Wide binary debate heats up again

73 thoughts on “Wide binary debate heats up again”

1. Figure 11 of the paper (Arxiv:2311.03436) is much more convincing in my opinion. Close binaries affect the result, but not in a manner that would depend on the internal acceleration of the wide binary. So in Newtonian mechanics, you are expecting a flat line on this figure, while MOND implies a trend like the dashed line. It is not too likely that systematics conspire to give a flat line.

   You are correct that the main analysis relies on fitting the v_tilde distribution. It is good that your post focuses on this crucial parameter. The next step is to focus on some typical measure like the median, and then quantify trends with the wide binary internal acceleration.

   LikeLike

   1. Indranil – you make a lot of the relative P-values. But what is the absolute chi^2? For how many degrees of freedom?

      LikeLike

      1. I have not worked this out, but I did comment in several places on how the model is not a perfect fit to the data. My model is fairly similar to that of the Pittordis & Sutherland (2023) paper (https://doi.org/10.21105/astro.2205.02846). The total chi^2 in the best Newtonian model with a thermal eccentricity distribution (2e) is given in table 6 of the cited paper and also plotted in its figure 16. The total chi^2 = 109.8+94.6+58.0+70.8 = 333.2. However, it should be clear from figures 13-15 that there are 70 v_tilde bins in each r_sky range and four bins in r_sky. The 70 v_tilde bins cover the range 0-7 with a step size of 0.1, as can be seen clearly in these figures. Therefore, the total number of degrees of freedom N_dof = 280. The chi^2 distribution with such a large number of degrees of freedom can be approximated as Gaussian with mean N_dof and dispersion sqrt(2*N_dof), which in this case is sqrt(560) or just under 24. This means it is entirely reasonable to get a total chi^2 of 333.2 for 280 degrees of freedom.

         As in the Newtonian model, the best MOND fit is for the thermal eccentricity distribution and the resulting chi^2 values are given in table 9. In this case, the total chi^2 = 355.8+258.2+114.1+130.5 = 858.6. The number of degrees of freedom still remains 280, so the expected distribution of the chi^2 statistic still has a mode at 280 and a variance of about 24 with a nearly Gaussian shape. As a result, instead of a 2.25 sigma deviation from the expected value as occurs in the Newtonian case, the discrepancy now becomes 24.45 sigma.

         Thus, focusing on the thermal eccentricity distribution which is expected on other grounds and which works best in both gravity theories, the analysis of Pittordis & Sutherland (2023) used N_dof = 280 bins in (r_sky, v_tilde) and found a total chi^2 of 333.2 in the Newtonian case but 858.6 in the MOND case. The discrepancy with the expectation that chi^2 should be roughly N_dof +/- sqrt(2*N_dof) is thus slightly over 2 sigma in the Newtonian case but about 24 sigma in the MOND case.

         For my own study, I have not given chi^2 values because the smaller sample size means one should really use binomial statistics. However, people can see for themselves that we get a fairly decent fit to the overall (r_sky, v_tilde) distribution. I also wrote in the conclusions about how rotation curves do not always fit perfectly in MOND either, but this is often considered just par for the course in astronomy. One would not expect the correct gravity theory to fit the wide binary data perfectly because obviously our model is not perfect and there may be some observational issues. But that is like looking at a spectrum to see if there is a spectral line somewhere (requiring a passable continuum model), seeing none, and saying that the best model with no spectral line does not fit the continuum well. There is other physics relevant to the continuum shape beyond whether there is a line at the critical wavelength. Similarly, wide binaries are affected by other things, like the eccentricity distribution and if it changes with wide binary separation. Having said that, these should not affect the results very much. So we should have a fairly good fit to the data.

         In short, the black curves on Figure D1 of the paper need to be compared with the histograms to get an idea of how well the best model fares. This gives an idea of how close our best model comes to matching the actual data. For the purpose of this discussion, please ignore the blue lines on this figure.

         LikeLike

         1. OK, thanks. Taking what you say about chi^2/dof at face value, then the absolute Newtonian fit is good enough (in the astronomy sense) that I’m persuaded that the relative preference for it is meaningful.

            My next questions is whether there is any independent evidence for what the CB fraction is. Both models want it to be high (66, 70%) in Table 1, and it apparently has to change a lot to make any difference (30% in Fig. D1). But do we have any idea what this number should be?

