Early large galaxy studies

[pinball1970]

TL;DR Summary

The standard model for how galaxies formed in the early universe predicted that the James Webb Space Telescope would see dim signals from small, primitive galaxies. But data are not confirming the popular hypothesis that invisible dark matter helped the earliest stars and galaxies clump together.

"What the theory of dark matter predicted is not what we see," said Case Western Reserve astrophysicist Stacy McGaugh, whose paper (12.11.24) describes structure formation in the early universe.

McGaugh, professor and director of astronomy at Case Western Reserve, said instead of dark matter, modified gravity might have played a role. He says a theory known as MOND, for Modified Newtonian Dynamics, predicted in 1998 that structure formation in the early universe would have happened very quickly—much faster than the theory of Cold Dark Matter, known as lambda-CDM, predicted.

https://iopscience.iop.org/article/10.3847/1538-4357/ad834d (open access)

There was also this in Nature (abstract only)

https://www.nature.com/articles/s41586-024-08094-5 (13.11.24)

A survey of 36 galaxies that yielded "three ultra-massive galaxies (logM★/M⊙ ≳ 11.0, where M★ is the stellar mass and M⊙ is the mass of the Sun) require an exceptional fraction of 50 per cent of baryons converted into stars—two to three times higher than the most efficient galaxies at later epochs."

and "The contribution from an active galactic nucleus is unlikely because of their extended emission."

Possibly a reference a couple of studies like the below in the last few months,

https://iopscience.iop.org/article/10.3847/1538-3881/ad57c1

 

[ohwilleke]

FWIW, the prediction of early structure formation with MOND is pretty generic to all gravity based explanations of dark matter phenomena.

Do Early Galaxies And The Hubble Tension Have A Common Source?

Early structure formation could also be related to the Hubble tension which is equally glaring. There are basically three ways to resolve the Hubble tension (and a solution could come, in part, from two or more of them and not just one):

1. Early CMB based calculations of the Hubble tension are off by about 9%.

McGaugh, for example, has suggested that this is a plausible full or partial explanation (although he hasn't explicitly made the connection between this possibility and early structure formation in anything that I've seen him write, as I spell out in the paragraphs in bold below).

For example, maybe the Planck collaboration omitted one or more theoretically relevant components of the formula for converting CMB observations to a Hubble constant value that were reasonably believed to be negligible (indeed, it almost certainly did so). But it could be that one or more of the components omitted from the Planck collaboration's calculated value of Hubble's constant from the CMB data actually increase the calculated value by something on the order of 9% because some little known factor makes the component(s) omitted have a value much higher than one would naively expect.

Also, since the indirect determination of the value of Hubble's constant from CMB measurements is model dependent, any flaw in the model used could cause its determination of Hubble's constant to be inaccurate.

An indirect CMB based determination of Hubble's constant is implicitly making a LamdaCDM model dependent determination of how much the universe had expanded since the Big Bang at the time that the CMB arose. If the LambdaCDM model's indirect calculation of Hubble's constant predicts that the CMB arose later than it actually did, then its indirect determination of the value of Hubble's constant would also be too low, and a high early time value of Hubble's constant would resolve the problem.

This is a plausible possibility because the James Webb Space Telescope has confirmed that the "impossible early galaxies problem" is real, implying that there is definitely some significant flaw (of not too far from the right magnitude and in the right direction) in the LambdaCDM models description of the early universe, although the exactly how much earlier than expected galaxies arose in the early universe (which is a mix of cutting edge astronomy, statistical analysis, and LambdaCDM modeling) hasn't been pinned down with all that much precision yet.

The impossible early galaxy problem is that galaxies form significantly earlier after Big Bang than the LambdaCDM model predicts that they should. The galaxies seen by the JWST at about redshift z=6 (about 1.1 billion years after the Big Bang) are predicted in the LambdaCDM model to apear at about redshift z=4 (about 1.7 billion years after the Big Bang).

If the CMB arose more swiftly after the Big Bang than the LambdaCDM model predicts it did but the amount by which the universe had expanded at that point was about the same, in much the same way that galaxy formation actually occurred earlier than the LambdaCDM model predicted that it would, then that could fully or partially resolve the Hubble tension.

