the same conclusions.

Anze Slosar
Posts: 183
Joined: September 24 2004
Affiliation: Brookhaven National Laboratory
Contact:

Re: [astro-ph/0506478] Evidence for Evolution or Bias in Hos

Post by Anze Slosar » July 08 2005

Syksy Rasanen wrote:CMB+LSS is consistent with \Omega_m=1 (see astro-ph/0304237 or hep-ph/0503271).
Well, Sean Carol just claimed the opposite here in Paris, but in principle I can buy that...
Syksy Rasanen wrote: I don't know about baryon oscillations. At what statistical significance can one say something about [tex]\Omega_m[/tex] from the new SDSS and 2dF data?

The following is taken from Eisenstein et al.:
This distance ratio is consistent with the familiar cosmological constant cosmology. It is grossly inconsistent with the Einstein-de Sitter (\Omega_m = 1) model, which predicts R_{0.35} = 0.133 (nominally 10 σ). A model lacking dark energy would require \Omega_m = 0.70 with \Omega_K = 0.30 to match the distance ratio. This would require h = 0.90 and \Omega_{m}h^2 = 0.57 to match the CMB peak location, implying an age of 8 Gyr. This is in complete disagreement with the observed shape of the CMB anisotropy spectrum, the galaxy correlation function (including these LRG data), the cluster baryon fraction (White et al. 1993),the observed value of H0 (Freedman et al. 2001), and the age of old stars (Krauss & Chaboyer 2003, and references therein), as well as other cosmological measurements. Hence, our measurement provides geometric evidence for dark energy.

Alessandro Melchiorri
Posts: 129
Joined: September 24 2004
Affiliation: University of Rome
Contact:

[astro-ph/0506478] Evidence for Evolution or Bias in Host Ex

Post by Alessandro Melchiorri » July 09 2005

Ehi, what about the recent ISW detections by cross correlating WMAP and LSS ?

Pablo Fosalba
Posts: 29
Joined: September 25 2004
Affiliation: Institut d'Estudis Espacials de Catalunya
Contact:

[astro-ph/0506478] Evidence for Evolution or Bias in Host Ex

Post by Pablo Fosalba » July 09 2005

I think the short answer is that evidence for ISW detection is
at the 3-\sigma level from independent analyses. The interpretation of this is that, either the universe is largely NOT flat with the same significance,
what is ruled out by CMB data, or \Omega_{m} < 1 and \Omega_{\Lambda} > 0 (at 3-\sigma again).

Anze Slosar
Posts: 183
Joined: September 24 2004
Affiliation: Brookhaven National Laboratory
Contact:

Re: [astro-ph/0506478] Evidence for Evolution or Bias in Hos

Post by Anze Slosar » July 10 2005

Pablo Fosalba wrote:I think the short answer is that evidence for ISW detection is
at the 3-\sigma level from independent analyses. The interpretation of this is that, either the universe is largely NOT flat with the same significance,
what is ruled out by CMB data, or \Omega_{m} < 1 and \Omega_{\Lambda} > 0 (at 3-\sigma again).
There is actually an interesting theoretical bound on ISW: detection will never be larger than 5-7 sigma due to primordial "noise".

Syksy Rasanen
Posts: 119
Joined: March 02 2005
Affiliation: University of Helsinki

Re: [astro-ph/0506478] Evidence for Evolution or Bias in Hos

Post by Syksy Rasanen » July 11 2005

Anze Slosar wrote:
Syksy Rasanen wrote:CMB+LSS is consistent with \Omega_m=1 (see astro-ph/0304237 or hep-ph/0503271).
Well, Sean Carol just claimed the opposite here in Paris, but in principle I can buy that...
If you assume that the primordial perturbations are adiabatic with a power-law spectrum, then you can rule out \Omega_m=1 from the CMB alone. If you allow for features in the spectrum (see e.g. astro-ph/0408138), then this is no longer the case. (I'm not sure whether the same is true of isocurvature. This obviously depends on the form and correlation taken for the adiabatic and isocurvature contributions.)

Alessandro Melchiorri
Posts: 129
Joined: September 24 2004
Affiliation: University of Rome
Contact:

[astro-ph/0506478] Evidence for Evolution or Bias in Host Ex

Post by Alessandro Melchiorri » July 12 2005

Ok so:

- There are features in the primordial spectrum that are mimicking Lambda.
- There are neutrinos or quintessence just to give LSS (galaxies, Ly-alpha, clusters) similar to LCDM.
- There are systematics in the determination of the age of the universe that are giving an higher age (just as Lambda).
- There are systematics in the determination of the Hubble constant which are giving larger values of h_0.
- There are systematics in the measurements of the matter density giving a smaller Omega_m.
- There are systematics in the SN-Ia that are mimicking Lambda.
- Again systematics in the determination of the ISW from cross CMB-LSS
- There are systematics in (add here your current favorite dataset) that are mimicking Lambda....

So or the interpretation of all current datasets is wrong or there is Dark Energy.
We may discuss if there is a cosmological constant or something different but as far as I know there is not a single dataset pointing towards Omega_m=1 !

Syksy Rasanen
Posts: 119
Joined: March 02 2005
Affiliation: University of Helsinki

Re: [astro-ph/0506478] Evidence for Evolution or Bias in Hos

Post by Syksy Rasanen » July 16 2005

I don't agree with your list of things which should be reinterpreted for observations to be consistent with an Einstein-de Sitter universe (nor the claim that systematics are the only resort), but I guess you were trying to be more polemical than accurate...

I agree that there is strong evidence for some form of non-clustering matter or modification of gravity on large scales. I don't think that it is contradictory to have the stand that in order to see where individual datasets are pointing, they should be analysed independently, and that one should be aware of the conditions under which given conclusions are drawn from the data.

Dragan Huterer
Posts: 27
Joined: July 18 2005
Affiliation: University of Michigan
Contact:

[astro-ph/0506478] Evidence for Evolution or Bias in Host Ex

Post by Dragan Huterer » July 18 2005

I have to say I fully agree with Alessandro here. The list is actually even longer since there are many individual items under "measurements of Omega_M", and most of them are pretty convincing.

For example, type Ia SNa measurements would need to be off by something like 0.5 mag or more, and in a very specific way with redshift, to accomodate for EdS. There is no known mechanism that works and can give this much extinction. But even if there was one, the one would still need an extremely low Hubble constant (to accomodate, say, measurements of the bend in the matter power spectrum and also CMB peak morphology). There is no controversy here.

As for the XMM paper, I think such measurements are still in relatively early stages and suffer from both theoretical and experimental systematics (but others may well disagree here). In particular, I am worried about their purely empirical fit to the L-T luminosity relation (how many parameters do you assign?), and also about our understanding of what the measured "L" and "T" really are.

Post Reply