I Observational evidence against expanding universe in MNRAS

[PeterDonis]

Thread closed for moderation.

 

[PeterDonis]

The thread is being reopened in order to allow continued discussion of the specific paper referenced in the OP, and to allow @elerner to respond to specific questions regarding that paper (and the 2014 paper that it is based on). Please limit discussion to that specific topic. This thread is not about the general methodology of science or the overall pros and cons of the current mainstream cosmological model of the universe.

@elerner, in responding to questions, please give specific cites to your papers rather than general claims and opinions. We understand your basic claims; we are looking for the specific evidence and arguments given in your papers that you think support those claims, not repetitions of the claims themselves. Additional fine details of methodology not provided in the papers are fine (since that is a large part of what other posters have asked about).

 

[PeterDonis]

it seems that there's rather a lack of interest in my proposal (to independently try to verify the GALEX results reported in L14)

This is outside the scope of PF. Independent replication of scientific results is original research, which is not what PF is for.

 

[PeterDonis]

There are a range of sophisticated galaxy formation simulations available today, they would be a much better comparison given they represent the leading edge of the field and that the selection function could be applied to them.

@elerner, this is one question that I did not see you respond to. Is there a reason why the particular model of galaxy size evolution used in the paper was chosen? Or is there further work planned to apply a similar methodology to a wider range of models of galaxy size evolution?

 

[PeterDonis]

sizes as we measure them at low z, where there is no cosmic distortion

Doesn't this contradict your claim that the "expansion hypothesis" applies at all scales, right down to $z = 0$?

 

[elerner]

ruarimac said: ↑ There are a range of sophisticated galaxy formation simulations available today, they would be a much better comparison given they represent the leading edge of the field and that the selection function could be applied to them.

@elerner, this is one question that I did not see you respond to. Is there a reason why the particular model of galaxy size evolution used in the paper was chosen? Or is there further work planned to apply a similar methodology to a wider range of models of galaxy size evolution?

I did reply to this, but I can elaborate. My goal was to test the predictions based on the expansion hypothesis against the galaxy size and related data sets. For disk galaxies, the dominant , in fact only, theory I could find that made predictions prior to the publication of the data sets (starting around 2005) was the Mo et al theory. This is also the one that is by far most referenced as a comparison in the literature.

The many more recent simulations of galaxy growth do not produce predictions that can be tested against this data set. The models contain many free parameters that, the authors describe, are adjusted to fit the available data, including data on galaxy size and growth. As I pointed out in a previous post, fits to a data set can’t be tested against that data set. They can be tested only against new, different data sets that don’t exist at the time that the fits are made.

For ellipticals, I looked at predictions made not only by Mo et al, but also by three different theories of elliptical galaxy growth: puffing up, major merger and minor mergers. These are all I found referenced. My paper shows in detail that these theories also make quantitative predictions that conflict with observations, such as merger rates and the velocity dispersions of high-z ellipticals. This last data set poses a particularly severe conflict as it implies that for any expanding universe model the gravitating mass of high-z ellipticals is less than their stellar mass.

 

[elerner]

sizes as we measure them at low z, where there is no cosmic distortion
Doesn't this contradict your claim that the "expansion hypothesis" applies at all scales, right down to z=0" role="presentation">z=0?

Clearly not, as the formula for distortion depends on 1+z. When 1+z differs only slightly from 1, the cosmic distortion is insignificant. Strictly speaking, "no cosmic distortion" is a misstatement. I meant "no significant distortion" . Also, I don't say that expansion makes predictions down to z=0--i.e. the room next to yours. It makes predictions for all scales where matter is not bound gravitationally. This is thus for scales more than a few Mpc. That definition of "small scale" is far smaller than the 200-800 Mpc scale measurements that are the low-z comparisons in my paper.

 

[ruarimac]

The many more recent simulations of galaxy growth do not produce predictions that can be tested against this data set. The models contain many free parameters that, the authors describe, are adjusted to fit the available data, including data on galaxy size and growth.As I pointed out in a previous post, fits to a data set can’t be tested against that data set. They can be tested only against new, different data sets that don’t exist at the time that the fits are made.