The relationship between Hubble's constant and the amount of expansion in the universe at any given point in time is non-linear (it's basically exponential). So, figuring out how much of a roughly 55% discrepancy at 1.1 billion years after the Big Bang in galaxy formation time translates into in Hubble constant terms, at about 380 million years after the Big Bang, is more involved than I have time to work out today, even though it is really only an advanced pre-calculus problem once you have the equations set up correctly. But my mathematical intuition is solid enough to suspect that the effect isn't too far from the 9% target to within the uncertainties in the relevant measurements.

2. Late time measurements of the Hubble tension are off by about 9%. But this is complicated by the fact that the late time measurements are confirmed by independent methods, so the source of the inaccuracy in the late time measurement would have to be due to a systemic error source common to all of them.

For example, one explanation that has been explored is that the little corner of the universe around the Milky Way from the perspective of solar system observers has some local dynamics, or has local distortions that impact light at the relevant wavelengths reaching us in the solar system (e.g. due to localized gravitational lensing or local distributions of interstellar gas and dust) that has nothing to do with the expansion of the universe, but is indistinguishable, by the most precise existing methods used to measure Hubble's constant in the late time universe, from an increase in Hubble's constant of about 6.4 (km/s)/Mpc.

3. There are new physics that explain it.

Another Possible Cause Of The Impossible Early Galaxies Problem

Another point related to the impossible early galaxies problem which hasn't, IMHO, received enough attention, is that star formation within galaxies appears to happen about ten times faster than expected, probably as a result of the influence of weak magnetic fields in star forming galaxies. See T.-C. Ching, et al., "An early transition to magnetic supercriticality in star formation" Nature (January 5, 2022) (open access) https://doi.org/10.1038/s41586-021-04159-x

It takes much longer for galaxies to coalesce from individual stars (on the order of hundreds of millions or billions of years) than it does for stars to form from hydrogen gas. The "classical view" is that it takes about ten million years for a typical star to form, while the observations in the article cited above, suggests that one million years is closer to the mark.

I have blogged this article and quoted secondary materials discussing it that post.

Given that we're only looking for a tweak of 9% or less to the current inference of the early time Hubble constant value from the CMB observations (about 0.1 dex in the terms astronomers like to use for uncertainties), it doesn't have to be a huge effect.

This star formation timing issue may very well be completely independent of the impact of dark matter and/or gravity based explanations for phenomena attributable to dark matter on how long it takes galaxies to form. And, while a star formation issue would still be a problem with the LambdaCDM based model of how galaxies form, it isn't a core assumption of the LambdaCDM model and could be incorporated into predictions of how long it takes galaxies take to form in the LambdaCDM with this one modification without greatly disturbing the other cosmological predictions of LambdaCDM.

On the other hand, it could be that some of the discrepancy between theory and observation in star formation rates could have a full or partial cause related to gravitational effects that explain dark matter phenomena that Ching (2022) does not consider as a possible explanation.

 

[ThickTarget]

Apologies for posting on an old thread, but you didn't get much of a response. I'm just going to quote a part to respond to multiple issues.

The impossible early galaxy problem is that galaxies form significantly earlier after Big Bang than the LambdaCDM model predicts that they should. The galaxies seen by the JWST at about redshift z=6 (about 1.1 billion years after the Big Bang) are predicted in the LambdaCDM model to apear at about redshift z=4 (about 1.7 billion years after the Big Bang).

There are a few problems with this; a) LCDM doesn't predict galaxies on its own, and b) the galaxy measurements are not as robust as they appear.

LCDM is a wonderfully predictive cosmological model, but it only tells you about the dark matter halos. Specifically, you can calculate the number density of halos at a given mass and redshift. There isn't a particular redshift when a halo of mass X forms, because it's a question of how rare that halo is. But the observations are not measuring DM, they are measuring light (a proxy for mass in stars). And so on to the dark matter halo one has to estimate how normal matter (baryons) are accreted, cool and eventually form stars. It gets even more complicated if you make the treatment more realistic, and include things like supernovae or magnetic fields. Older simple analytic models are being replaced by complex numerical simulations, and different simulators have different methods, and even changing the resolution has an effect. This is to say, LCDM predicts the halos, but not the galaxies. One cannot say "LCDM predicts galaxies form at redshift X", because that calculation depends on many more assumptions than just LCDM, largely the efficiency of star formation. There are some physical limits, for example one can assume a galaxy has the universal ratio of baryons-to-DM which is likely an upper limit on the mass of stars that could form. None of the confirmed objects fall in this category.

