Acceleration of expansion as unknown reddening bias

[zonde]

Does anybody know if there is some analysis of type Ia supernovae data for possibility of unknown intergalactic reddening bias that would result in apparent acceleration of expansion.

Naturally such unknown reddening bias would be function of redshift and it could spoil calculated luminosity distances for SNs with different redshifts.

I found that in this paper http://arxiv.org/abs/astro-ph/0510155" there is mentioned some unknown reddening but it's unclear for me how significant is this anomaly.

 

[Vanadium 50]

First, redshift is not the same as reddening. The former is a shift in spectral lines, the latter in overall color (which can be affected by the medium).

Second, I didn't see anything about "unknown reddening" in the paper, but even if there is some, it's not really relevant, because redshift is the determiner of distance.

 

[Chalnoth]

Does anybody know if there is some analysis of type Ia supernovae data for possibility of unknown intergalactic reddening bias that would result in apparent acceleration of expansion.

Naturally such unknown reddening bias would be function of redshift and it could spoil calculated luminosity distances for SNs with different redshifts.

I found that in this paper http://arxiv.org/abs/astro-ph/0510155" there is mentioned some unknown reddening but it's unclear for me how significant is this anomaly.

Nearly all of the reddening of supernovae is due to dust in the host galaxy. This reddening has a very distinct spectral signature (it looks nothing like a redshift, but is instead a suppression of the higher-frequency part of the spectrum).

Anyway, as I mentioned in the other thread, observations of very distant supernovae put the nail in this coffin:
http://iopscience.iop.org/0004-637X/607/2/665

Evidence of past deceleration from observing these very distant supernovae demonstrates that the appearance of acceleration is not just a matter of something like dust.

 

[zonde]

First, redshift is not the same as reddening.

Yes

The former is a shift in spectral lines, the latter in overall color (which can be affected by the medium).

Reddening is higher extinction at shorter wavelengths than at longer.

Second, I didn't see anything about "unknown reddening" in the paper,

Look at p.11
"Figures 6 to 10, in addition, detail the basic elements used to match the observed spectra. These elements are the spectrum of a nearby SN Ia at early phases and the spectrum of a galaxy from the catalog of Kennicutt (1992), but we found it necessary to add an additional slope to make its continuum redder. All of the observed spectra appear too red to match the model of a SN contaminated by a galaxy."

A bit later they say:
"We have convinced ourselves that the spurious slopes are not intrinsic to the spectra, but unfortunately we cannot offer a reasonable explanation."

but even if there is some, it's not really relevant, because redshift is the determiner of distance.

Redshift is proportional to distance only if expansion rate is linear. However if expansion rate is not linear or even not smooth then you can not determine distance only from redshift.
Redshift gives you separation speed between (past) source and receiver at the moment of measurement. If separation speed was varying you have no idea about that from redshift.

Therefore we need standard candles i.e. type Ia SN.
That's because luminosity depends from redshift and distance. You can imagine it that way - all the light from source is distributed over spherical surface after it traveled some distance R (or simply appears at distance R). Therefore intensity of light will drop proportionally to square of R if we observe portion of surface with fixed size.
Redshift however affects luminosity in two different ways. First because of redshift energy of photons drops linearly so the same goes for luminosity. Second separation speed linearly affects photon to photon frequency as well so we detect smaller number of photons and have liner drop for luminosity.

So we have that luminosity drop is proportional to square of distance times square of separation speed. Knowing separation speed (redshift) and luminosity drop we can calculate distance. Even if separation speed have nonlinear history.

But this is so if we can neglect extinction. Obviously extinction is still another factor that affects luminosity.

 

[zonde]

Nearly all of the reddening of supernovae is due to dust in the host galaxy. This reddening has a very distinct spectral signature (it looks nothing like a redshift, but is instead a suppression of the higher-frequency part of the spectrum).

Anyway, as I mentioned in the other thread, observations of very distant supernovae put the nail in this coffin:
http://iopscience.iop.org/0004-637X/607/2/665

Evidence of past deceleration from observing these very distant supernovae demonstrates that the appearance of acceleration is not just a matter of something like dust.

Please understand that I am not suggesting that reddening can be mixed for redshift.
I am saying that if there is unknown reddening bias i.e. extinction of the higher-frequency part of the spectrum, then luminosity distances would be calculated incorrectly.
To be more specific luminosity drop due to unknown reddening will affect nearby SN more that distant SN if we divide it proportionally to distance. Thats because distant SN after they spectra have shifted to long wavelengths will not be affected by reddening bias any more and as you divide luminosity drop by distance the effect of bias will diminish.

So qualitatively it produces exactly the same effect as recent acceleration of expansion.