The fact that a model was published after HUDF does not mean it was calibrated to it your observations, that is simply false. As a counter example I can point to the EAGLE simulations which are calibrated to the z~0 stellar mass-size relation from GAMA and SDSS. It was not calibrated on the redshift evolution and yet it matches data very well in size evolution from CANDELS which it was not calibrated to, nor UDF for that matter.

http://adsabs.harvard.edu/abs/2017MNRAS.465..722F

The calibration is described here:

https://arxiv.org/abs/1407.7040

It simply isn't true that this is the only model available. I do recommend you compare to a simulation in future, at the very least they can test the systematics of your analysis.

Furthermore your argument is illogical. You said you don't want to compare recent models with your data because they're fit to this data (which is false). But your claims about cosmology rest on the claim that there is no way to fit the data in an expanding universe. But then it doesn't matter if the galaxy evolution model was fit to the data, the parameters still have to be physical. If a fitted physical model can match the data then the data can be explained in an expanding universe, your whole argument about this being a problem for cosmology falls apart. I'm afraid you can't make both arguments simultaneously.

 

[laymanB]

If this hypothesis was correct and the universe is not expanding, is eternal, and is evolving as a function of time, why would not all the stars have burned out an infinite time ago?

 

[elerner]

If a fitted model can match the data then the data is can be explained in an expanding universe, your whole argument about this being a problem for cosmology falls apart. You can't make both arguments simultaneously.

Of course I can. It concerns different sets of data that still relate to the same size question. If you have just the disk size data, you can certainly fit it ex post facto with sufficient numbers of free variables. But for the ellipticals, the merger theories also make predictions about the rate of mergers, which can be tested against merger observations. In addition any size evolution for the ellipticals that matches the data also produces the paradox from the velocity data that the implied gravitational mass is less than the stellar mass. So for ellipticals you can't fit the size data, the merger data and the velocity data no matter what you do.

In the case you cited of Eagle, they claim a good fit to van der Wels. But in section 7 of my paper, I show why van der Wel's disk data disagrees with both Shibuya and Lerner et al 2014 because it does not take into account the RS effect at the longer wavelengths used.

 

[elerner]

If this hypothesis was correct and the universe is not expanding, is eternal, and is evolving as a function of time, why would not all the stars have burned out an infinite time ago?

In the Big Bang model, evolution slows down without limit going forward in time. So as you go forward in time, you would eventually get to a time without stars, without galaxies, etc. But evolution in the universe can speed up with time, as we know to be the case here on Earth. Similarly, if you extrapolate the observed accelerating evolution backwards in time, you get a slower evolution the further back you are. Therefore any stage of evolution of the universe--the formation of galaxies, the formation of stars etc. could have begun a finite time ago, but not the universe itself.

 

[Eric Bretschneider]

The assumption that galaxies should appear larger in an expanding universe of course requires that you know the current distance to the galaxy. The issue I have is that regardless of whether or not the universe is expanding, galaxy size is measured based on the distance to the galaxy when the light was emitted. We don't see galaxies at their current distance, only the distance long ago in the past.

 

[Jean Tate]

it seems that there's rather a lack of interest in my proposal (to independently try to verify the GALEX results reported in L14)

This is outside the scope of PF. Independent replication of scientific results is original research, which is not what PF is for.

Thanks. Got it.

 

[Jean Tate]

While this thread was locked, there was an exchange in a thread in International Skeptics Forum that I think is highly pertinent:

Eric;
For the sake of brevity, could you please provide maybe a link to the specific Galex dataset(s) you used?
Thanks

http://galex.stsci.edu/casjobs/

If Jean Tate provides a place to put them I can at some point--not the next few days but the next few weeks--post an excel file of our GALEX data.

My only suggestion, for now, is as before: GitHub. While I do have an account there, I would rather not host your data @elerner. Could I suggest that you set up a GitHub account (if you don't already have one)? Also, may I suggest that you convert any excel file to a CSV before posting it?

 

[Jean Tate]

From before this thread was temporarily locked:

<snip>
However, I'm even more curious about how you "fitted the radial brightness profile with a disk law excluding the central 0.1 arcsec for HST and 5 arcsec for GALEX".

For example, did you write your own code? Or use a publicly available tool or package? Something else??

