Extragalactic Distances from the J-region Asymptotic Giant Branch - Abby Lee (ft Freedman & Madore)

[@talks_curator728]

*Index to Key Parts of the Talk* [0:00] Intro comments by Shaun [01:06] Abby's opening comments on paper and CCHP [01:37] Two takeaways to remember [02:36] Background and motivation for this work [04:50] Asymptotic Giant Branch Stars , HR diagram, carbon stars as standard candles [06:38] The J-Region Asymptotic Giant Branch (JAGB) Method [08:30] History of the JAGB Method [17:08] Comparison of brightnesses between distance indicators [18:58] The JAGB Method Advantages [20:25] TRGB, Cepheids, and JAGB are independent methods [22:45] How accurate/precise are JAGB distances vs TRGB and Cepheids? [25:38] Comments on the paper [30:32] How do you know where the outer disk is? [33:32] *Final H0 and Distance Comparisons* [37:21] Summary [37:57] Is JAGB the gold standard method of the future? [43:36] What current work in cosmology is interesting but underappreciated?

[@tb82957]

Amazing talk yet again! Too bad you didn't get to ask them about the 'reply' from the SH0ES team earlier today on the arxiv. It's really difficult for bystanders to keep track of what is going on, but at least future JWST data should homogenize things further. It'd be great if at some point both groups could actually analyze an identical dataset.

[@CosmologyTalks]

Thanks! Yeah, the SH0ES paper came out after recording, though not before I published the video - I actually re-recorded my intro after noticing that SH0ES had their own JAGB paper earlier this year. It is hard as a bystander, absolutely - things are so nuanced! Hopefully the data eventually makes it very clear what the local Hubble value actually is. There have been various re-analyses of each other's data over the years - and as far as I understand, the SH0ES paper from yesterday didn't take issue with the measured values of any of the three CCHP+JWST results. Instead (again as far as they understand) they claim it is what is expected! Their claim appears to be that these analyses use a subset of the full set of local galaxies with supernovae, and this particular subset has a random scatter which, even with old HST data, pushes H0 down in all three methods. With JWST measurements in other galaxies they say one gets larger H0 values and the full combined H0 of all data in all three methods favours an H0 in tension with LCDM+CMB. (I think CCHP would argue that this subset actually contains the most trustworthy galaxies and the deviation between this sample and the full one might be systematic not random scatter.) So, essentially, the claim is that this result is correct, but within an accepted range of statistical fluctuations if H0 actually was ~72-73-ish. They do also claim that the combined H0 value in the CCHP "status report" doesn't take into account that the three methods use the same supernova data (and thus aren't independent of each other). So, even if SH0ES take the individual measurements at face value and aren't surprised, they do challenge the combination of the data. I thought I asked Wendy and Barry about this in the video last week and they said they did take that covariance into account, but perhaps my question wasn't clear, or perhaps SH0ES are wrong and CCHP did combine the data sets correctly.

[@madeleinebirchfield7658]

It is also possible that H0 is not a constant but depends on redshift, and then we need new physics to explain why H0 would change value from around 73 at small redshift to around 67 at larger redshifts, since JAGB can be studied out to larger redshifts than Cepheids can. If there is a redshift dependence, then I don't know if the 10 galaxies in the CCHP paper or the 42 galaxies in the SH0ES paper would be sufficient to show the redshift dependence due to lack of data - one may presumably want to take hundreds or thousands of galaxies with various different redshifts between z = 0 and some redshift limit, and then calculate the H0 using the standard candles at each specific redshift, resulting in an H0 per redshift value, and see what the relationship is between the calculated H0 vs redshift.

[@CosmologyTalks]