 

[Chalnoth]

Please understand that I am not suggesting that reddening can be mixed for redshift.
I am saying that if there is unknown reddening bias i.e. extinction of the higher-frequency part of the spectrum, then luminosity distances would be calculated incorrectly.
To be more specific luminosity drop due to unknown reddening will affect nearby SN more that distant SN if we divide it proportionally to distance. Thats because distant SN after they spectra have shifted to long wavelengths will not be affected by reddening bias any more and as you divide luminosity drop by distance the effect of bias will diminish.

So qualitatively it produces exactly the same effect as recent acceleration of expansion.

a) Reddening is visible in the spectrum.
b) As I mentioned, further-away supernovae appear brighter than a "coasting" universe, which rules out this sort of explanation for the appearance of acceleration.

 

[zonde]

a) Reddening is visible in the spectrum.

Well, not exactly. You can see in spectrum characteristic features due to absorption and emission because it involves specific frequencies. But reddening is generally due to scattering. And scattering is rather not specific to frequency. Only large differences in frequencies change result of scattering.

So usually you compare observed spectrum in different bands for some object with spectrum of similar object that you believe is not reddened.
As you can see result depends from quality of your reference spectrum and intrinsic similarity between your observed object and reference object.

b) As I mentioned, further-away supernovae appear brighter than a "coasting" universe, which rules out this sort of explanation for the appearance of acceleration.

Estimating luminosity distances for supernovae is quite complicated process. Here it again depends heavily from you reference objects. Let's say if you overestimate luminosity distance of nearby supernovae because there is unaccounted luminosity drop you can get too much luminosity for another distant object when you compare the two if this other object is not affected by that unaccounted luminosity drop (because of redshift).

And I am not trying to justify "coasting" universe. I just don't buy jerking universe where this jerking is synchronized across the whole universe (or considerable portion of it). Something like that would require very contrived physical mechanism.

 

[Chalnoth]

Well, not exactly. You can see in spectrum characteristic features due to absorption and emission because it involves specific frequencies. But reddening is generally due to scattering. And scattering is rather not specific to frequency. Only large differences in frequencies change result of scattering.

So usually you compare observed spectrum in different bands for some object with spectrum of similar object that you believe is not reddened.
As you can see result depends from quality of your reference spectrum and intrinsic similarity between your observed object and reference object.

Yes, but to get a significant bias you'd need far-away supernovae to be intrinsically different from more nearby ones in such a way that it looks like more or less redshift (depending on distance). There's no reason to believe this is so at present.

Estimating luminosity distances for supernovae is quite complicated process. Here it again depends heavily from you reference objects. Let's say if you overestimate luminosity distance of nearby supernovae because there is unaccounted luminosity drop you can get too much luminosity for another distant object when you compare the two if this other object is not affected by that unaccounted luminosity drop (because of redshift).

And I am not trying to justify "coasting" universe. I just don't buy jerking universe where this jerking is synchronized across the whole universe (or considerable portion of it). Something like that would require very contrived physical mechanism.

Um, a small but nonzero cosmological constant predicts a universe that first decelerates, then accelerates.

 

[zonde]

Yes, but to get a significant bias you'd need far-away supernovae to be intrinsically different from more nearby ones in such a way that it looks like more or less redshift (depending on distance). There's no reason to believe this is so at present.

I don't see your point.
I am talking about biased luminosity distances not about biased redshift.
Can you explain your point a bit?

Um, a small but nonzero cosmological constant predicts a universe that first decelerates, then accelerates.

Fine. What about "flatness" of that change in acceleration/deceleration across the universe?

 

[Chalnoth]

I don't see your point.
I am talking about biased luminosity distances not about biased redshift.
Can you explain your point a bit?

You're right, I misspoke. It's about a bias in luminosity distance.

Fine. What about "flatness" of that change in acceleration/deceleration across the universe?

I don't understand your point here. First of all, the degree of expansion just isn't that well-measured to even determine if there is some spatial variation in the expansion rate. At least, not yet.

But the reason why it works this way in the standard model is because the expansion rate is density driven, and on large scales our universe is nearly the same density everywhere.

 

[zonde]

But the reason why it works this way in the standard model is because the expansion rate is density driven, and on large scales our universe is nearly the same density everywhere.

Expansion by itself changes density. So if expansion is density driven then it can kind of "feed itself". So what prevented it from accelerating long time ago at small scale where density is not even?

On the other hand if you implement some mechanism that prevents it from getting out of line at small scale if should be effected with speed of light limit in mind. That in turn would make any changes in behavior of expansion really crawling.

 

[Chalnoth]

Expansion by itself changes density. So if expansion is density driven then it can kind of "feed itself". So what prevented it from accelerating long time ago at small scale where density is not even?

Well, to get what happens when you add in the fact that the universe isn't perfectly smooth, you basically have to use computer simulations. But in the end, the evidence seems to be that for the most part, the overall expansion just depends upon the average density. You'd have to have some pretty severe inhomogeneities on large scales for there to be any significant difference.