I will try to ask you only one question at a time @elerner, and wait for an answer before asking another.

 

[elerner]

This is an easy one, said Humpty Dumpty. My colleagues wrote their own code. The basic algorithm is simple. Once you convert the profile to a logarithmic one, the exponential disk fit is just a straight line, so easy to fit.

 

[Jean Tate]

This is an easy one, said Humpty Dumpty. My colleagues wrote their own code. The basic algorithm is simple. Once you convert the profile to a logarithmic one, the exponential disk fit is just a straight line, so easy to fit.

Thanks!

My follow-on is more than one question, but all related, so I hope that's OK (all about the GALEX data).

Did you deconvolve the PSF? If so, how? And what did you use for the PSF?

Did you do a 1D or a 2D fit? If the former, which axis through the center did you use? If the latter, how did you de-project the disks (since very few will be seen close enough to face-on)?

 

[ruarimac]

Of course I can. It concerns different sets of data that still relate to the same size question. If you have just the disk size data, you can certainly fit it ex post facto with sufficient numbers of free variables.

These simulations have been run. They were not calibrated to your data. Comparing your results to them is not fitting anything, as I explained.

So for ellipticals you can't fit the size data, the merger data and the velocity data no matter what you do.

Then you have no reason not to look at recent simulations.

In fact this claim seems to be to be entirely untested in the paper. The only physical model you plot against elliptical data is the Mo et al. model, which is for disks, not ellipticals. You seem to have concluded that no model can fit this data, by testing no models of ellipticals whatsoever. Your cosmological claims rest on this elliptical data not being able to be fit by any expanding universe model, but you haven't tested any models of it. Furthermore people have actually shown that simulations (EAGLE) reproduce van der Wel's data. You claim van der Wel's has made mistakes but I don't think you have demonstrated that.

In the case you cited of Eagle, they claim a good fit to van der Wels. But in section 7 of my paper, I show why van der Wel's disk data disagrees with both Shibuya and Lerner et al 2014 because it does not take into account the RS effect at the longer wavelengths used.

I'm honestly don't believe this "resolution-size" effect really exists. Your only demonstration of this in the paper is to compare GALEX and SDSS galaxy sizes. You don't seem to consider the possibility that galaxy sizes can be different in different bands, asserting that because the relation is not 1-to-1 it must be an effect of the surveys. That doesn't follow. You also don't consider that catalogue sizes may not be measured consistently. Measuring sizes with barely resolved objects will always be difficult but if you actually model the process of fitting sizes any bias can be simulated. I'm aware that people have tested the ability to measure sizes of between ground based imaging and HST, the results correlation is much tighter than in your test, the bias is much smaller and has the opposite sign. If this "RS effect" were real then the all lower resolution imagery would be biased to higher radii, but that isn't what other people have found.

http://adsabs.harvard.edu/cgi-bin/bib_query?arXiv:1708.00005

van der Wel on the other hand studied the wavelength dependence of their size measurements with the different HST bands, the PSF is different between different bands and so any substantial effect there would be folded in. They fit the sizes themselves so this is a much better test that your GALEX SDSS comparison because they're measured consistently. Given that you have different selection functions and are measuring in different bands it's hardly surprising there is disagreement, it doesn't imply van der Wel is wrong. Just to summarise my key point, the Almaini paper I linked demonstrated this RS effect is not some blanket effect. This disproves your criticism of van der Wel, therefore the comparison with the EAGLE simulations has already demonstrated that recent simulations can fit the size evolution. The claim that this is evidence against an expanding universe just doesn't hold any water.

 

[SelfSim]

Eric;

We are pursuing the data from the link you posted at ISF. We sampled similar data (for the M31 region) and found that it contained very high levels of backgound noise (confirmed by a Pixinsight measurement of σ = 1.358 x 10 ̄ ² for a sample of the M31 region). We used AIP4WIN software to compare resulting FWHM figures across both software packages.

In both cases the FWHMs are rejected due to the high background noise, due to the difficulty of accurately determining a radius value in order to calculate the FWHMs.
We conclude that this would have also have been the case in Lerner's etal's radius measurements.

Which leads us to our rather simple question of:

'Did you perform your radius measurements on single image data, or stacked data?'