Observations themselves also have assumptions. For example, a value which is often quoted is the stellar mass, this is not a direct observable. Instead, one fits the galaxy fluxes with a galaxy model with some formation history and an assumed stellar population (distribution of stellar masses). The latter is key because it can change the mass-to-light ratio (of the stars), assuming a different initial mass function (IMF) can significantly change the estimated stellar mass. In these young galaxies, the light is dominated by massive stars, which make up a tiny fraction of the mass in a typical IMF. Do these galaxies really have stellar populations like the Milky Way? So far, it cannot be measured. There are other things too, like having a bursty star formation history, which you linked.

https://arxiv.org/abs/2208.07879

Along the same lines, often in some of these early studies the analysis assumes that the light production comes from stars, whereas it's now clear that supermassive black holes (active galactic nuclei) exist in these populations. AGN are the most luminous objects in the universe. If you incorrectly attribute the light from the AGN to be powered by stars, you would erroneously assume it was a very massive galaxy. This has already happened to some extent. In the first few months of JWST there was a lot of press about impossibly massive galaxies, which originated with the Labbe et al. 2023 Nature paper. Labbe et al. found very red, compact galaxies which they claimed were old and massive. It has since emerged that there is a huge population of faint AGN at high redshift (Little Red Dots), revealed in JWST data. At least one Labbe galaxy is an AGN, possibly all of them are (TBC). These objects will exist in older datasets, but it's only now that this can be confirmed.

https://arxiv.org/abs/2302.00012

Lastly there is the effect of interlopers, where the redshift of the object turns out to be totally wrong. Most of these objects have their redshifts estimated from images (photometry) alone, red and/or dusty galaxies can be misidentified as higher redshift. This happened dramatically with the z=17 "Schrödinger" galaxy, which turned out to be a dusty z=5 galaxy (as some predicted). With JWST, there are spectroscopic samples, but the Steinhardt paper you liked is heavily based on photometric redshifts. A lot of candidate massive high-redshift galaxies turned out to be lower redshift dusty galaxies, because they are much more numerous. This recent paper shows a sample of candidate massive early galaxies followed up with Keck spectroscopy, among the super-massive 16 candidates all of them were confirmed to have wrong photometric redshifts. A few were AGN. Based on this, they estimate the number of these early supermassive galaxies is hugely overestimated (>4 times at 99%, >17 times at 68%), easing the impossibly early galaxy problem. The number densities they found are in good agreement with some simulations. The paper you linked finds a high density, but they only have three galaxies, there are huge uncertainties. The most massive object in that paper is also suspect as it lacks other confirming emission lines, it may be an interloper as discussed.

https://arxiv.org/abs/2404.19018
https://arxiv.org/abs/2208.02794

This is a plausible possibility because the James Webb Space Telescope has confirmed that the "impossible early galaxies problem" is real, implying that there is definitely some significant flaw (of not too far from the right magnitude and in the right direction) in the LambdaCDM models description of the early universe, although the exactly how much earlier than expected galaxies arose in the early universe (which is a mix of cutting edge astronomy, statistical analysis, and LambdaCDM modeling) hasn't been pinned down with all that much precision yet.

I don't think the problem has been confirmed, the jury is still out. There have been claims, some object has been debunked, some await spectroscopic confirmation. As I explained, predictions about galaxies are the product of LCDM and galaxy formation modelling, if there is a discrepancy it doesn't imply the fault lies with LCDM. It's really only if physical limits are violated, and even then there are assumptions like the IMF. One clause you cut out of your quote from the nature paper was, "We find no tension with the Λ cold dark-matter model in our sample." I think that's about as much as one can fairly say currently. Comparing to simulations, very high redshift galaxies do seem to be more luminous than estimated, but not to the extent that cosmology is violated. There are many hypotheses for the origin of this tension, but it doesn't have to be cosmology. In general, trying to do cosmology with galaxy evolution is extremely difficult.

 

[ohwilleke]

Apologies for posting on an old thread, but you didn't get much of a response. I'm just going to quote a part to respond to multiple issues.



