Szapudi reasons to drop alternatives to dark energy

[marcus]

http://space.newscientist.com/article/dn13975-dark-energy-imaged-in-best-detail-yet.html

Ever since accelerated expansion was observed in 1998 some people (like e.g. David Wiltshire) have been working hard to think up alternative explanations that could account for the observations without invoking a positive cosmo constant or a uniform dark energy density.

This is good, it is good to try out alternative explanations.

But finally we have a study that begins to look like it shoots down these alternatives, and that there really is a positive cosmo constant.

The author is Istvan Szapudi et al.
(Granett/Neyrinck/Szapudi)
Szapudi is at Uni Hawaii--Honolulu

I think the study is an extensive analysis of integrated Sachs-Wolfe effect on the CMB temperature variation. Other people have done this, I believe, but this time it is done more thoroughly. It sounds like it begins to persuade some of the alternative-seekers that dark energy is real.
http://arxiv.org/abs/0805.2974
Dark Energy Detected with Supervoids and Superclusters
Benjamin R. Granett, Mark C. Neyrinck, István Szapudi (IfA, Hawaii)
(Submitted on 20 May 2008)

"The observed apparent acceleration of the universe is usually attributed to negative pressure from a mysterious dark energy. This acceleration causes the gravitational potential to decay, heating or cooling photons traveling through crests or troughs of large-scale matter density fluctuations. This phenomenon, the late-time integrated Sachs-Wolfe (ISW) effect, has been detected, albeit at low significance, by cross-correlating various galaxy surveys with the Cosmic Microwave Background (CMB). Recently, the best evidence has come from the statistical combination of results from multiple correlated galaxy data sets. Here we show that vast structures identified in a galaxy survey project an image onto the CMB; stacking regions aligned with superclusters produces a hot spot, and supervoids, a cold spot. At over 4 sigma, this is the clearest evidence of the ISW effect to date. For the first time, our findings pin the effect to discrete structures. The ISW signal from supervoids and superclusters can be combined with other cosmological probes to constrain dark energy and cosmological parameters. In addition, our findings make it more plausible that the extreme Cold Spot and other anomalies in the CMB are caused by supervoids."

===sample quote from NewSci===
...A photon gains energy when it enters a dense region with enhanced gravity – such as a galaxy cluster – as though it is falling into a well. When it leaves the cluster and climbs back out of the gravitational well, it loses energy.

In a universe without dark energy, the energy gained and lost during the crossing would be equal and would cancel out. But in the presence of dark energy, the universe expands quickly enough to stretch the gravitational well while the photon is still inside. This makes the well shallower and easier for the photon to climb out.

That means that a photon traveling through a cluster gains more energy than it loses, giving it a little energy kick so that it creates a hotter spot than would be expected on images of the CMB. Similarly, a photon that has passed through a void would leave a cold spot.

It's tough to detect this effect because dark energy gives only a slight nudge to the temperature, which is easily swamped by the normal temperature variations seen in the CMB, says Szapudi.
Extreme density

To get around this, his team looked at regions of extremely high and extremely low density, where you would expect to see the biggest effect.

Using data from the Sloan Digital Sky Survey, they chose over 3000 superclusters of galaxies and 500 "supervoids" of relatively empty space, and they found that the regions did indeed tally with enhanced hot and cold spots in the CMB.

Other teams have reported signs of this effect in the past, but those have been open to alternative explanations, says Szapudi. By contrast, his calculations suggest that there is less than a 1 in 200,000 chance that the match up his team saw is down to anything other than dark energy...
==endquote==

 

[Wallace]

This is indeed a very good and important paper. Once the Planck satellite has done its mapping of the CMB we will really be able to know for certain, since we will have even more detail of the CMB anisotropies on very small angular scales. Until then though this seems pretty conclusive.

On the other hand this has been an issue that has been much debated for the last few years, and this paper uses effectively the same data sets that others have used to find no evidence for any significant ISW. Be prepared for some negative responses in the literature to this paper! (not that I'm suggesting there is anything wrong with it, but I suspect it will generate some controversy).