            LikeLike

            1. I am not too familiar with that, but I was asked to cite the Offner 2023 paper in relation to this, so you might want to check that out and the context in which it is cited. Stars do seem to be in binaries quite often. The thing I am a little concerned about is that we should be skipping some of the close binaries because more distant ones should be resolved and very tight ones would not impact the analysis due to orbits looping around within the Gaia timeframe (hence the need for a fairly narrow range in close binary separations). So f_CB does seem a bit high. But first of all, it does not make that much difference. And secondly, one can alter the assumptions a little bit and get f_CB to drop by a lot more than its formal error within any individual analysis variant. The referee did ask about this in some detail and it is the main reason for the long time spent at the refereeing stage. You can get f_CB down to about 0.5 within our model, and perhaps with further improvements, it could be a little bit less. There is barely any degeneracy with the gravity law, which is why I am still fairly convinced by the Newtonian result despite uncertainty over just how common undetected close binaries are. As for independent studies of how plausible the estimated f_CB is, you will have to ask observers. Of course, it is a bit tricky because we are talking about undetected companions. In my opinion, it is a little on the high side, but not implausibly so. It is also extremely difficult to understand the properties of the extended tail to the scaled velocity distribution any other way.

               LikeLike

               1. OK, thanks. I have no opinion on what f_CB is, and I’m not suggesting it will swap one gravity law for the other. I am saying it gives me the heebie-jeebies because it has a big impact on the shape of the model distribution. That let’s it soak up a lot of other effects, including systematics we may not have considered modeling. So you could come up with a perfectly reasonable fit for f_CB and reality could have some completely different number and it might not matter or it might be the result of systematics rendering all this moot. I have no idea how big a concern that is here, but I’ve seen things like this happen more times than I can remember.

                  LikeLike

                  1. As far as the gravity law is concerned, systematic effects are relevant only if there is some relation with the wide binary internal acceleration, so obviously a link with the separation would be like that. A changing f_CB with wide binary separation could then be a concern. This is why I was asked by the other authors to prepare something like Figure 16. I have also provided arguments in the paper why significant trends in f_CB with the wide binary separation are unlikely. And one would need a very particular type of effect to cancel out the MOND signal and get a flat median v_tilde with the wide binary internal acceleration. In general, because we get a flat line but some trend was expected in MOND, systematics that hide a MOND signal would need to cancel it out somehow. Personally, I do not think that is very likely.

                     There are also significant differences with rotation curve studies because the inclination is handled statistically rather than needing to be estimated for each system, the distance is known much better, the mass estimates require M/L for single stars rather than for stellar populations, we do not need to worry about gas, etc. On the other hand, there are new types of uncertainties specific to wide binaries, an obvious example being undetected close binary companions.

                     LikeLike

2. There is a nice YouTube video explaining my work on the wide binary test.

   I was not involved with its production and did not answer queries related to my paper to help with producing this video. But I did email it as soon as it was accepted, giving more time to make the video and post it as soon as my paper appeared on Arxiv. It is now also on the MNRAS website:
   https://dx.doi.org/10.1093/mnras/stad3393

   I am giving a talk about this on first December at four in the afternoon British time. The talk is supposed to be one hour with another hour of questions. I am expecting this to be similar to the wide binary lecture on my YouTube channel where I explained the results to my department a few months ago, though we made several improvements to the analysis in relation to referee comments.

   LikeLike

   1. so are you a dark matter advocate ?

      LikeLike

   2. Having grown up and learned my science at a time when (almost) everyone used frequentist statistics rather than Bayesian statistics, I’m not quite sure how one models the orbital eccentricity. The inclination part seems straight-forward – all orbital planes are equally likely, but if you treat the orbit as an ellipse, isn’t this assuming Newtonian gravity? If you ignore binaries where the stars’ accelerations are close to a0, so may be in the Newtonian regime for part of their orbits, can you model non-circular orbits in the deep MOND regime and use that to test the fit to MOND? I appreciate that it doesn’t allow you to use alpha-grav as a parameter, but it would be good to be able to show that even if you make the most favourable assumption for MOND, that it still does not fit the observational results.

      LikeLike

      1. Eccentricities are defined in a careful way for a generalised gravity theory, with the definition also valid in Newtonian gravity. Please see section 2.3.1 of MNRAS, 480, 2660 in relation to this. Regarding the distribution of eccentricities, power-law forms are considered. The exponent is allowed to vary over a wide range and is inferred simultaneously with the other parameters. There are other constraints on what this exponent gamma can be. I have discussed this in the discussion section. It is shown that plausible changes to the eccentricity distribution cannot mask a MOND signal. In general, the eccentricity distribution does not alter the v_tilde distribution all that much. Moreover, there would need to be a change in this distribution depending on the wide binary separation. This is unlikely given independent constraints on the eccentricity distribution from the angle between the projected separation and relative velocity of each wide binary.

         LikeLike

         1. Thanks for your comment, Indranil, which answers my question.

            LikeLike

   3. Indranil, the video takes the position that MOND is falsified. I hope you don’t want to pose that (as it will sound to outsiders) “MOND is false, see? We all said so, didn’t we?” Surely you still see the great predictive value of MOND in the many unexpected phenomena it predicted?

      Secondly, I don’t think your paper rebuts Chae’s paper. While I’m pretty sure the result you give is true or at most accidentally false, the 5 sigma result of Chae hasn’t disappeared. The focus should therefore be on the search for theories that explain both statistical results at the same time, rather than on how MOND was false and GR right (which is an unfair lie).