There are a few problems with this; a) LCDM doesn't predict galaxies on its own, and b) the galaxy measurements are not as robust as they appear.

LCDM is a wonderfully predictive cosmological model, but it only tells you about the dark matter halos. Specifically, you can calculate the number density of halos at a given mass and redshift. There isn't a particular redshift when a halo of mass X forms, because it's a question of how rare that halo is. But the observations are not measuring DM, they are measuring light (a proxy for mass in stars). And so on to the dark matter halo one has to estimate how normal matter (baryons) are accreted, cool and eventually form stars. It gets even more complicated if you make the treatment more realistic, and include things like supernovae or magnetic fields. Older simple analytic models are being replaced by complex numerical simulations, and different simulators have different methods, and even changing the resolution has an effect. This is to say, LCDM predicts the halos, but not the galaxies. One cannot say "LCDM predicts galaxies form at redshift X", because that calculation depends on many more assumptions than just LCDM, largely the efficiency of star formation. There are some physical limits, for example one can assume a galaxy has the universal ratio of baryons-to-DM which is likely an upper limit on the mass of stars that could form. None of the confirmed objects fall in this category.

Observations themselves also have assumptions. For example, a value which is often quoted is the stellar mass, this is not a direct observable. Instead, one fits the galaxy fluxes with a galaxy model with some formation history and an assumed stellar population (distribution of stellar masses). The latter is key because it can change the mass-to-light ratio (of the stars), assuming a different initial mass function (IMF) can significantly change the estimated stellar mass. In these young galaxies, the light is dominated by massive stars, which make up a tiny fraction of the mass in a typical IMF. Do these galaxies really have stellar populations like the Milky Way? So far, it cannot be measured. There are other things too, like having a bursty star formation history, which you linked.

https://arxiv.org/abs/2208.07879

Along the same lines, often in some of these early studies the analysis assumes that the light production comes from stars, whereas it's now clear that supermassive black holes (active galactic nuclei) exist in these populations. AGN are the most luminous objects in the universe. If you incorrectly attribute the light from the AGN to be powered by stars, you would erroneously assume it was a very massive galaxy. This has already happened to some extent. In the first few months of JWST there was a lot of press about impossibly massive galaxies, which originated with the Labbe et al. 2023 Nature paper. Labbe et al. found very red, compact galaxies which they claimed were old and massive. It has since emerged that there is a huge population of faint AGN at high redshift (Little Red Dots), revealed in JWST data. At least one Labbe galaxy is an AGN, possibly all of them are (TBC). These objects will exist in older datasets, but it's only now that this can be confirmed.

https://arxiv.org/abs/2302.00012

Lastly there is the effect of interlopers, where the redshift of the object turns out to be totally wrong. Most of these objects have their redshifts estimated from images (photometry) alone, red and/or dusty galaxies can be misidentified as higher redshift. This happened dramatically with the z=17 "Schrödinger" galaxy, which turned out to be a dusty z=5 galaxy (as some predicted). With JWST, there are spectroscopic samples, but the Steinhardt paper you liked is heavily based on photometric redshifts. A lot of candidate massive high-redshift galaxies turned out to be lower redshift dusty galaxies, because they are much more numerous. This recent paper shows a sample of candidate massive early galaxies followed up with Keck spectroscopy, among the super-massive 16 candidates all of them were confirmed to have wrong photometric redshifts. A few were AGN. Based on this, they estimate the number of these early supermassive galaxies is hugely overestimated (>4 times at 99%, >17 times at 68%), easing the impossibly early galaxy problem. The number densities they found are in good agreement with some simulations. The paper you linked finds a high density, but they only have three galaxies, there are huge uncertainties. The most massive object in that paper is also suspect as it lacks other confirming emission lines, it may be an interloper as discussed.

https://arxiv.org/abs/2404.19018
https://arxiv.org/abs/2208.02794



I don't think the problem has been confirmed, the jury is still out. There have been claims, some object has been debunked, some await spectroscopic confirmation. As I explained, predictions about galaxies are the product of LCDM and galaxy formation modelling, if there is a discrepancy it doesn't imply the fault lies with LCDM. It's really only if physical limits are violated, and even then there are assumptions like the IMF. One clause you cut out of your quote from the nature paper was, "We find no tension with the Λ cold dark-matter model in our sample." I think that's about as much as one can fairly say currently. Comparing to simulations, very high redshift galaxies do seem to be more luminous than estimated, but not to the extent that cosmology is violated. There are many hypotheses for the origin of this tension, but it doesn't have to be cosmology. In general, trying to do cosmology with galaxy evolution is extremely difficult.