(We also note that irrespective of how Lerner etal performed the measurements, the cutoff results are around 50% of the specified FWHM .. instead of being the lower limit of derived FWHM values).

Looking forward to your response.
Cheers

 

[PeterDonis]

In the Big Bang model, evolution slows down without limit going forward in time.

I have not seen this in any description of the standard Big Bang model. Where are you getting it from?

 

[elerner]

Did you perform your radius measurements on single image data, or stacked data?'

(We also note that irrespective of how Lerner etal performed the measurements, the cutoff results are around 50% of the specified FWHM .. instead of being the lower limit of derived FWHM values).

Selfsim, I have said before that the difference between FWHM and radius is just the same as diameter and radius--a factor of 2. That is where your 50% comes from. Every galaxy was measured individually. Jean Tate, the algorithm took annuli cenetered on the center of the galaxy image and calculated the light in each, took the log, plotted and fitted a straight line. The central areas that are affected by PSF were cut out before the fit was done. Look all, we did check our measurements against GALEX catalog at one end and several HUDF measurements at the other. They all come out the same. We just wanted to make sure ours were identical for both HUDF and GALEX. But as my paper points out in comparing our results with Shibuya, they are robust with respect to the exact measurement technique.

 

[elerner]

elerner said: ↑
In the Big Bang model, evolution slows down without limit going forward in time.
I have not seen this in any description of the standard Big Bang model. Where are you getting it from?

Peter, lots of Big Bang papers describe how huge changes in the universe occur in tiny fractions of a second during inflation, slowing down through many magnitudes through decoupling, etc, finally getting to hundreds of millions of years for formation of stars and earliest galaxies and now changes taking many billions of years. Many papers, including some of the earliest ones by Eddington, relate the Big Bang to a general increase in entropy and reduction in energy flows with time, leading to slower and slower change in the future and eventual "heat death" of the universe.

 

[PeterDonis]

lots of Big Bang papers describe how huge changes in the universe occur in tiny fractions of a second during inflation, slowing down through many magnitudes through decoupling, etc, finally getting to hundreds of millions of years for formation of stars and earliest galaxies and now changes taking many billions of years. Many papers, including some of the earliest ones by Eddington, relate the Big Bang to a general increase in entropy and reduction in energy flows with time, leading to slower and slower change in the future and eventual "heat death" of the universe.

Okay, but that's a much weaker claim than the claim you made, which was:

In the Big Bang model, evolution slows down without limit going forward in time. So as you go forward in time, you would eventually get to a time without stars, without galaxies, etc. But evolution in the universe can speed up with time, as we know to be the case here on Earth.

None of the things you mention in the first quote above--rates of inflation, decoupling, star formation, galaxy formation, eventual heat death--are inconsistent with evolution on Earth, or in some other local region, speeding up with time. Nor does the standard Big Bang model claim that all "evolutions" must get slower with time. So whatever "Big Bang model" you think you are refuting, it appears to be a straw man model you made up, not the actual model that cosmologists use.

 

[SelfSim]

Did you perform your radius measurements on single image data, or stacked data?'

Eric;
Could you please answer the only explicit question I asked in post#119?
(Repeated above again, for clarity).

Thanks

 

[timmdeeg]

@elerner How is the observational evidence that CMB temperature dropped down from initially (last scattering) 3000 K to 2.7 K as measured today consistent with the assumption that the universe does not expand?

 

[Jean Tate]

The thread is being reopened in order to allow continued discussion of the specific paper referenced in the OP, and to allow @elerner to respond to specific questions regarding that paper (and the 2014 paper that it is based on). Please limit discussion to that specific topic. This thread is not about the general methodology of science or the overall pros and cons of the current mainstream cosmological model of the universe.

@elerner, in responding to questions, please give specific cites to your papers rather than general claims and opinions. We understand your basic claims; we are looking for the specific evidence and arguments given in your papers that you think support those claims, not repetitions of the claims themselves. Additional fine details of methodology not provided in the papers are fine (since that is a large part of what other posters have asked about).

(my bold)

I fully support this.