      LikeLike

      1. I explained the problem with the Chae result in Section 5.2. General Relativity is sufficient to explain both papers. The result of Chae can be understood as due to a systematic effect, as explained very carefully in the paper and briefly touched upon in the video.

         LikeLike

         1. OK, your method of Monte Carlo approach is not my expertise and I’m not well-versed in it so I can’t criticize it. It does seem strange that the remaining sample size is larger than in Chae’s tightened cut of 0.3%, but I don’t know enough to comment on that.

            Let me for one minute assume that on average, also the less reliable data from Chae should be around the correct value so Chae’s result is still there but less reliable. Then for a theory that can explain both Chae’s result: consider the approach of Alexandre Deur. It should be a comfortable theory for you, since it uses GR with as only modification that gravity has self-interaction. And how should gravity not self-interact? It carries energy, which is mass.

            Deur argues in his papers that many MOND predictions arise (especially for galaxies) automatically from the self-interaction. By example, for galaxies approximately in a 2D mass distribution, he derives a 1/r scaling gravitational force which starts to show and takes over the 1/r2 law below ~a_0. This is because the self-interaction of gravity pulls the gravitons toward the 2D mass distribution which means that the gravity spreads in a circle (circumference scales with r) instead of a ball (surface scales with r2).

            What follows is how I imagine Deur’s theory should work in wide binaries (WB). The stars are far enough from each other that the gravity emanating from them is also pulled to the 2D distribution – the gravitons from the star don’t encounter each other but do encounter the external gravitons that pull the gravity towards the 2D distribution. Integrated over each direction from WB star A to B, the gravity is just like in GR giving your result with alpha_grav ~ 0. But the pull towards the 2D distribution will boost the gravity in the directions that go not or just a little in the 3rd dimension, at the cost of gravity in the directions that go a lot in the 3rd dimension – the SI (1/r) term there is withdrawn from the Newtonian gravity and doesn’t reach the other star anymore, I’d imagine.

            This means for wide binaries that orbit each other in a plane with a greater inclination to the MW disc plane that gravity is much stronger around the ‘horizontal’ position than the other positions, leading to high eccentricity (which Chae’s study filters out, giving him only the orbits that are quite aligned to the MW plane!). Therefore Chae’s results suggest MOND effects. However, your paper doesn’t apply an eccentricity cut as far as I could find, giving all orbit planes and therefore an average of zero deviation from GR.

            Hope this thought analysis is interesting to you all 🙂

            LikeLike

         2. Indranil,
            section 5.3 of your paper states that a sufficient rapid change of the interpolation function could avoid a falsification of MOND. Is it conceivable that WB have a different interpolation function than spiral galaxies? Theoretically I can conceive the MOND/Newton transition as a delta-function ( phase transition?!) for point mass objects like binary stars but much more smooth for extended objects like spirals ( Newton’s “iron theorem” might not work here). I recall Hernandez saying something to the effect that he observed the MOND/Newton transition to be very rapid in his WB sample. My concern is that Chae/ Hernandez and your paper Banik et al. have actually measured very different ” tenets” of MOND. Chae/Hernandez have checked and verified the existence of the acceleration constant a0 in WB systems, whereas Banik et al. have only falsified a smooth transition function in WB systems.

            LikeLike

   4. Hi Indranil, I watched Dr. Becky’s video. (Before coming here. (I mean Dr. Becky… what nerd can resist her?)) In the video she describes you as a MoND proponent. And I thought, yes and no, your ‘preferred’ flavor of MoND (This is my understanding, which is why I’m asking.) is a flavor with some other light-ish particle (A few eV’s.) and so you would be fine with the wide binary test. As it fits your theory. Is that right? Please correct my errors. Thanks, George

      LikeLike

      1. ..with the wide binary test being negative, (only Newton).

         LikeLike

         1. Indeed, my MOND review does advocate a hybrid model with light (11 eV rest energy) sterile neutrinos to fit the cluster data properly:
            arxiv.org/abs/2110.06936

            This would not affect the wide binary test, or indeed galaxy rotation curves. That is why I have barely mentioned the possible existence of hot dark matter in a MOND context when discussing how Milgromian wide binaries should behave. Hope that answers your question, but please reply if not.

            LikeLike

            1. Yes Thankyou. So you have no bias against the WB showing MoND behavior. That does add credence to your paper. Thanks again.

               LikeLike

3. I could honestly see it going either way. MOND is incomplete and needs further modification so it is not exactly the death toll some people say it is for modified gravity if binaries don’t support, just disappointing. I am just baffled and frustrated the results are so all over the place: it would have been nice to have a single consensus by now, but I suppose cosmology is quite messy. Alas, we can’t shove a solar system into a lab.

   What sticks out to me from the paper is the suggestion of an additional parameter based on the phantom dark matter density, which ultimately derives from the matter (hence ‘phantom’). It may not have the cosmological niceness of a0 but it does have a kind of ‘this sounds like what a covariant version of MOND with fields might do’ feel to it.

