Observational evidence against expanding universe in MNRAS

[elerner]

<< Mentor Note -- after a very long Mentor discussion, we acknowledge that this paper, while potentially controversial, has been published in a reputable peer-reviewed journal. We believe that a discussion of this paper can be useful and constructive. Thanks >>

This paper of mine was published online in Monthly Review of the Royal Astronomical Society.
The title is a good summary: "Observations contradict galaxy size and surface brightness predictions that are based on the expanding universe hypothesis".
For a non-technical description, see our press release here.

 

[rootone]

Interesting, but is this implying that Fred Hoyle's idea was right?
That the Universe is static and eternal (mostly), but we have not yet discovered how light (EM generally) behaves on the cosmic scale?

 

[elerner]

No.Hoyle hypothesized that the universe did not evolve. There is plenty of evidence that it does change over time. But evolution does not require expansion--the Earth we live on has been evolving at an accelerating pace but is not expanding. The universe too can be evolving in an accelerating way without either having an origin in time or undergoing expansion. The redshift-distance relationship does imply something we don't yet understand is draining energy from the EM radiation as it travels long distances.

 

[PeterDonis]

evolution does not require expansion--the Earth we live on has been evolving at an accelerating pace but is not expanding

The Earth is a gravitationally bound system so I'm not sure it can be usefully compared to the universe as a whole.

Can you describe what sort of model you envision for the universe as a whole evolving but not expanding? I'm particularly curious as to whether you expect such a model to be consistent with the Einstein Field Equation, or whether you think some new physics beyond that will be needed.

 

[elerner]

The evolutionary processes occurring on Earth have changed it a lot, but have nothing to do with changes in the structure of space. Similarly, in the universe on large scales, gravitational, electromagnetic and nuclear processes can change it over time, independently of any expansion. You don't need expansion to form galaxies, stars, or planets or to burn hydrogen to helium in fusion.

I don't think the second question can be answered right now. One possibility is that matter is distributed fractally at all scales with a dimension of 2 or less. In that case, GR effects would be small on all scales. (In other words, if the density keeps dropping the curvature of space is negligible at all scales. In that case GR does not predict any expansion or contraction.) A second possibility is that gravitation weakens at very large distances. If some process drains photons of energy over large distances, it may well weaken EM forces and perhaps gravitation as well. Again, that would result in no overall expansion or contraction.

Unlike some of my colleagues, I don't claim to have a Theory of Everything. Observations are the test of theories. If the predictions of the theory don't hold up, then either the theory needs to be changed, or the assumptions, like homogeneity, that go into the predictions need to be changed. To figure out what needs to be changed requires more work. The first step is to decide that predictions are wrong. What this paper is saying is that the predictions based on expansion don't work, at least not for this data set. What works are predictions based on the hypothesis of no expansion and a linear relation between z and distance.

 

[rootone]

It sounds a bit like a 'tired light' hypothesis
Not utterly implausible, but it requires an undefined parameter when there is no reason or evidence to assume it.

 

[elerner]

If tired light means that something happens to the light over long distances, yes. But no new parameter is needed, just the same old Hubble constant. With just the Hubble constant, the linear, no-expansion hypothesis fits the supernova data as well as LCDM does with three parameters (Hubble, dark matter, dark energy). No parameters at all are needed for the linear hypothesis to fit the galaxy size/surface brightness data. The value of the Hubble constant does not matter for that prediction of no change--there are no free parameters at all. And that data set can't be fit with LCDM hypotheses without violating other observational constraints, as I show in the paper.

 

[PeterDonis]

You don't need expansion to form galaxies, stars, or planets or to burn hydrogen to helium in fusion.

That's true, but I don't think this observation counts for much either way. It just means those particular phenomena, in themselves, aren't relevant to the question of whether the universe is expanding or not.

One possibility is that matter is distributed fractally at all scales with a dimension of 2 or less.