 

[marcus]

another paper by the same authors (Granett, Neyrinck, Szapudi)
it looks like a condensed version, to meet a page limit. they wanted one short enough that they could submit to Astophyscial Letters which I guess means 4 or 5 pages max.

http://arxiv.org/abs/0805.3695
An Imprint of Super-Structures on the Microwave Background due to the Integrated Sachs-Wolfe Effect
Benjamin R. Granett, Mark C. Neyrinck, István Szapudi (IfA, Hawaii)
5 pages, 1 figure
(Submitted on 25 May 2008)

"We measure hot and cold spots on the microwave background associated with supercluster and supervoid structures identified in the Sloan Digital Sky Survey Luminous Red Galaxy catalog. The mean temperature deviation is 9.6 +/- 2.2 microK. We interpret this as a detection of the late-time Integrated Sachs-Wolfe (ISW) effect, in which cosmic acceleration from dark energy causes gravitational potentials to decay, heating or cooling photons passing through density crests or troughs. In a flat universe, the linear ISW effect is a direct signal of dark energy. The statistical significance of our detection is over 4 sigma, making it the clearest detection to date using a single galaxy dataset. Moreover, our method produces a compelling visual image of the effect."

the earlier paper was 17 pages and had a big sky map.

 

[nrqed]

...

Other teams have reported signs of this effect in the past, but those have been open to alternative explanations, says Szapudi. By contrast, his calculations suggest that there is less than a 1 in 200,000 chance that the match up his team saw is down to anything other than dark energy...
==endquote==

I am confused by this last statement. What is ruled out, exactly? What is meant by "dark energy" versus theories without dark energy? The only way for me to make sense of the statement is if they meant "GR without a cosmological constant is ruled out compared to GR with a cosmological constant".

For example, is MOND included as the ruled out theories? What about other modified gravity theories?

Puzzling...

 

[Wallace]

Good question nrqed, I will try and respond.

If you allow the possibility of MOND and other modified gravity theories (TeVeS, DPG etc etc...) then what this result claims to rule out is a GR model dominated at all times after recombination by pressure less dust (i.e. baryonic and dark matter). This is because the ISW effect is caused by the potential well of a perturbation in the density field (either an under or over dense region) changing as a photon traverses the region. It turns out that for a matter dominated universe, the expansion of the universe and the growth of the perturbation exactly cancel each other in terms of their effect on the potential and the potential remains constant within the perturbation (at least in the linear growth regime).

In terms of GR, the only way for the potential to change is for the Universe to be dominated at late times by a component with significantly negative pressure, i.e. dark energy. In this case the potential does change as the photon traverses the potential well, leading to a net gravitational red or blue shift (since the gravitational blueshift caused 'going in' to the well is not balanced by the redshift 'going out') which we call the ISW or Rees-Sciama effect.

Now, modified gravity theories that mimic the existence of DE do in general predict an ISW effect, however the specifics of the effect in general will be different, you would expect a different kind of ISW signal. Since we a still arguing about whether we really see any ISW effect we obviously don't have the sensitivity to decide between these models at present. In the future though it is thought the ISW effect will be an important cosmological probe.

 

[Sanjay87]

This sounds like interesting stuff. One thing I'm unclear about: the article says that LRGs make good indicators of matter distribution. I don't really understand how.

 

[Wallace]

LRG's (Luminous Red Galaxies) are very bright and hence we can see them more easily than galaxies in general. Therefore they are convenient to use as tracers of the matter distribution. The tricky part is that they, like any galaxies, will be a biased tracer of the mass distribution. This is because LRG's, as the 'L' part implies, are large galaxies and hence tend to be found in the most dense environments. This means that they are likely to be more clustered than galaxies in general.

Using shallow surveys (low redshift) were we can easily see LRG's and other galaxies, we have been able to compare the clustering of LRG's and galaxies generally and understand this bias, which if you like calibrates LRG's as tracers of the mass distribution for use at higher redshifts when you can't see the dimmer more average galaxies as well as you can the LRG's.

 

[jonmtkisco]

Hi Wallace,

The methodology used for the ISW analysis isn't clear to me so I hope you can explain a bit more.