   To break my pet rule of not posting personal theories since it is at least relevant, I have been toying around for years with the idea of a universe that demands a minimum level of interaction to be met, and it seems the kind of thing that would (indirectly) care about densities whilst also having the side effect of demanding an effective acceleration minimum (since interactions cause accelerations, f ~ ma). On the downside the composition (hot self-interacting gas versus cold dust) would matter in such a theory, but that might not be a bad thing for such scenarios like the Bullet Cluster where all the added gravity appears very much to not be in said hot gas but with the cold matter that flew out mostly collision-free; the concept of an interaction minimum works wonderfully there.

   LikeLiked by 1 person

   1. I’d be interested in reading more about your work, especially in regards to minimal interactions. In my non-professional study of cosmology, I’m rather convinced there’s high degrees of emergent physics driven by layers of relativity. Some versions of MOND may only be describing a portion of the complicated nature of (emergent?) gravity at various levels of interactions; our solar system’s gravity is extremely predictable and leads us to believe gravity should be extremely consistent across the universe, even when we’re constantly presented with data showing that gravity’s surrounding physics are far more nuanced. Those who are attached to “dark matter” need to, at the very least, reconsider renaming it “dark physics.” MOND models on the other hand likely need to focus on more parameters beyond their current EFE – in other words, there’s likely more interactions that aren’t being accounted for; perhaps emergent physics that our predictable solar system can’t/won’t reveal. I’ve been closely studying dense planetary cores, neutron stars, black holes and especially powerful GRBs and their downwind interference with densities – I think we can learn a lot more from these extreme, dynamic systems.

      LikeLike

      1. More parameters? I hope not. Preferably instead a non-fudgable transition function between Newtonian and Milgromian dynamics. With parameters one can only explain more outliers, without them we have exact testable predictions.

         LikeLike

      2. Oh, goodness, I probably shouldn’t have broken my rule. I never did quite work out vigorously enough the densities part and don’t even know if the ‘extra parameter’ part suppressing any MOND effects is even true; rather, the minimal interaction part was motivated by something entirely different, as a result from trying to derive a mathematical system that solves the nothingness problem which has, naturally, some ‘interaction/force like behavior is necessary’ behavior as a result of self equalities of the initial nonclassical 0, although that is a gross oversimplification as the actual statement is pure number theory in a ‘locally binary, globally nonbinary’ logic system and it’s entirely possible I misunderstood how to extrapolate it; I have been far more focused on the pure mathematics foundations and ‘fixing incompatible superpositioned metrics problem’ part and haven’t really focused on the MOND part that seemed to potentially come out as a side effect. I actually think I have made very good progress – but it’s been a years long effort and every time I think I am close to finishing, I get distracted. Also my inner perfectionist isn’t really happy ’bout stuff that isn’t in proof form.
         If I am eventually successful, then I am certain you will hear about my paper since there is a dearth of purely mathematical solutions to the nothingness problem. Technically speaking, it would actually be a 0 parameter model, as I try to derive everything /from/ a 0 (I say a 0 because 0 is not equal NULL, there are multiple 0-likes which is important), but in practice I don’t imagine doing a calculation for 13 billion years for every single particle would really be feasible.

         However, as an amusing teaser and because being a hermit for years kinda sucks, let us imagine a universe that is continually splitting from a single entity until it is at a tipping point where everything is basically worth a Planck mass. The moment it splits further and creates a non-Planck mass and allows for free observable interactions, parameters such as c, h, and electric permittivity of space must be set. The observable universe has 10 to the 60 such masses if we took all the mass and chopped it up that way.
         Add the splits together, 1/2,1/3, etc, imagining that the original fields of the less split versions never actually truly go away (the first field term 1/1 being excluded and uninteresting here because everything is identified with it including uncharged particles). This is the Harmonic function minus 1, which approximates the natural logarithm at a given cut off. Cut it off at 10 to the 60 terms.
         60 x ln(10) – 1 ~ 137.
         The inverse is approx the low energy value of the fine structure constant, and it popped right out. If we imagine some object at low energies ‘seeing’ its relationships as 1 out of this value set by higher energies, we have our explanation for why it’s inverted. Neat, right? I was actually surprised when it came out because I said “Alright, this is my first guess motivated by my model, probably a bit too simple, and using observable masses seems a little weird, because observable, so this will probably be way off…”
         But then it actually worked. So. I don’t know, I feel like this paper is driving me literately insane. I never know when I’ve done enough and it feels like someone else should be writing it, and it sounds nuts to be even trying, but then it keeps outputting cool stuff like that. The only proper thing to do is to finish it up and throw it at some reviewers and they should confirm if it’s, like, my brain just hallucinating solutions or if there’s something concrete to it, or if it’s almost concrete and it needs more work…

         …alright, I apologize for having a mental breakdown publicly and for outputting such a huge post. Thanks for listening I guess. I don’t really have words for how freaky it was to see that 137 pop out, even if it’s just an approximation. This has stressed me out so I’m gonna disappear for awhile (maybe forever) and try to make myself actually do the boring remaining editing work and double-checking.