I have to start with a pedantic comment, but it is important so we make sure we're taking about the same things. H0 can't depend on redshift, no matter what new physics there might be. This is because its definition is "the value of the Hubble parameter at z=0". Note, the Hubble parameter itself does depend on redshift, in every model, and this is taken into account when people quote an "H0" value by looking at objects not at z=0. But I think what you mean by "H0 is not a constant but depends on redshift" is "the value we would infer for H0, by studying objects at a specific redshift and assuming ΛCDM, might depend on that redshift" (correct me if I'm wrong). This is true, and important to note, yeah. In fact, if the Hubble tension is not due to astrophysics, then almost certainly the value we get for H0 when we assume ΛCDM *will* depend on the redshift we observe at, because we're evolving it to z=0 with the wrong model. The problem is that it is really hard to get a model that can fix this without breaking other observations. You mention that 10, or even 42, galaxies aren't enough to see how things change with redshift. That's true, but a few things should be clarified. The 10, or 42, galaxies are all *very* close to us. So close that we don't expect them to be moving with the Hubble flow. These are just the galaxies where supernovae have gone off close enough that we can calibrate the supernovae to things like cepheids, TRGB and JAGB. The full SN data sets then used from Pantheon+ and the Carnegie Supernva Program to measure the expansion rate in the Hubble flow have many more SN and extend much further away (100s in CSP and 1000s in Pantheon+). So the redshift dependence of the Hubble parameter itself can be more accurately mapped out than one could with just 42 SN. In fact, this is the problem, and why it is so hard to fix the Hubble tension with some modification of the universe's expansion history. The data is too good and too constraining, rather than not informative enough. The even bigger problem comes from when supernovae are calibrated via the BAO. The recent Dark Energy Survey catalogue of around 1800 supernovae is the most constraining here. If one calibrates the DES SN via the BAO scale, and then look at the DES SN near z=0, the measured expansion rate is very tightly constrained to be close to the ~67 CMB value (+- 1, or even less). The number of supernovae involved make it very hard to fix the Hubble tension with anything cosmological between the redshifts where BAO are measured and z=0, so it requires something happening before the BAO observations in order to modify the BAO scale itself (or something wrong in the supernova calibration due to astrophysics, but if SN calibration is going wrong then the Hubble tension evaporates too). (see the recent Cosmology Talk on this: kzbin.info/www/bejne/oau3YaeNjMyaiNE) The DES SN+BAO inverse distance ladder is so constraining though that there *would* be enough statistical power to break the SN into redshift bins and ask "what H0 value do we infer from SN in this redshift range" (as you suggest). I'm unaware of anyone having done that, but it would be cool to see, yeah! (the dataset itself is very new - maybe someone is doing it already?)

[@CosmologyTalks]

I thought I recognised your username so I looked through past comments and see now that you commented on the inverse distance ladder video I linked above, so you've obviously already watched it! As you pointed out in that comment, the highly constraining nature of the inverse distance ladder does assume the FLRW metric (i.e. homogeneity and isotropy). So, if homogeneity and/or isotropy was sufficiently violated that we need to go beyond that metric then that might be a path to resolve the Hubble tension without astrophysics, yes! Given how constraining the CMB is in the early universe and the inverse distance ladder is in the late universe I might even go as far as to say that beyond FLRW is one of the most compelling directions to look in. Calculating anything without assuming the FLRW metric is notoriously difficult though, and no numerical relativity codes have shown (in my opinion) a convincing need to go beyond FLRW in a statistically isotropic universe. But neither has anyone (in my opinion) conclusively proven that we definitely don't need to either. People are actively looking into this, but, yeah, it is notoriously difficult to calculate beyond FLRW so progress is slower than progress within FLRW.

[@madeleinebirchfield7658]

​@@CosmologyTalks Yeah, when I mean by "H0 might vary by redshift" in the original comment is that "the value we would infer for H0, by studying objects at a specific redshift and assuming ΛCDM, might depend on that redshift".

[@madeleinebirchfield7658]

@@CosmologyTalks I also commented on the Jenny Wagner talk on the case against the cosmological principle about an year ago, asking her about alternative metrics to FLRW. The first time I took seriously the idea of moving beyond FLRW for a solution to the Hubble tension was when it was mentioned as an option in section VII.H in the Snowmass 2021 paper titled "Cosmology Intertwined: A Review of the Particle Physics, Astrophysics, and Cosmology Associated with the Cosmological Tensions and Anomalies". At the time it was more appealing to me than many of the more mainstream approaches like early dark energy which tried to resolve the Hubble tension but was running into problems with the S8 tension, and moving beyond FLRW would at the same time also potentially resolve the various cosmic dipole anomalies discussed in section VIII.F of the Snowmass paper.