It is really interesting to discuss general ideas about, and evidence for, LCDM cosmological models ("the Big Bang Theory"), why general challenges to these might be flawed, and incredibly easy to veer away from the specific scope of the two papers (Lerner 2018, and Lerner+ 2014). So I think it's really important to read what the two papers actually say, and where you can, quote from them.

 

[elerner]

selfsim, I did answer your question:

Every galaxy was measured individually.

I also agree that, while the general topic deserves its own thread, this thread should focus on these two papers. So I will not respond further here to general questions on cosmology models.

 

[PeterDonis]

It is really interesting to discuss general ideas about, and evidence for, LCDM cosmological models

That is off topic for this thread, as @elerner has correctly pointed out.

 

[SelfSim]

Did you perform your radius measurements on single image data, or stacked data?'

Every galaxy was measured individually.

Apologies for my persistence here, but your answer is somewhat ambiguous.
The question pertains to the data you used .. not the measurements you performed on that data.

 

[elerner]

Selfsim, you should take the time to read the papers. I am re-summarizing stuff described clearly there. Here is the quote from Lerner et al 2014:

To measure total flux and half light radius, we extracted the average surface brightness profile for each galaxy from the HUDF or GALEX images.

My answer had no ambiguity at all. GALEX images and HUDF images of individual galaxies were measured. That is the data used. The images are not stacked. Stacking is the process where many images from different galaxies are added together. We measured each galaxy separately. The galaxy measurements were then combined into samples as described in detail in the papers and a median radius for each sample determined. I am not going to answer further questions that are explicitly answered in the papers. I am assuming people read these two papers before asking. Otherwise everyone's time will be wasted--not the purpose of this forum.

 

[SelfSim]

Selfsim, you should take the time to read the papers. I am re-summarizing stuff described clearly there. Here is the quote from Lerner et al 2014:

To measure total flux and half light radius, we extracted the average surface brightness profile for each galaxy from the HUDF or GALEX images.

My answer had no ambiguity at all. GALEX images and HUDF images of individual galaxies were measured. That is the data used. The images are not stacked. Stacking is the process where many images from different galaxies are added together. We measured each galaxy separately. The galaxy measurements were then combined into samples as described in detail in the papers and a median radius for each sample determined.

Ok thanks for your response.

So, my next question is:

'Did you then subtract the sky background data corresponding to the individual image files?

(.. because this would then impact the cut off values).

I am not going to answer further questions that are explicitly answered in the papers.

Umm .. I think you need to, if you expect others to follow your atypical methodology and thence consider, at least as a possibility, accepting your results(?)

I am assuming people read these two papers before asking. Otherwise everyone's time will be wasted--not the purpose of this forum.

We are attempting to bridge gaps here between your paper, which contains a non-standard approach, and others with which we are more familiar.
Your generosity in assisting in this process is appreciated.

 

[Dale]

As I pointed out in a previous post, fits to a data set can’t be tested against that data set. They can be tested only against new, different data sets that don’t exist at the time that the fits are made.

Given the size of these data sets k fold cross validation or other similar approaches would be entirely appropriate. Such approaches are specifically designed to address the statistical issue you raise here. So you could, in fact, both fit and test using these techniques. The time when the data set is made is not relevant.

 

[Copernicuson]

Elearner,

I want to believe your model, but I don't understand why you don't copy and paste the appropriate sections of your paper to answer the questions.

 

[SelfSim]

Eric;

'Would you also please provide a reference source (a link preferably) for the specific HUDF datasets you used?'

Thanks.

 

[Jean Tate]

I need to quote an earlier post, to provide adequate background:

While this thread was locked, there was an exchange in a thread in International Skeptics Forum that I think is highly pertinent:

Eric;
For the sake of brevity, could you please provide maybe a link to the specific Galex dataset(s) you used?
Thanks

http://galex.stsci.edu/casjobs/
<snip>

<snip>

From L14:

We have determined the minimum measurable angular radius of galaxies, m, for each of the telescopes by plotting the abundance of galaxies (with stellarity index<0.4) [...]

Checking the GALEX Schema Browser, per the link you provided, @elerner, I could find no "stellarity index" ("No Column Names or Descriptions Contain: 'stellarity'" and "No Table Names or Function Names Contain: 'stellarity'").