I would disagree that the jury is out.

For example, while the author's solution to the problem is speculative, Rajendra Gupta in "JWST early Universe observations and ΛCDM cosmology" arXiv:2309.13100 (September 22, 2023) (published at MNRAS 524, 3385-3395 (2023) which states:

Deep space observations of the James Webb Space Telescope (JWST) have revealed that the structure and masses of very early Universe galaxies at high redshifts (z~15), existing at ~0.3 Gyr after the BigBang, maybe as evolved as the galaxies in existence for ~10 Gyr. The JWST findings are thus in strong tension with the ΛCDM cosmological model.

is still the mainstream position. Similarly, in arXiv:2402.16646 the authors state:

James Webb Space Telescope's (JWST) observations since its launch have shown us that there could be very massive and very large galaxies, as well as massive quasars very early in the history of the universe, conflicting expectations of the ΛCDM model. This so-called ''impossibly early galaxy problem'' requires too rapid star formation in the earliest galaxies than appears to be permitted by the ΛCDM model.

Similarly, Feldmann (2025) Monthly Notices of the Royal Astronomical Society, Volume 536, Issue 1, January 2025, Pages 988–1016, https://doi.org/10.1093/mnras/stae2633 states in the abstract that:

Recent observations with JWST have uncovered unexpectedly high cosmic star formation activity in the early Universe, mere hundreds of millions of years after the big bang.

The body text goes on to explain that:

The first billion years of cosmic time were a pivotal epoch in the history of our Universe that witnessed the formation of the first galaxies, the rapid growth of supermassive black holes, and the reionization of intergalactic hydrogen (Stark 2016; Dayal 2019; Inayoshi, Visbal & Haiman 2019; Robertson 2022). Before the launch of the JWST, the ultraviolet (UV) luminosity and star formation rate (SFR) of galaxies at high redshifts were primarily constrained by observations from the Hubble Space Telescope (HST) and ground-based facilities. These observations provided crucial insights into the galaxy population up to z∼10⁠, revealing a steep faint-end slope of the UV luminosity function (LF) and a fast decline in their overall normalization with increasing redshift (e.g. McLure et al. 2013; Bouwens et al. 2015; Finkelstein et al. 2015; Oesch et al. 2018; Bouwens et al. 2022). However, the number densities and properties of galaxies at z≳9 were not well constrained with challenges arising from, e.g. low number statistics and lensing uncertainties (Bouwens et al. 2017; Atek et al. 2018).

The advent of JWST has transformed our understanding of high-redshift galaxy evolution, thanks to its excellent sensitivity, resolution, and wavelength coverage. Among many other findings, JWST revealed a higher-than-expected density of UV-bright galaxies at high z (Finkelstein et al. 2022; Naidu et al. 2022; Finkelstein et al. 2023; Labbé et al. 2023; Casey et al. 2024), confirmed spectroscopically the presence of galaxies up to z∼14 (Curtis-Lake et al. 2023; Robertson et al. 2023; Carniani et al. 2024; Castellano et al. 2024; Harikane et al. 2024), and uncovered the population of galaxies most likely responsible for reionization (Atek et al. 2024). Furthermore, JWST-based studies showed an elevated UV luminosity density ρUV at z>10 compared with both previous observational estimates and modelling predictions (Finkelstein et al. 2023; Harikane et al. 2023; Donnan et al. 2023a; Bouwens et al. 2023b; Donnan et al. 2023b; Chemerynska et al. 2024; Conselice et al. 2024; Harikane et al. 2024), suggesting that an important component of galaxy theory is either missing or not sufficiently understood. Solving this conundrum is critical, not only to gain a better understanding of the physical processes driving galaxy evolution at high z but also to constrain the impact of galaxies on cosmic reionization.

The emerging consensus is that structure formation as detected by the JWST is occurring much sooner than models that assumed LambdaCDM as a baseline predicted, and that the impossible early galaxies problem is real.