I can understand why supervoids are good subjects for this study, because they represent expanding regions which are highly transparent to CMB photons passing through them. But the relevance of superclusters is less clear to me. First, superclusters are gravitationally bound regions which presumably are currently contracting, not expanding. Therefore it would seem that the existence or nonexistence of dark energy inside a supercluster would have little or no noticeable effect on the expansion rate while a CMB photon traverses it. Second, there is the question of whether superclusters are suffused with dust and gas which would filter the traversing CMB photons in some way.

Is the nearby CMB radiation being measured that which actually passes through the supercluster, or just nearby? Why would CMB photons that pass near, but not through, a supervoid or supercluster closely model the behavior of CMB photons which actually pass through it?

I also question whether the results of this survey have a direct bearing on Wiltshire's theory. While I personally do not advocate Wiltshire's theory, it seems to me that it might well generate ISW effects which are similar to dark energy.

Jon

 

[Wallace]

Hi Wallace,

The methodology used for the ISW analysis isn't clear to me so I hope you can explain a bit more.

I can understand why supervoids are good subjects for this study, because they represent expanding regions which are highly transparent to CMB photons passing through them. But the relevance of superclusters is less clear to me. First, superclusters are gravitationally bound regions which presumably are currently contracting, not expanding. Therefore it would seem that the existence or nonexistence of dark energy inside a supercluster would have little or no noticeable effect on the expansion rate while a CMB photon traverses it.

Dark energy most certainly affects the rate at which overdense regions collapse. Remember that the superclusters they are talking about are large regions, not individual collapsed objects, so the density contrast to the background is not overwhelming. Dark energy can only be ignored once an individual halo has collapsed and virialised, which is something that happens on scales orders of magnitude smaller than the regions contributing to the ISW effect.

Second, there is the question of whether superclusters are suffused with dust and gas which would filter the traversing CMB photons in some way.

Again, on these scales this isn't a problem. However, on the scales of individual clusters there is an important secondary anisotropy in the CMB called the Sunyaev-Zeldovich (sp?) effect. Inverse Compton scattering by hot free electrons in the halo of clusters is the cause of this. This is expected to be a good way of finding clusters when the next generation CMB telescope, Planck, has finished its survey with greater angular resolution. The SZ effect makes much smaller 'blobs' on the CMB than the ISW.

Is the nearby CMB radiation being measured that which actually passes through the supercluster, or just nearby? Why would CMB photons that pass near, but not through, a supervoid or supercluster closely model the behavior of CMB photons which actually pass through it?

I don't understand this question? The point of the idea is that the photons that go through the under/over dense region get an additional red or blue shift that the 'nearby' photons do not.

I also question whether the results of this survey have a direct bearing on Wiltshire's theory. While I personally do not advocate Wiltshire's theory, it seems to me that it might well generate ISW effects which are similar to dark energy.

From memory I think he mentions this somewhere in the original paper, but I may be mistaken. However I struggle to see how it would work. The ISW effect can be explained from linear perturbation theory, since it deals with large scale regions of relativity mild under or over density. Wiltshire's theory aims more at the non-linear region, where local densities depart much more from the mean. This has nothing to do with the ISW effect. Linear perturbation theory has been shown to model the growth of structure on large scales very well, so it's hard to see how there is much wriggle room to question those results in the framework of Wiltshire's ideas.

 

[oldman]

...In terms of GR, the only way for the potential to change is for the Universe to be dominated at late times by a component with significantly negative pressure, i.e. dark energy. In this case the potential does change as the photon traverses the potential well, leading to a net gravitational red or blue shift (since the gravitational blueshift caused 'going in' to the well is not balanced by the redshift 'going out') which we call the ISW or Rees-Sciama effect...

I understand that dark energy is sometimes attributed in GR to a non-zero cosmological constant (lambda), which is (non-quantitatively) related to the zero-point energy of the vacuum. If this were a correct interpretation, would the results of Neyrinck et al., taken in isolation, require (a) that Lambda indeed has a non-zero value and (b) that this value changes with time (as implied by your phrases "for the potential to change" and "dominated at late times")? Or, if interpreted in a Newtonian context, would they provide evidence that G changes with time?