What is this index? How did you determine the stellarity index for each of the GALEX objects (galaxies?) you used at this stage of your sample selection?

I also checked the SDSS DR5 Schema Browser, with the same result, "The expression has not been found in the column and flag names, their units and descriptions, or in the SDSSConstants table.".

 

[Jean Tate]

Thanks, @elerner.

<snip>
Jean Tate, the algorithm took annuli cenetered on the center of the galaxy image and calculated the light in each, took the log, plotted and fitted a straight line. The central areas that are affected by PSF were cut out before the fit was done. Look all, we did check our measurements against GALEX catalog at one end and several HUDF measurements at the other. They all come out the same. We just wanted to make sure ours were identical for both HUDF and GALEX. But as my paper points out in comparing our results with Shibuya, they are robust with respect to the exact measurement technique.

I have several follow-on questions, but they are of lower priority and importance than some others I've been waiting to ask. Like the one on "stellarity index" I just posted.

 

[Jean Tate]

This post has no explicit question to @elerner (though your response would be welcome); rather, it's my attempt to make the first para of L14 somewhat less confusing (I will be editing this later, to try get the symbols and formatting right). Of course, others' inputs and comments are most welcome.

As Tolman1,2 demonstrated, the dependence of the bolometric surface brightness (SB) of identical objects as a function of redshift z is independent of the specific parameter of the adopted cosmology, e.g., Hubble constant, dark matter Ω_M and dark energy Ω_Λ content of the Universe. For this reason the comparison of the surface brightness of similar objects at different distance was seen as a powerful tool to test for the expansion of the Universe. In fact, in any expanding cosmology, the SB is expected to decrease very rapidly, being proportional to (1+z)^-4, where z is the redshift and where SB is measured in the bolometric units (VEGA-magnitudes/arcsec⁻² or erg sec⁻¹cm⁻²arcsec⁻²). One factor of (1+z) is due to time-dilation (decrease in photons per unit time), one factor is from the decrease in energy carried by photons, and the other two factors are due to the obΩject being closer to us by a factor of (1+z) at the time the light was emitted and thus having a larger apparent angular size. (If AB magnitudes or flux densities are used, the dimming is by a factor of (1+z)³, while for space telescope magnitudes or flux per wavelength units, the dimming is by a factor of (z+1)⁵). By contrast, in a static (non expanding) Universe, where the redshift is due to some physical process other than expansion (e.g., light-aging), the SB is expected to dim only by a factor (1+z), or be strictly constant when AB magnitudes are used.

For me, confusion set in early: "the bolometric surface brightness (SB)" is ambiguous ... does "SB" refer to "surface brightness" of any kind? Or is it strictly limited to "bolometric surface brightness"? It would seem the former ... but is it used consistently throughout the rest of the paper? Stay tuned.

Then there are the magnitudes, and fluxes: "bolometric units (VEGA-magnitudes/arcsec⁻² or erg sec⁻¹cm⁻²arcsec⁻²)", "If AB magnitudes or flux densities are used,", and "While for space telescope magnitudes or flux per wavelength units,".

One aid to disentangling these terms: the Wikipedia article on Luminosity. Another, Wikipedia on AB magnitudes. Fundamentally, this is all about energy, or power (energy per unit time). Sadly, the terms "flux" and "flux density" are not always used consistently, though I think them both being used inter-changeably in the one paper is rare these days; between papers? well you have to keep a sharp eye out. In either case, "flux" ~= perpendicular through a unit surface. Not relevant for L14, but definitely for radio astronomy, is whether there's also a "per steradian" aspect, or is isotropy assumed (and so an implicit 4π).

Then there's the system of units and zero points. As the above extract makes clear, L14 uses the cgs system (MKS is far more common); however, while both VEGA and AB are used - their zero-points are fixed - that of "space telescope magnitudes" is not.

Finally, there's the bandwidth and filter: filters do not have infinitely sharp boundaries, nor is the wavelength (or frequency) response perfectly uniform; conversions between observations made using one system on one facility (telescope, filter, camera) and another are a bane of astronomers. And converting to bolometric ("absolute") magnitudes even more full of a shopping list of assumptions (to quote from an earlier post).