         LikeLike

         1. Hi Kirk,
            Nice to see a little of what you’re working on! It’s good to see people trying out new ideas. I honestly reacted only to “What sticks out to me from the paper is the suggestion of an additional parameter based on the phantom dark matter density”, I thought this was independent of your pet theory.

            LikeLike

4. So forget statistics for a bit, all you need is one ‘good’ wide binary. Can we find one (or more) ‘good’ WB’s in the data and look at those more?

   LikeLike

   1. You mean like Proxima Centauri? We know its parameters pretty well, but that alone cannot test MOND because if it is slower than the Newtonian escape velocity, it could just be Newtonian. But if it is in between the Newtonian and MOND escape velocities, it could be unbound. A small fraction of unbound systems is to be expected, but this would be rare, so you would have to find more such systems. In general, one should not draw strong conclusions about the behaviour of gravity from any single system, be it a galaxy or a wide binary. But one can alter the tradeoff between data quality and quantity.

      A single system can test MOND if you get the acceleration directly:
      arxiv.org/abs/1906.08264

      This is however not what the wide binary test is about.

      LikeLike

      1. Proxima Centauri? Isn’t that too close to the galactic center and not really in the mond regime of our galaxy? (OK I obviously need to read your paper.)

         LikeLike

   2. This is what Hernandez has emphasized in his contributions – not a single binary, but the very few that are most trustworthy. Figuring out which those are is the trick.

      LikeLike

      1. Right, finding the ‘good’ ones is hard. And I would guess we have to then watch them for a significant part of their orbit.

         LikeLike

      2. if just 1 or a few quality and indisputable wide binary show MOND isn’t it enough

         LikeLike

5. Online lecture series `Octonions, standard model, and unification’ [OSMU23]

   We will have two online talks on `test of law of gravitation using wide binaries’. Each talk will be one hour long, followed by an hour long Q&A session.

   1st talk:
   Kyu-Hun Chae
   Friday, November 24, 2023 at 2.00 pm London time

   2nd talk
   Indranil Banik
   Friday, December 1, 2023 at 4.00 pm London time

   Everyone is welcome to attend. For the ZOOM link please search for OSMU23 and follow the webpage of the lecture series.

   Tejinder Singh
   co-organiser, OSMU23

   LikeLiked by 2 people

   1. You should also hear from Xavier Hernandez, if it can be arranged.

      LikeLiked by 1 person

      1. Definitely. We can do it early year. This year all slots are taken until December 15. Many thanks for your suggestion.

         LikeLike

   2. As I have already said, if you consider the Banik et al. results seriously, I must cancel my talk. You didn’t tell me that you were also inviting Banik, even when I told you my intention. I have found your intention only here. The fact that you invited Banik right after my talk shows that you are considering it seriously. So I must cancel my talk based on what I have said already. As I said in the personal email communication, I’m not interested in persuading people because science has nothing to do with people’s opinions or politics. So, if anyone is wondering who’s results are correct, then he/she should read papers first and then do the analyses themselves, if necessary. This is how I got into this research a year ago. You cannot judge just from discussions, blogs, etc. The reason I cancel my talk is that even if you listen to both talks, you can’t quite decide. To tell the truth, even coauthors can’t quite judge what are lies and what are really true. So, I don’t want to waste my time. I can give a talk only when you want to know details about my analysis and results.

      The accepted version of https://arxiv.org/abs/2309.10404 has a response to Banik et al. It will be available from Tuesday. Appendix B explains why the Banik et al. kinematic cuts don’t matter. Banik et al. made a false claim deliberately based on appearances of v^tilde without detailed calculations. A lot of people seem to be deceived by their v^tilde figures (in particular, their fig 19 and 21) as if they have a real truth.

      Finally, please note that Hernandez and myself cannot respond to all the comments here mainly because we want respond formally with real results, some of which can be found in the new paper mentioned above. Since this is not a political issue, the answer cannot be found by exchange of words.

      LikeLiked by 1 person

      1. Considering the organizer’s advice and my responsibility to the science community, I take back my intention to cancel my talk. I will give the talk as originally scheduled. If you have any questions, please bring any. I’ll be ready to answer all questions without any limits as along as the organizer permits the time.

         LikeLiked by 3 people

         1. Thanks so much Kyu.

            Tejinder

            LikeLike

6. I have seen lots and lots and lots of claims to falsify MOND over the years. Literally hundreds, the vast majority of which go away on close inspection. This has happened so often that I swore to give up fixing other people’s mistakes after the DF2 debacle (https://tritonstation.com/2018/04/04/the-dwarf-galaxy-ngc1052-df2/). Binaries are a new test, so maybe that experience doesn’t apply. Neither do binaries have any bearing on all the other evidence to date: we still need to understand the MONDian phenomenology in galaxies, which dark matter does not and, as far as I can tell, cannot satisfactorily explain.

