Physicists Move One Step Closer to a Theoretical Showdown

August 10, 2023 11:09 AM

The deviance of a tiny particle called the muon might prove that one of the most well-tested theories in physics is incomplete. (NYT gift link)
posted by praemunire (53 comments total)

Interesting, thanks praemunire! Here's Fermilab's Muon g-2 page with press releases and such, and here's an article at Scientific American with a slightly different general audience take from the NYT link in the post.
posted by indexy at 11:23 AM on August 10, 2023 [2 favorites]

Also, the name of my next band is Deviant Muons, and our album Standard Model Showdown drops next year.
posted by indexy at 11:24 AM on August 10, 2023 [26 favorites]

Take my money, indexy. Please!
posted by Quasirandom at 11:54 AM on August 10, 2023 [3 favorites]

I love watching muons go by in the cloud chamber at the Exploratorium.
posted by njohnson23 at 12:00 PM on August 10, 2023 [1 favorite]

Every few weeks someone gets all press-releasey on *OMG the Standard Model is toast!* and then nothing comes of it.

Of course the Standard Model is incomplete -- if it were complete it would be compatible with Relativity.

But I must admit after about the 1000th time someone makes huge waves about the imminent demise of the Standard Model, I'm sorely tempted to just ask each one to bet me $1 that it'll come to nothing. I'd have a couple hundred bucks by now if I started just a couple years ago.
posted by tclark at 12:05 PM on August 10, 2023 [4 favorites]

Fascinating news, and pretty well reported by the NYT. The SciAm article helps get a different perspective. I'm still a bit lost on the details; do we have any domain experts who can comment? I get that there are several different ways to calculate g-2, some based more on theory and some based on experiments. And all of them agree (including this new one?) except the older, most familiar method? Are the details of that interesting? Why?
posted by Nelson at 12:10 PM on August 10, 2023 [2 favorites]

i agree with nelson. for these types of articles, as an engineering-friendly art school graduate, i typically read and do my best to make sense of things, but then come back here to really understand by absorbing everyones' posts. :)
posted by rude.boy at 12:35 PM on August 10, 2023 [3 favorites]

and here's an article at Scientific American with a slightly different general audience take from the NYT link in the post.

Hadrons interact recursively with themselves such that they create what physicists call a “hadronic blob,”

Aight, now they're just messing with me.
posted by clawsoon at 12:35 PM on August 10, 2023 [7 favorites]

In all honesty, I posted this not because I had perfect understanding, but in the hopes of luring in some local physicists to comment further.
posted by praemunire at 12:51 PM on August 10, 2023 [9 favorites]

I'm qualified to teach high-school physics, but quantum vacuum theory is above me. On the sidebar of the Fermilab page I posted above are some links which may be helpful for further reading.

clawsoon: Hadrons interact recursively with themselves such that they create what physicists call a “hadronic blob,”

Aight, now they're just messing with me.

No, it's true! Hadronic Blob will be opening for Deviant Muons on our tour.
posted by indexy at 1:00 PM on August 10, 2023 [27 favorites]

No, it's true! Hadronic Blob will be opening for Deviant Muons on our tour.

They’d be a better band if they fixed all that instrument feedback they have.
posted by notoriety public at 1:03 PM on August 10, 2023 [12 favorites]

Well, a narrower band, anyway.
posted by The Bellman at 1:04 PM on August 10, 2023 [38 favorites]

I'm trying to figure out if this means that a) the current theory is correct and complete, but you can only see that if you do months of supercomputer calculations with great precision, or b) there are particles out there that we haven't detected yet.
posted by clawsoon at 1:18 PM on August 10, 2023 [1 favorite]

Muons … that’s what cows are made of, right?
posted by caviar2d2 at 1:26 PM on August 10, 2023 [17 favorites]

clawsoon, if I'm reading correctly, the traditional way of calculating muon g-2 which uses the standard model and years of experimental data differs from the new supercomputer lattice method, and theoretical physicists aren't sure why. But since the lattice method matches the Fermilab data quite well, they're not sure if this means new particles are out there or if the method of traditional calculating needs tweaking. So the ball's now in the theory side of the court. But I could be wrong about that.
posted by indexy at 1:33 PM on August 10, 2023 [1 favorite]