I don't understand what this means. The universe has some average density of matter. That isn't a "fractal", it's just an average density. Unless you are claiming that the average density is zero, which basically means all the matter we can see is a finite "island" of matter surrounded by an infinite expanse of empty space. Is that what this hypothesis is referring to?

if the density keeps dropping the curvature of space is negligible at all scales

The relevant curvature is the curvature of spacetime, not space. Yes, a completely empty universe with zero density (and zero cosmological constant) is just Minkowski spacetime, which is not "expanding" and has zero spacetime curvature everywhere. Is this the model you are suggesting?

A second possibility is that gravitation weakens at very large distances.

Ok, this falls into the second category I mentioned: new physics that is not consistent with the Einstein Field Equation. See below.

that would result in no overall expansion or contraction.

But the only way we have of making any predictions at all regarding overall expansion or contraction, or lack thereof, is by means of the Einstein Field Equation. And if you're hypothesizing that that no longer works for the universe as a whole, you have no way of making any predictions at all, unless you propose some specific alternate model for how gravity weakens at large distances, how photons lose energy, etc. I'm not saying that can't be done; but it seems to me that unless and until it is done, you won't make much headway in trying to challenge current cosmology, which does have a model, and whose model is based on the EFE, which is well confirmed over a wide domain.

 

[PeterDonis]

What this paper is saying is that the predictions based on expansion don't work, at least not for this data set.

Can you give a brief summary of what mistakes you think the mainstream cosmology community (which says that the predictions based on expansion do work) is making in interpreting this data?

 

[PeterDonis]

Unlike some of my colleagues, I don't claim to have a Theory of Everything.

But you do obviously have some concrete model, since you are making predictions, and you can only make predictions if you have a concrete model. "I don't have a Theory of Everything" isn't a prediction. That's why I'm asking questions about your model: to try to understand what concrete model you are using to generate the predictions that you say match the data.

 

[Dale]

The redshift-distance relationship does imply something we don't yet understand is draining energy from the EM radiation as it travels long distances.

Please limit the discussion to the actual content of the paper. Such speculations were not accepted by MNRAS nor are they appropriate here.

For other posters, please try not to force the conversation into speculation. This is a controversial topic, so extra care must be made to keep it within the bounds of the professional scientific literature. As much as possible, support points with references in this thread, even before being specifically requested.

 

[rootone]

If tired light means that something happens to the light over long distances, yes

If red shift is not due to expansion then it must be due to something else.
As far as we know light does not change it's wavelength depending on only distance of the light source.
There is no reason I know of to think that it might be so.

 

[Dale]

@elerner I am not actually convinced that this statement in your paper is correct “To fit the data by excluding the nearest galaxies invalidates the comparisons. Rather, a fit has to be required to go through the low-z point”.

The key assumptions of the cosmological model are that at some large enough scale the universe is homogenous and isotropic. The cosmological models are not expected to work below that scale, and we know that those assumptions are invalid at small scales.

Perhaps the discrepancy in the low z values merely shows where the appropriate scale cut-off lies. In any case, it is certainly not required to fit nearby galaxies, and the comparisons are not invalidated by doing so. It merely informs the domain of applicability (in a rather expected and acceptable way)

http://www.tapir.caltech.edu/~chirata/ph217/lec01.pdf

 

[elerner]

A fractal of dimension n=2 means that matter is distributed in such a way that the total mass measured increases as D^2 (distance squared) when measured from any point. Average density then decreases with radius measured. Observations in peer-reviewed publications show that galaxies are distributed in this fractal way at least to scales of 200 Mpc, but it remains unclear whether that distribution is extended to still larger scales. There is no "special location" in such a distribution, but it would not be homogeneous at any scale. Prediction of the contraction or expansion of the universe requires a homogeneous distribution. (Indeed Edgar Allen Poe pointed out that Newtonian gravitation makes a homogeneous universe unstable to gravitational collapse.) If the universe is not homogeneous, the prediction is not valid.