 

[elerner]

Hi all, I will be replying but somewhat delayed as we are busy at the lab making fusion.

 

[elerner]

Jean Tate--the stellarity index has the name "NUV_class_star" and "FUV_class_star"
Selfsim--the HUDF data was from Coe D., Bentez N. Sanchez S. F., Jee M., Bouwens R., Ford H. 2006, AJ 132, 926
Jean Tate--they key thing is that bolometric units are just energy. AB units are energy per unit frequency, so since the light is redshifted, that means different exponents for the surface brightness. No physical difference, just how you measure it.

 

[Jean Tate]

Jean Tate--the stellarity index has the name "NUV_class_star" and "FUV_class_star"

Thanks.

<snip>
Jean Tate--they key thing is that bolometric units are just energy. AB units are energy per unit frequency, so since the light is redshifted, that means different exponents for the surface brightness. No physical difference, just how you measure it.

Well, you and I are going to have to disagree, I think, on both this and (more importantly) on how consistent L14 is, in its words and how they have been applied to both the data and analyses. The words L14 seems to use rather loosely and/or inconsistently include "intrinsic luminosity", "distance d", "apparent magnitudes", "absolute magnitudes", "flux-luminosity relation", "bolometric luminosity", and "flux" ... and that's just in Section 2. However, I'll leave my concerns on this topic for a later time.

Instead, I have just one question about something in Section 3. Again, concerning only GALEX. Here's the full context:

Moreover, to avoid biasing the comparison of data obtained with telescopes having different resolutions, we also require that the minimum measurable physical size of galaxies r_m is the same, in each pair of samples, for GALEX (low z) and HUDF (high z). We have determined the minimum measurable angular radius of galaxies, θ_m, for each of the telescopes by plotting the abundance of galaxies (with stellarity index <0.4) vs. angular radius for all GALEX MIS3-SDSSDR5 galaxies and for all HUDF galaxies and determining the lower-cutoff angular radius for each.

For each and every GALEX MIS3-SDSSDR5 galaxy, how did you determine the "angular radius"?

OK, two questions: did you cut on NUV_class_star and FUV_class_star first (selecting those with values <0.4 only), and then determine the angular radii?

 

[elerner]

For each and every GALEX MIS3-SDSSDR5 galaxy, how did you determine the "angular radius"?

The GALEX catalog provides a 50% flux radius--half light radius--for both FUV and NUV. We used those for this calculation. So, yes first we eliminated the >0.4 stellarity entries and then we plotted the numbers. there was a sharp fall-off at the minimum measurable radius.

 

[Jean Tate]

Thanks.

The GALEX catalog provides a 50% flux radius--half light radius--for both FUV and NUV. We used those for this calculation. So, yes first we eliminated the >0.4 stellarity entries and then we plotted the numbers. there was a sharp fall-off at the minimum measurable radius.

Searching the GALEX Schema Browser, I found the following Column names, Units, and Descriptions which seem to match your description (the Tables in which they appear in curly brackets):

NUV_FLUX_RADIUS_2, {blank}, Fraction-of-light radius ( 0.5000) {PhotoObjAll, VisitPhotoObjAll}
FUV_FLUX_RADIUS_2, {blank}, Fraction-of-light radius ( 0.5000) {PhotoObjAll, VisitPhotoObjAll}
fuv_ncat_flux_radius_2, {blank}, FUV FLUX_RADIUS #2 (-fd-ncat)(px)[0.50] {PhotoObjAll, VisitPhotoObjAll}

Can you please confirm that what you used was NUV_FLUX_RADIUS_2 and FUV_FLUX_RADIUS_2?

L14 has HUDF and SDSS references, but apparently no GALEX ones. Only if you can fit it in (otherwise feel free to skip this question), may I ask why you did not include a GALEX reference?

The sentence immediately after the one I quoted in my last post is ("this cutoff" refers to "the lower-cutoff angular radius"):

We took this cutoff to be the point at which the abundance per unit angular radius falls to 1/5 of the modal value.

I guess "abundance" means something like number, or relative frequency. For GALEX - NUV and FUV - what were the units of angular radius that you used? Why did you choose "1/5 of the modal value"?

This seems - to me - to be strange and arbitrary.