 

[Wallace]

No, a constant Lambda and constant G leads naturally to an early epoch of matter domination followed by domination by the Lambda term. This is due how the energy density of matter and lambda evolve. For matter, the density drops with the inverse of the volume of the Universe, i.e. if the universe doubles in volume the matter density halves. On the other hand, the lambda term has a constant energy density.

This means that at early times in the Universe, the average matter density is much higher than lambda, but as the Universe expands and dilutes the matter density the lambda term eventually becomes larger than the average matter density. This is the description of standard LCDM cosmology, in which lambda and G are both constant. This leads to an expected ISW effect.

 

[Garth]

No, a constant Lambda and constant G leads naturally to an early epoch of matter domination followed by domination by the Lambda term. This is due how the energy density of matter and lambda evolve. For matter, the density drops with the inverse of the volume of the Universe, i.e. if the universe doubles in volume the matter density halves. On the other hand, the lambda term has a constant energy density.

This means that at early times in the Universe, the average matter density is much higher than lambda, but as the Universe expands and dilutes the matter density the lambda term eventually becomes larger than the average matter density. This is the description of standard LCDM cosmology, in which lambda and G are both constant. This leads to an expected ISW effect.

This only leaves the interesting coincidence with a constant Lambda that [itex]\Omega_{Matter}[/itex] ~~ [itex]\Omega_{\Lambda}[/itex] in today's epoch, remembering the matter density has fallen from about 10⁹³ gm/cc (OOM) at Planck time to about 10^-30 gm/cc in the present epoch, i.e. over 120 Orders of Magnitude, while the (dark) energy density has remained constant!

Garth

 

[Wallace]

Co-incidence or not, the data keeps stacking up in favour of the model, the study under discussion in this thread being just one more example.

 

[jonmtkisco]

Thanks Wallace.

Dark energy most certainly affects the rate at which overdense regions collapse.

Aha, I was under the mistaken impression that large superclusters were gravitationally bound. Grannet, Neyrink & Szapudi say not so on p.1 of their 17 page paper, but their description is a bit confusing:

"... we identified the largest over- and under-densities in a galaxy catalogue. These
structures are still undergoing gravitational collapse and expansion, respectively. The
superclusters we detect are unlikely to be gravitationally bound, which distinguishes
them from their constituent clusters."

Obviously a structure can't be "still undergoing gravitational collapse" if it is not gravitationally bound.

The way the ISW is described in various technical papers makes clear that the superclusters they are talking about are actually expanding during the course of the CMB photons' passage. For example Giannantonio, Scranton, Crittenden et al on p.2 of their 5/6/08 paper say:

"In the usual case, the potential amplitudes decrease at late times, so that a temperature increase results from passing through potential wells, while a temperature deficits results from traversing potential hills."

So presumably little or no ISW effect would be detected if there were such a thing as a gravitationally bound supercluster.

Regarding Wiltshire's theory (which he has named the "Fractal Bubble" (FB) theory):

From memory I think he mentions this somewhere in the original paper, but I may be mistaken. However I struggle to see how it would work.

In his 12/07 paper Wiltshire says:

Detailed construction of the Doppler peaks, to enable comparison to WMAP, and determination of parameters such as σ8, is a matter of urgency. It is a rather non–trivial exercise in recalibration of standard quantities, in which all steps need to be carefully reconsidered. Detailed quantitative calculations of CMB ellipticity and the integrated Sachs–Wolfe effect can only be performed in conjunction with such an analysis.

While the calculation of the integrated Sachs–Wolfe (ISW) effect will differ in the FB model, one must be careful not to base intuition on that of perturbations on FLRW backgrounds. In particular, the observed ISW signal is believed to confirm dark energy since its consequence is a suppression of the gravitational collapse of matter at relatively recent times. If one replaces the words “dark energy” by “voids” in the standard qualitative explanation, then a probable description of the ISW effect in the FB model emerges. The same correlation of the ISW signal with clumped structures is expected; what is important is the magnitude of the effect. Mattsson has recently proposed an extension of the Dyer–Roeder formalism, which may have some relevance for such calculations.