In this paper, and an earlier one with my colleagues, I tested the hypothesis that z in linearly proportional to D at all D against observations. That is a perfectly good test of the hypothesis that this is the actual relation between these two quantities. I don't have, and don't have to have, a model that explains why this relationship holds. That is a further step. It is just like when the formula for the Balmer series was discovered long before quantum theory, which explains that formula, was perfected.

Right now, no experimental test is sensitive enough to show if light does change just by traveling a large distance. But such a test could be done if the proper equipment were put on the planned LISA experiment.

 

[PeterDonis]

A fractal of dimension n=2 means that matter is distributed in such a way that the total mass measured increases as D^2 (distance squared) when measured from any point.

Ok, got it.

There is no "special location" in such a distribution

I'm not sure I understand this. Average density would be highest at the spatial origin and would decrease in all directions out from it, so that point is clearly a "special location".

Prediction of the contraction or expansion of the universe requires a homogeneous distribution.

Why do you think this? The fact that the standard cosmological model predicts expansion based on homogeneity does not mean there can't also be models which are not homogeneous but which predict expansion.

 

[PeterDonis]

I tested the hypothesis that z is linearly proportional to D at all D against observations.

Ok, but we don't directly measure D, so you have to make some other assumptions in order to extract D from the data. What are those assumptions?

 

[elerner]

Dale, the whole point of the Tolman test is to compare the apparent size or surface brightness of nearby objects with distant ones. If you throw out the nearby ones (which number in the thousands and go out to a z of 0.14) you are saying that we happen to live in the center of the one part of the observable universe that has galaxies twice as big as everywhere else. You want to make that hypothesis? I think Copernicus would have a good chuckle over that.

 

[elerner]

Peter, and others, look up fractals online. They are very common structures in nature. There is no special point of origin. If you go out from ANY point, you get the same results. In my paper, the necessity for assuming a relationship of z and D in the non-expanding case is in order to have a formula for deriving absolute luminosity from measured redshift and apparent luminosity. We do that so we can compare galaxy samples that have the same absolute luminosity. So we don't have to measure distance to test the predictions. The prediction is just that galaxies of the same luminosity have the same radius independently of z.

Signing off for tonight. Good night to all!

 

[kimbyd]

<< Mentor Note -- after a very long Mentor discussion, we acknowledge that this paper, while potentially controversial, has been published in a reputable peer-reviewed journal. We believe that a discussion of this paper can be useful and constructive. Thanks >>

This paper of mine was published online in Monthly Review of the Royal Astronomical Society.
The title is a good summary: "Observations contradict galaxy size and surface brightness predictions that are based on the expanding universe hypothesis".
For a non-technical description, see our press release here.

Eh. Galaxy dynamics are tremendously complicated.

The expansion of the universe is not, and is consistent with a wide array of observational evidence.

If there is a discrepancy between predictions based upon galaxy dynamics and the uniform expansion, by far the most likely resolution of that discrepancy is that the model of galaxy dynamics is wrong. There is potentially some value in this paper in highlighting a discrepancy that needs to be understood to gain a better understanding of our universe. But it's far, far more likely to result in increased understanding of galaxy dynamics rather than increased understanding of expansion.

If you want to try to draw some conclusions about the expansion rather than galaxy dynamics, it is an absolute necessity to start drawing in other data sets, from the CMB to primordial light element abundances to baryon acoustic oscillations to supernova measurements.

 

[Dale]

If you throw out the nearby ones (which number in the thousands and go out to a z of 0.14) you are saying that we happen to live in the center of the one part of the observable universe that has galaxies twice as big as everywhere else.

Do galaxies have a size distribution? Or are they all exactly the same size? If they have a size distribution then there will in fact be many pockets throughout the universe with large or small galaxies. And in any of those pockets that nearby datapoint will be off.

In any case, a model is based on assumptions, and when the assumptions are violated you expect poor fits. If a model designed for large scales fits well at large scales and not well at small scales then you use it for large scales and not for small scales, as it was intended!

I think you are incorrect in asserting that the previous comparisons are invalid. Certainly the professional community seems to consider it to be valid. How is it justified in the previous literature?