 

[zonde]

Well, to get what happens when you add in the fact that the universe isn't perfectly smooth, you basically have to use computer simulations.

And can such simulations model accelerating expansion?

But in the end, the evidence seems to be that for the most part, the overall expansion just depends upon the average density.

About what kind of evidence are you talking?

 

[Chalnoth]

And can such simulations model accelerating expansion?

Yes.

About what kind of evidence are you talking?

Supernovae, baryon acoustic oscillations, and the CMB, primarily. They all fit the standard model, but have difficulty fitting cosmologies that, for instance, try to produce acceleration due to us living in a large void.

 

[zonde]

Yes.

Can you provide a reference?
I have tried to look myself and couldn't find anything even close to that. I suspect that you are making empty statements.

Supernovae, baryon acoustic oscillations, and the CMB, primarily. They all fit the standard model, but have difficulty fitting cosmologies that, for instance, try to produce acceleration due to us living in a large void.

This too asks for reference.
There is simply not enough supernovae data to make some spatial expansion maps.
And it is not clear how CMB can provide means to correlate density with expansion.

Not to mention general problem that most part of energy density is only hypothetical and independently from expansion is not observable at all.
Another logical problem is that if you intend to correlate two quantities with zero variations you can not speak about any correlation at all.

 

[Chalnoth]

Can you provide a reference?
I have tried to look myself and couldn't find anything even close to that. I suspect that you are making empty statements.

It's a well-known result that you can't get the appearance of accelerated expansion without us sitting very near the center of an extremely large void. I don't understand the difficulty here.

This too asks for reference.
There is simply not enough supernovae data to make some spatial expansion maps.
And it is not clear how CMB can provide means to correlate density with expansion.

Not to mention general problem that most part of energy density is only hypothetical and independently from expansion is not observable at all.
Another logical problem is that if you intend to correlate two quantities with zero variations you can not speak about any correlation at all.

See here:
http://arxiv.org/abs/1007.3725

Basically, it's not anyone data set, but the combination of them that demonstrates this. Supernovae alone, for instance, do not constrain cosmological parameters very well. But they do so in a way that is very different from other probes like Baryon Acoustic Oscillations, such that the combination of BAO and supernovae is much, much better than BAO alone.

 

[zonde]

It's a well-known result that you can't get the appearance of accelerated expansion without us sitting very near the center of an extremely large void. I don't understand the difficulty here.

You are saying that we can't have accelerated expansion without void hypothesis and in the link you gave it is said that data does not support that hypothesis.
What I am supposed to conclude? That you are saying - accelerated expansion is ruled out?
Previously you was stating exactly the opposite. I am confused.

Anyways the question was about computer simulation of accelerated expansion that takes into account spatially uneven density at small scale.
You said yes that's possible.
Can you provide reference for such simulation?

See here:
http://arxiv.org/abs/1007.3725

Basically, it's not anyone data set, but the combination of them that demonstrates this. Supernovae alone, for instance, do not constrain cosmological parameters very well. But they do so in a way that is very different from other probes like Baryon Acoustic Oscillations, such that the combination of BAO and supernovae is much, much better than BAO alone.

Again we were not talking about some large void hypothesis but about evidence that "the overall expansion just depends upon the average density".
To gather some evidence for correlation between expansion and density we would have to get some results from research like the one proposed in this paper:
http://arxiv.org/abs/0812.0376"

From paper:
"For comparison, in Fig. 1 we also plot the shot-noise for two potential SN surveys (dashed lines): 3,000 SNe in 10 deg^2, as can be achieved with deep repeated observations from space, and 10^6 SNe in 20,000 deg^2, from a shallow, but wide, SN survey."

Obviously right now there is no such data. Such data however can become available after http://snap.lbl.gov/science/detailed.php" is completed.

If you think otherwise please provide reference.

 

[Chalnoth]

You are saying that we can't have accelerated expansion without void hypothesis and in the link you gave it is said that data does not support that hypothesis.
What I am supposed to conclude? That you are saying - accelerated expansion is ruled out?

Sorry that I wasn't more clear. I'm saying that the appearance of acceleration being due to the universe not being uniform is basically ruled out. This is evidence for there being some form of modified gravity or dark energy (with dark energy typically preferred by theory, because it's very difficult to modify gravity on large scales without leaving observable effects at short scales).

Anyways the question was about computer simulation of accelerated expansion that takes into account spatially uneven density at small scale.
You said yes that's possible.
Can you provide reference for such simulation?
Again we were not talking about some large void hypothesis but about evidence that "the overall expansion just depends upon the average density".

Well, you'd have to go into the depths of one of the large-scale n-body simulations to see this, such as the Millennium Run. Alternatively, from the theory side, you could look into some of the spherically-symmetric inhomogeneous universe models, or the swiss cheese universe.