   Perhaps these results are informative of a deeper theory (including potential hybrids), but first we have to figure out what to believe. Simply choosing the option that one prefers is no option at all. But I don’t see how to resolve this myself, and hope the contending parties can ultimately come to some agreement.

   LikeLike

   1. Thank you so much Dr. McGaugh for hosting this blog. As a lay person, it is fascinating to see the making of the sausage in real time. I’ve always been intrigued by cosmology and galaxy dynamics.

      As a long time lurker, I now enter the factory with trepidation …

      Looking at both papers, albeit cursory, including the latest addition by Chae, it appears to me the results are dependent on the fraction of close binaries estimated in the WB cohort. Banik+’s estimate is over 60%, Chae’s more like 35% or less. Chae highlights this in the addition. The last row in Table 3 in Banik+ is instructive, changing fcb to 30% elicits a mild MOND signal, changing the 16 sigma nominal confidence to 3 sigma in favour of Newton only. Banik+ do have an explanation why this was discarded, but it seems to me that if there were some way to lower fcb in cohort selection it may produce a more robust signal either way.

      But what do I know, I’m just an interested amateur. I like this blog because I believe it tries to tell it like it is, unlike many science communicators. It is also nice to see the authors weigh in directly.

      LikeLiked by 3 people

7. It’s a heated debate, but I like it a lot. Scientists are earnestly trying to solve the problem. They may have a preferred solution, but they’re also open to considering alternatives. And I like that Indranil comments on and explains his paper. And I think the confusion about the results is not accidental and will be a manifestation of the underlying theory..

   LikeLike

8. I remember reading here on Tritonstation that our sun is just within the galactic radius where Newton rules. So it must be that the binary samples in the paper are all outward from the sun’s location where it transitions/enters the MONDian regime.

   With a thin snow cover having melted and the sun shining I’m eager to add to a cumulative 16,775 miles for nine consecutive cycling seasons, before storing the bike till our club reopening in May. So, it’s out the door.

   LikeLike

   1. I forgot to mention MOND’s External Field Effect (EFE) in the prior comment. I had printed out the EFE section of “The MOND Pages” several days ago. From reading that I’m pretty sure even wide binaries are subject to Newtonian dynamics when embedded in the Newtonian regime of our galaxy.

      LikeLike

      1. This is something that also bothers me – in the end, it depends on how distant are the investigated stars compared to the Sun.

         LikeLike

         1. Yes, the ideal place to conduct this test would be deep in the MOND regime. We don’t live there, and so far we can only do it locally. We live at about 1.8 a0, which is on the Newtonian side of the transition region but not entirely Newtonian. So one expects nearby wide binaries to be in the quasi-Newtonian regime dominated by the EFE. In this regime, orbits are Keplerian, but with a higher effective value of Newton’s constant. For where we live, the boost in G (and acceleration) is a factor of ga,,a~1.5. This is what Hernandez and Chae claim to detect. The twiddle-v parameter scales as sqrt(gamma), so roughly a 22% boost. This is the signal Indranil is looking for, and by his analysis should be able to detect but does not.

            LikeLike

            1. Thank you for this clarifying statement

               LikeLike

9. I just watched Sabine Hossenfelder’s latest Science News video on YouTube, where she talks about the Banik paper. Then she makes a comment that I have not heard before – that in galaxy rotation curves, “the higher the uncertainty of the data, the better MOND seems to work” (https://youtu.be/i4lu9AxRtqA?si=QBR4B5-8sOoFz-MY&t=230). Stacy, you are afaict a galaxy rotation curve expert. Any comment on this claim?

   LikeLike

   1. That’s a rather strange assertion from her – I saw it too. As far as I can remember, dr. McGaugh had some sliders in which he showed that as the data improved, most of the predictions were really spot on.
      So I don’t know what to say about Sabine. This video left me with the impression that she just wanted an excuse to ditch MoND and this paper was a perfect opportunity. I’ve seen her in some other videos in which she was opened to MoND, but in this one she seemed, I don’t know how to say it – happy, relieved… that MoND was disproved (by just one paper and in the middle of a debate).
      To me, it looks like confirmation bias at work (she is a proponent / supporter of a specific flavor of dark matter).

      LikeLike

      1. I can’t speak to Sabine’s motivations, but as I’ve mentioned before on my own part, there is an enormous temptation to go along to get along: my life would be a lot easier if I could honestly say that MOND was wrong. Sometimes I wish I had never heard of MOND, then the RAR would be an important empirical discovery that we might get appropriate credit for. Instead, MOND already existed and we had the intellectual honesty to say “gee, Milgrom predicted this.” Dark matter advocates hate that, and they can’t seem to separate their hate for MOND from the messenger of the empirical results.