 

[Dale]

The expansion of the universe is not, and is consistent with a wide array of observational evidence.

This is a critical point that must not be forgotten. An alternative cosmology must not just explain one type of observation, it must explain all of the different kinds of observations. In fact, even this one kind of observation is well explained by the current model at sufficiently large scales.

 

[elerner]

Dale,
Galaxies of the same luminosity have significant spread in size, but the expected variation in the mean or median of a sample decrease with N^0.5, where N is sample size. So with samples of the size used here, we can get pretty small statistical uncertainty. If you look at the size of the error bars, you get an idea. To hypothesize that regions as large as 800 Mpc across, the nearby samples or even 200 Mpc across have galaxies 10 sigma away bigger than elsewhere is a big leap. If you really want to hypothesize that, the easy test from the data is to compare galaxies in different parts of the sky. If they are the same, then you would also need to hypothesize we are right in the center of this very odd patch of galaxies. Is that the pocket you want to hypothesize? Ptolemy would doubtless approve. Or would you accept isotropy of galaxy size as falsification of your idea we are in some special pocket?

There were no justifications for leaving out the nearby galaxies. It was just not part of most of those studies. They compared HST observations at large z with each other. They omitted to compare them with nearby galaxies, which necessarily would be observed with other telescopes (at low z the HST survey volumes are too small).

 

[Dale]

but the expected variation in the mean or median of a sample decrease with N^0.5, where N is sample size. So with samples of the size used here, we can get pretty small statistical uncertainty.

That assumes that the measurements of nearby galaxies are uncorrelated or independent. But you are still missing the point.

I start with assumptions X, Y. With those assumptions I generate model Z. Model Z is shown to fit a wide variety of data under conditions where assumptions X and Y are expected to hold. You diligently show that model Z does not hold in all cases, the specific case being one extreme data point of one data set where assumption X may not hold. By your own analysis model Z does hold for the remainder of the data set where assumption X holds. It is only when the assumptions are violated that the model fails. I agree and point out the fact, recommending that model Z be used only when the assumptions X, Y hold. It works as designed!

It is like you are buying a car and complaining that it doesn’t float well. It wasn’t intended to float well, it was intended to drive well.

I believe that the usual length scale for cosmological scale is something like hundreds of Mpc. Isn’t z=0.027 smaller than that?

which necessarily would be observed with other telescopes

And which could therefore have measurements which were non-randomly different from the remainder of the dataset, thereby negating the statistical benefit of large sample sizes.

Don’t get me wrong. I do accept your paper as evidence against the standard cosmology, just not as very strong evidence due to the issues mentioned above. So (as a good Bayesian/scientist) it appropriately lowers my prior slightly from “pretty likely” to “fairly likely”, and I await further evidence.

 

[elerner]

"It is important to note that any over-all comparisons of cosmological models must be based on all available data-sets."
That is the last sentence of my paper--hope you all read to the end. But, given that, it is essential to state when each data-set contradicts the overall model. If you want to look at an informal list of other contradictions, based on peer-reviewed literature, see <unacceptable link deleted>

 

[elerner]

By the way, Dale, z=0.027 defines a region 200 Mpc across and the nearby data goes out to z=0.14, which defines a region 1100 Mpc across.

 

[Dale]

By the way, Dale, z=0.027 defines a region 200 Mpc across

Ok, so that is getting towards the lower end of what is generally considered “cosmological scale” to my knowledge. It appears then that the primary contribution of your work may be to show that the lower limit of the “cosmological scale” is a little larger than previously assumed. Or it may be that the HST data is systematically different in some way than the nearer data.

Your data does not seem to contradict previous studies showing the validity of current models at substantially larger scales.

 

[Dr. Courtney]

@elerner I am not actually convinced that this statement in your paper is correct “To fit the data by excluding the nearest galaxies invalidates the comparisons. Rather, a fit has to be required to go through the low-z point”.

The key assumptions of the cosmological model are that at some large enough scale the universe is homogenous and isotropic. The cosmological models are not expected to work below that scale, and we know that those assumptions are invalid at small scales.