While astronomers and astrophysicists have "kicked the tires" of some specific observations, the overwhelming evidence is that the stage of structure development seen is several z higher than LambdaCDM based models predicted before JWST was available. Hubble had strongly hinted at this, but it wasn't definitive then. Now, it is.

Now, those are just the facts about what level of structure formation exists at particular redshift levels. There is certainly not a consensus on why this is the case. It is an "unsolved problem" in astrophysics. Proposed solutions that deviate from, or supplement, the LambdaCDM model are all over the map.

But none of the assumptions used to estimate the level of structure formation at a given redshift z assuming the LambdaCDM model have been seriously cast in doubt and have won favor yet as a widely accepted explanation.

Furthermore, as a relatively simple six parameter model, whose parameters have been measured to reasonable precision, there isn't a lot of room to tweak that model to post-dict observations. The CMB data, in particular, tightly fixes those parameters.

There is also good reason to point the finger at LambdaCDM is a likely the source of the problem:
* The Hubble tension
* The σ₈ tension
* LambdaCDM seriously underestimates of the number and velocity of colliding galaxies like the Bullet cluster
* The failure to come up with proposed dark matter candidates that produce the right halo shape (since the NFW halo distribution predicted analytically does not resemble what is seen in real life)

And, another dozen or two other serious conflicts between LambdaCDM's predictions and observations.

When the paradigmatic model is already suffering a death by a thousand cuts, the possibility that whatever is flawed in that model is also causing the impossible early galaxies problem has the same source is "well-motivated."

 

[ThickTarget]

I would disagree that the jury is out.

For example, while the author's solution to the problem is speculative, Rajendra Gupta in "JWST early Universe observations and ΛCDM cosmology" arXiv:2309.13100 (September 22, 2023) (published at MNRAS 524, 3385-3395 (2023) which states:

is still the mainstream position.

No, it's not the mainstream position. Who does Gupta cite when he makes this claim? No one. None of the papers in the following section say "as evolved", or anything like it. And does he do any analysis himself to prove galaxies are as evolved? No. It is a totally baseless claim. High redshift galaxies have smaller sizes, lower stellar masses, lower chemical enrichment than modern ones. In what way are they as evolved? They aren't.

https://arxiv.org/abs/2309.04377
https://arxiv.org/abs/2304.08516

The only "analysis" in the paper is complete nonsense. Gupta copies some measurements of galaxy sizes. But this "test" assumes galaxies don't evolve in size over time, which is not the case in big bang cosmology, including LCDM and Gupta's models. It only makes sense in a non-evolving pure tired light models, which is where Gupta copied the data and idea from. But even if you ignore this, the analysis is doubly wrong, because even at fixed redshift there are big galaxies and small galaxies. To make a comparison like this at all one needs to carefully select analogous galaxies, and as I said it makes no sense whatsoever in LCDM. Total gibberish.

This paper should have never been accepted in its current state. Ned Wright also pointed out that Gupta's model of a time varying gravitational constant is rejected by Lunar ranging data.

https://www.astro.ucla.edu/~wright/cosmolog.htm

The emerging consensus is that structure formation as detected by the JWST is occurring much sooner than models that assumed LambdaCDM as a baseline predicted, and that the impossible early galaxies problem is real.

Those are two completely different statements. Saying that the star-formation rate density is higher than some models does not mean it conflicts with the cosmology. And saying that does not imply that these galaxies are "impossibly early". As I explained, it is very difficult to actually show a galaxy violates cosmology, and none of the confirmed galaxies do.

LambdaCDM seriously underestimates of the number and velocity of colliding galaxies like the Bullet cluster

That's also not true. It was a long debate over the bullet cluster. Initially, people assumed that the velocity of the merger was equal to the velocity of the shock seen in x-ray data. But when people sat down and did hydrodynamical simulations of the merger, they found that a lower velocity merger could make a faster shock. People also found that in the case of merging clusters there are problems with the codes used to find clusters in the simulations, that when they collide they become one object. With different codes, the rarity of such mergers rose dramatically. The most recent papers find no tension with LCDM. This was all figured out a decade ago.

adsabs.harvard.edu/abs/2015MNRAS.452.3030T