         LikeLike

   2. This is a classic example of physicists failing to understand astronomical data. Further comment will have to be a whole post, I guess. Sigh – I was hoping to do some work on a new paper today, not explain about this yet again.

      LikeLike

      1. I was hoping to do some work on a new paper today

         what is the topic of your paper ?

         LikeLike

         1. Yesterday’s plan was to look further into the influence of gravitational lensing on the best-fit Hubble constant in the CMB. Today, I’m reminded of Antlia 2, which is a case like Crater 2. So many things to do.

            LikeLike

      2. Stacy, maybe you shouldn’t allow yourself to get triggered by any and all things your scientist colleagues say – and maybe I shouldn’t have alerted you to Sabine’s claim, either . If you have explained it before, just refer to that explanation – or if you feel something new should be said, let it wait and write your paper first. I think it is a good rule of thumb that the more offensive one finds a statement, the longer one should pause and breathe before responding to it 🙂

         LikeLike

         1. Yes, that part of the point of this blog; to have it to point to in such circumstances.

            LikeLike

10. Question – so far, over what mass and distance scales has MOND provided successful predictions?

    Yes, MOND’s scale is set by a0; but a0 can be a result of millions of solar masses and thousands of parsecs or by a few grams and a few centimeters. So, what are the well-established bounds of mass and distance within which we know that MOND works?

    Thanks in advance!

    LikeLike

    1. I would like to merely point out that the no-planet nine MOND explanation of the sednoid anomalies (see https://www.scientificamerican.com/article/modified-gravity-may-make-planet-nine-disappear/) suggests MOND may very well also apply to WB scales.

       LikeLike

    2. MOND works well in galaxies over masses from 100,000 solar masses up to the largest known, nearly a trillion (1E12 Msun). It also seems to work in groups of galaxies, but not in clusters of galaxies, which range from 1E13 to 1E14 solar mass of baryonic material.

       As for distance scales, that depends on the answer to the debate here, as wide binaries probe sub-parsec scales. Setting that aside, it seems to work from sizes of 100 pc up to 1 Mpc. Neither of those are strictly bounded; they’re just the limits we’ve been able to test.

       Since the relevant scale is a0, the corresponding quantity is mass surface density. The lower the surface density, the lower the acceleration, and the more pronounced MOND effects become. This is why I noticed it while studying low surface brightness galaxies: I had unwittingly devised the ideal experiment to test it.

       LikeLike

11. If we put all of Indranil’s work together, we seem to have GR at interstellar scales, MOND at galactic scales, MOND + sterile neutrinos at cluster scales. One way to tie that together might be GR + MONDian dark matter + sterile neutrinos?

    LikeLike

12. Epicycles all over again.

    Pretty sure one could come up with a tweak to basic Newtonian gravity that would explain Mercury orbit. But Einstein didn’t do that. He introduced whole new framework for gravity. Still, obviously, an approximation but a better one than Newton’s. So until somebody with enough imagination and intellect comes along and does the same you are, pardon my French, just leeching.

    Look at what happened to quantum mechanics. Some proposed mechanism explained an observation that couldn’t be explained “in the old tongue”. And then we decided to milk that cow to where it’s on life support.

    Universe didn’t invent math and then decided to orbit clumps of energy around each other following some formula. We invented math and for some inexplicable reason we expect universe to conform. Zero doesn’t exist in the universe. It can’t. And yet our language that was to describe said universe does contain zero. Funny that. Math is extremely useful tool. Tool. Like a knife. Knife can’t feed you but it certainly makes the process of procuring food easier.

    If I showed you a 400 x 400 pixels image of a fractal, could you tell me what was the formula that generated it?

    LikeLike

13. Why do most astrophysicists and cosmologists assume that exactly one particular kind of dark matter would resolve both galaxy dynamics discrepancies and cosmic structure formation discrepancies at the same time? On one hand, the cosmological evidence suggests some kind of non-baryonic dark matter. On the other hand, the evidence from galaxy dynamics suggest some form of MOND or some form of baryonic dark matter which could replicate the effects of MOND at the level of galaxies.

    If Lambda CDM ever gets replaced with some other model of cosmology in the future due to existing tensions like the Hubble tension or the cosmic dipole tensions becoming so severe that it is clear that Lambda CDM is falsified, there is no guarantee that non-baryonic cold dark matter would remain in the successor model of cosmology. However, it seems that the MOND vs dark matter debate would nevertheless continue in that case, since galaxy dynamics, and wide binaries for that matter, have nothing to do with cosmology.

    LikeLike

    1. With each new parameter, dimension, and type of dark matter, Ockham turns once in his grave.

       LikeLike

    2. On the one hand, they do seem like incommensurate problems. MOND makes no claim to explain cosmology, but one would sure like a theory that does. LCDM is built to explain larger scale observations, but does make a claim to explain galaxies, at least so far as they are the “small” structures that form in dark matter halos. The basic predictions fail, and the rest is the deus ex machina of feedback.