Perhaps the discrepancy in the low z values merely shows where the appropriate scale cut-off lies. In any case, it is certainly not required to fit nearby galaxies, and the comparisons are not invalidated by doing so. It merely informs the domain of applicability (in a rather expected and acceptable way)

http://www.tapir.caltech.edu/~chirata/ph217/lec01.pdf

Emphasis added. This view is convenient. But it renders the current cosmological model invalid on the basis that it is not falsifiable. In the face of any purported falsification, proponents can always claim, "Maybe you've just found where the cut off is." To be falsifiable, a scientific model must specify the domain of its validity.

This is a critical point that must not be forgotten. An alternative cosmology must not just explain one type of observation, it must explain all of the different kinds of observations. In fact, even this one kind of observation is well explained by the current model at sufficiently large scales.

But you are breaking your own rules by going outside the bounds of the current paper. The current paper does not purport to be a comprehensive alternative cosmology, but merely pointing out where the current consensus cosmology (expanding universe) is contradicted by observations. Yes, it also contains some hints and elements that may represent progress toward an alternate cosmology, but the current consensus cosmology did not emerge in a single paper.

 

[Dale]

Emphasis added. This view is convenient. But it renders the current cosmological model invalid on the basis that it is not falsifiable. In the face of any purported falsification, proponents can always claim, "Maybe you've just found where the cut off is." To be falsifiable, a scientific model must specify the domain of its validity

I don’t think this objection is a strong one. The domain of validity always depends on the specific purpose intended, and is not a hard and fast boundary that is generally explicitly specified. For example, what is the domain of validity of Newton’s laws? Precisely who “must specify” it, and what is the specification?

In any case, all your suggestion would lead to is a series of models Z1, Z2, Z3, ... identical except for their specified their domains of validity. Each model would be falsifiable per your criterion, and you would simply reject Zn and use Zn+1 as needed. You would wind up at the same point.

The objection is particularly not strong here since the scale of the discrepancy reported is on the order of the previously known limit. It is perhaps a small refinement to the domain, not a radical change.

 

[Dr. Courtney]

I don’t think this objection is a strong one. The domain of validity always depends on the specific purpose intended, and is not a hard and fast boundary that is generally explicitly specified. For example, what is the domain of validity of Newton’s laws? Precisely who “must specify” it, and what is the specification?

In any case, all your suggestion would lead to is a series of models Z1, Z2, Z3, ... identical except for their specified their domains of validity. Each model would be falsifiable per your criterion, and you would simply reject Zn and use Zn+1 as needed. You would wind up at the same point.
oun
The objection is particularly not strong here since the scale of the discrepancy reported is on the order of the previously known limit. It is perhaps a small refinement to the domain, not a radical change.

One can frame the discussion that way, but then one needs to acknowledge that moving what were previously thought to be firm boundaries of well established theories is not usually treated as an insignificant accomplishment. "On the order of ..." That's astrospeak for a factor of 10.

The domain of Newton's laws was thought to be truly universal at one time. Now we know that the validity depends on the accuracy needed for a specific prediction and is never absolute. The current cosmological model has nowhere near the supporting evidence as Newton's laws. It may be that the current paper merely represents moving the domain of it's validity. But it also may mean an alternate cosmology can emerge which better explains the available evidence AND accurately predicts new findings.

I like to keep an open mind.

 

[Dale]

one needs to acknowledge that moving what were previously thought to be firm boundaries of well established theories is not usually treated as an insignificant accomplishment

So acknowledged! Although this particular boundary is not that firm and I suspect that the OP considers that to be an insignificant accomplishment, particularly compared to what he wanted to accomplish.

But it also may mean an alternate cosmology can emerge which better explains the available evidence AND accurately predicts new findings.

That is not in the MNRAS paper.

I like to keep an open mind.

Me too, but I try to do so in a roughly Bayesian framework. As Bayesian evidence this is a small update to my priors.