Jon

 

[Garth]

Co-incidence or not, the data keeps stacking up in favour of the model, the study under discussion in this thread being just one more example.

The study under discussion here reports on the alignment of superclusters and supervoids with CMB hot and cold spots respectively, attributed to the ISW effect, thus validating some kind of DE.

(As a side note the existence of such large structures at high z might introduce an age problem in the early universe.)

The astounding coincidence that I have mentioned surely tells us that DE evolves in some sense so that [itex]\Lambda[/itex] is almost certainly not constant, unless an anthropic argument is invoked.

The problem is discussed in this (now rather old) paper: A Dynamical Solution to the Problem of a Small Cosmological Constant and Late-time Cosmic Acceleration.

Garth

 

[oldman]

No, a constant Lambda and constant G leads naturally to an early epoch of matter domination followed by domination by the Lambda term... This is the description of standard LCDM cosmology, in which lambda and G are both constant. This leads to an expected ISW effect.

Wallace, thanks for this simple explanation, which does answer my first question.

But could you, or maybe someone else, perhaps respond as simply to the second question I asked, which was: "if interpreted in a Newtonian context, would (the observations under discussion here) provide evidence that G changes with time?"

I'm not suggesting that GR should be replaced by old physics. But Newtonian gravity often provides an easiy understood description, in broad outline, of what is happening. Such an example is: "The expansion of the universe slows down, like a stone thrown upwards that is being retarded by gravity" . (Which of course leaves the late-time acceleration unexplained)

Such explanations are very unsophisticated and sometimes inaccurate, but they often clarify the bare bones of what is happening. One should not forget that gravity remains a mysterious phenomenon which is only described rather than fully understood, even using the accurate and beautiful grammar of GR. A simple perspective is sometimes useful. Therefore I ask again:

"if interpreted in a Newtonian context, would (the observations under discussion here) provide evidence that G changes with time?"

 

[Wallace]

Good question, and of the top of my head I guess dark energy and a changing G could look the same, but only in the ideal homogeneous model for the Universe. In other words, you might be able to get the global expansion rates to agree for a dark energy and a varying G model.

However, in the inhomogeneous Universe the two models would look different I think. For instance, the orbits of the planets around the Sun almost completely unaffected by dark energy, since on the scale of the Milky Way matter dominates over dark energy even if globally dark energy is dominant. What this means is that once an object such as the Milky Way forms the global expansion rate ceases to makes any real difference to the internal dynamics. In the case of varying G, clearly this would make a difference.

Again I'm not sure of the top of my head how different structure formation would be in varying G vs dark energy models, but I feel it should be big enough to be detectable in the current data. I'm sure there must be some papers written on this in the literature somewhere...

The only alternative would be if G was variable and scale dependent (i.e. G becomes a function of distance). Such a model might be able to be tuned to completely spoof a dark energy model but then it would be terribly fine-tuned by comparison.

 

[jonmtkisco]

The way the ISW is described in various technical papers makes clear that the superclusters they are talking about are actually expanding during the course of the CMB photons' passage.

It seems to me that if superclusters are expanding, then their expansion rate must be accelerating, with a net acceleration rate less than or equal to the cosmic Hubble rate. As superclusters expand, they incorporate increasingly more dark energy over time while simultaneously decreasing their internal gravitational density. So their acceleration rate increases over time, eventually asymptotically approaching the cosmic Hubble rate.

It is interesting to think about whether the overdensity of such superclusters, compared to the average cosmic density, has been increasing or decreasing in the past. Presumably both: At the time of last scattering their overdensity was minute, then it probably increased by multiple orders of magnitude during the linear phase, and now it is decreasing again due to the growing influence of dark energy with scale.

My impression is that superclusters formed "top down" from initially overdense regions. As compared to galaxies and galaxy clusters, which seem to have formed "bottoms up" as very local density concentrations lead to star formation.

Jon