[@thorntontarr2894]

If JAGB data offers a new method for distance measurements with the claimed advantages, it is not obvious to me in this presentation how the distance is determined and whether that technique is independent of the TRGB and Cephid methods. It wasn't until the end of the talk (about 40 mins in) that a geometric (parallax) distance technique is employed in four local galaxies and that three of those are too bright for JWST to use the JAGB. I hope I understood that correctly. I will read Lee's paper and Weinberg and Nikolaev to try gain clarity here. Aside: I made the comment about the misuse of significant figures in the earlier video.

[@CosmologyTalks]

(caveat that this is all just my own understanding I'm not a niche-subject expert on this...) The three methods JAGB, cepheids and TRGB) aren't entirely independent as they all need to use the same supernovae and the same local geometric calibrator. They do, however, more or less independently provide the link between those two distances. The JAGB method is to take the mode (i.e. peak value) of the histogram of the luminosities of the galaxies in the blue band of this video's thumbnail. This absolute luminosity shouldn't change over time, so the actual measured luminosity will get brighter/dimmer as a galaxy gets closer/father away, thus it provides a distance indicator (calibrated on a nearby galaxy where we know from geometric means how close it is). The cepheids and TRGB methods use different stars and features in the colour-magnitude diagram and so are independent methods to go from geometric distances to SN distances. CCHP can isolate stars within that blue band in any galaxy and, as far as I understand, the amount the x-axis value it should redshift is negligible over the distances being considered so they use the same band in all galaxies. The mode makes sense to me as the best measure because both median and mean will be affected by stars in the tails of the distribution, whereas the mode will not. It might have a larger scatter than median/mode but will be less likely to be biased. Re: your aside, yes I replied to your comment in the other video saying I brought this up 😅

[@diegofloor]

You might want to know that the audio is a bit low.

[@CosmologyTalks]

Thanks, yeah I've noticed that myself sometimes. I need to remember to boost the volume before uploading.

[@chestercurtis7548]

I am somewhat perplexed by the lack of need for spectroscopy. The new standard candle is identified as such by color (red) however color varies with the distance and at 13 GLY woldn't you need a very deep red spectroscopy check to ID them?

[@chestercurtis7548]

I realized after posting that these are NOT like supernovae and not bright enough to be seen individually at 13GLY so that is an unfair application. However in mid range (like cepheids) color is still a function of distance. Seems like spectroscopy makes them a "safer" standard candlr where cepheids don't need a "color check".

[@CosmologyTalks]

Yeah, as I'm sure you noticed, I did bring this up in the video as my naive thought was also that one would be better able to isolate the galaxies with spectroscopy. *My* thought was that it would be easier to isolate the carbon stars if one could see there was actually lots of carbon there. However their point that one can do this all empirically and just isolate the stars in the blue band of the figure in the thumbnail of this video and work with that population of stars defined in that way does seem persuasive to me. One doesn't need to understand the feature in order to use it, so long as one is confident it won't change with time. Your concern that the colour itself will redshift is true. I'm not sure by how much one expects the x-axis to shift though for the velocities that are relevant (as you point out in your clarification reply, we're only considering things ~10-50 Mpc away, not Giga Lightyears). The x-axis is called the "colour", but it isn't a measurement that is proportional to the wavelength/frequency of the light, it is the difference in measured magnitude between two photometric bands. So what is relevant is how much one expects the magnitude in each band to change when the light is redshifted, not how much one expects the wavelength of light to shift. I am very very far from an expert on this so I'll just have to speculate that this is small for the bands considered (if the luminosity as a function of wavelength in the wavelengths within the band is relatively constant this is indeed what I would expect because as some light is redshifted out of the band, other light is redshifted in). This stack exchange Q&A claims that the effect is indeed small (0.01 mag, which will bring very few galaxies in/out of the blue band): astronomy.stackexchange.com/questions/43581/hr-diagram-x-axis-color-and-redshift

[@chestercurtis7548]

@@CosmologyTalks Thanks for the thoughtful response. Abby did mention that because of JAGB brightness and abundance that they could be used much further out than cepheids which raised the concern over photometric versus spectral identification. The ability to make photometric narrow band filters with various transmission bands could be used to tease out these stars for galaxies beyond cepheid capabilities is intriguing. Best I could tell from other sources (en.wikipedia.org/wiki/Carbon_star), the carbon star "red" is due to blue absorption by the 3rd dredge up and not by distinctive (sharp in spectroscopy) emission spectra from the carbon thus spectroscopy might not be conclusive anyways? No need to answer, wonderful that this exciting new work is underway.