       I don’t think we can treat them as separate problems, tempting as that may be. We can’t expect to keep cosmic dark matter out of galaxies if it exists. MOND can’t be a theory we just turn on in galaxies but isn’t active elsewhere. The nuHDM model is a non-starter in this regard. There, one imagines a low enough mass DM particle that it won’t stick to galaxies, but MOND is just grafted on as if it won’t have any impact on larger scales, as it surely will as a longer [effective] range force. I’m sure you could screen it, but that leads to other unsatisfactory situations, like AeST working for the CMB but not really for gravitational lensing.

       So we seem to be up the proverbial creek, with a large part of the community (and perhaps all of it) paddling in the wrong direction.

       LikeLike

14. Is it practical for Gaia to track a handful of WBs for enough time to exclude/confirm their elliptical orbits? Ellipses remain ellipses no matter how you rotate them in 3D, hence this test eliminates the great uncertainties in mass/light ratio and eccentricity, as well as the tacit assumption that the planes of WBs are isotropically distributed despite there being a preferred global (galactic) orientation.

    LikeLike

    1. I’m no expert, but a write-up on Sky and Telescope from 2021 shows that the red dwarf Proxima Centauri C orbits the A and B components of the triple star system in 550,000 years. It also indicates that Proxima Centauri C is 13k astronomical units (a.u.) from its A/B companions which are relatively close together and each about the size of our sun. Since the Banik et. al. paper covers binaries that are 2k to 30k a.u. apart, that orbital period must be roughly typical of their binary star samples. So it probably wouldn’t be practical, even for their closely spaced binary samples, to track them for an extended period to garner info on the degree of their orbital ellipticity.

       LikeLiked by 1 person

       1. According to Banik and Kroupa (https://ui.adsabs.harvard.edu/abs/2019MNRAS.487.1653B/abstract), this requires a precision at least ten times better than that of GAIA:
          “The wide binary orbit of Proxima Centauri around α Centauri A and B differs significantly between Newtonian and Milgromian dynamics (MOND). By combining previous calculations of this effect with mock observations generated using a Monte Carlo procedure, we show that this prediction can be tested using high precision astrometry of Proxima Centauri. This requires ≈10 yr of observations at an individual epoch precision of 0.5 μas, within the design specifications of the proposed Theia mission. In general, the required duration should scale as the 2/5 power of the astrometric precision.”
          That’s why
          “Instead of using the acceleration, it is much more promising to use the relative velocity, but then the wide binary test (WBT) of gravity has to be done statistically by considering a large number of WBs to average over orbital phases and projection effects.” (Banik et al., 2023).
          But if the various teams can’t agree on the results of this statistical WBT, we may have to wait for the hypothetical (?) Theia mission…

          LikeLike

15. Indranil Banik wrote an article for IAI on MOND being falsified:

    https://iai.tv/articles/the-challenge-to-dark-matter-mond-is-wrong-auid-2676

    LikeLiked by 1 person

    1. I finally realized what I don’t like in that piece.

       The article is written for an educated audience – for people who know what is the issue with MoND vs LCDM and understand that if MoND is falsified (as Indranil Banik is saying), this does not make LCDM correct.

       But the general audience does not know that. The general audience sees that MoND is falsified so it means that LCDM is good. I wonder what will Ethan Siegel comment on this (I’m quite confident he will).

       LikeLike

16. Pingback: A post in which some value judgements are made about the situation with wide binaries – Triton Station

17. Thanks Madeleine for bringing this article by Dr. Banik to our attention. Since it was intended for a general audience, and beautifully written, it was very easy to digest. The mention of a solar system test of MOND, from precise Cassini Mission data, is something entirely new to me. Definitely want to look into that. One thing that bothers me, though, is the astounding degree to which MOND is excluded in the wide binary study. As Dr. Becky put it – the chances of this being a statistical fluke are one part in a 100,000, trillion, trillion, trillion. Considering the nature of astronomical data this seems extravagant, particularly in comparison to the field of particle physics, where a 5 sigma signal (one part in 3.5 million, if memory serves) is considered the threshold for a discovery. But then I don’t remotely understand the statistical analysis that Dr. Banik and his colleagues clearly do, so that concern may be entirely mute.

    LikeLike

    1. If the MOND regime is an emergent property/behavior of a big assembly of stars as galaxies are then wide binaries star systems will not provide a definitive answer, exactly as a small set of metallic molecules will not exhibit rigid metallic properties.

       But obviously only consistent independent observations will settle this problem.

       LikeLike

       1. metallic molecules -> metallic atoms.

          LikeLike

18. Read Stacy’s post today (“A post in which some value judgements are made about the situation with wide binaries”)

    LikeLike

    1. wide binaries need to get resolved before any talk of MOND regime

       LikeLike

    2. Stacy McGaugh hasn’t enabled comments on the newest blog post yet.

       LikeLike

Comments are closed.