But...but... will that discrepancy get us to functional FTL?
posted by sammyo at 1:37 PM on August 10, 2023 [1 favorite]

The vague descriptions of the lattice method are reminding me of my vague understanding of computational fluid dynamics, in which the correct equations can't be solved analytically so you have to put oodles of computing power to work in order to get approximate numerical solutions for things like lift and drag. Is there any value to that analogy, or am I completely off-base?
posted by clawsoon at 1:48 PM on August 10, 2023 [2 favorites]

Here's an article in Nature on the lattice method, but it's not for the timid. May be paywalled, as I'm at work on a university computer.
posted by indexy at 2:02 PM on August 10, 2023 [1 favorite]

I'm sure the PBS Space Time YouTube channel has done something on Lattice computational modeling at some point in the past.
posted by zengargoyle at 2:03 PM on August 10, 2023 [5 favorites]

Muons … that’s what cows are made of, right?

Only the spherical ones.
posted by Greg_Ace at 2:05 PM on August 10, 2023 [20 favorites]

Only the spherical ones.

I think we can all safely make that assumption
posted by Insert Clever Name Here at 2:09 PM on August 10, 2023 [3 favorites]

Q: Does a particle have a Buddha nature?
A: μ

Mu-on: Apply directly to forehead!
posted by indexy at 2:11 PM on August 10, 2023 [8 favorites]

I'm sure the PBS Space Time YouTube channel has done something on Lattice computational modeling at some point in the past.

I'm watching the video they put out about this stuff the last time that results were announced a couple of years ago: Why the Muon g-2 Results Are So Exciting

I'm just up to the part where they're explaining it all with electrons. Apparently if an electron could only interact with a magnetic field in one way, represented by one Feynman diagram, g would exactly equal 2. But each additional interaction possibility bumps the number up a tiny bit, and:

Each layer of complication involve many more Feynman diagrams, but also added less and less to the interaction. The latest calculations rely on powerful computers to add many thousands of Feynman diagrams, and get us our g-factor to 12 significant figures. And a latest calculated electron g-factor of 2.001159652181643.

That was as of 2 years ago. I'm still not sure if they're talking about lattice methods or the traditional calculations, but the video is definitely going into more depth than either article.
posted by clawsoon at 2:11 PM on August 10, 2023 [2 favorites]

...hadronic blob...

That’s the stuff in a lava lamp, right?
posted by Thorzdad at 2:16 PM on August 10, 2023 [2 favorites]

Goddamnit, those Trisolarians keep messing with everything.
posted by danhon at 2:43 PM on August 10, 2023 [2 favorites]

Here's an article in Nature on the lattice method, but it's not for the timid.

Skimming (and not understanding 90% of it), what I think I'm getting is that the normal way to calculate it is to use simpler formulas combined with more experimental data, while the way that they're doing it is to use more complicated formulas combined with less experimental data.

(...the less complicated formulas being quantum electrodynamics, and the more complicated formulas being quantum chromodynamics, which is a generalization of quantum electrodynamics... but also they use some things from QED that I guess aren't in QCD? And then there are a whole bunch of sources of error in the calculations that they have to account for, in part because they're chopping the universe into little boxes and doing the calculations on those little boxes, and the smaller you make the boxes the more accurate you get but also the longer the calculation takes.)

Or something like that?
posted by clawsoon at 2:49 PM on August 10, 2023

OK, I just fell down a rabbithole because "supercomputers" and "lattice QCD" proved too tempting.

Back in the early 2000s, lattice QCD looks like it ran on your regular garden massively parallel supercomputers, until some researchers figured out a way to network a bunch of custom ASICs and deliver around 10 tflops of compute, back in 2005. They called that system QCDOC, for quantum chromodynamics-on-a-chip.

(Fifteen years later, a 2020-era Apple M1 Pro with a 14-core GPU will hit around 4.6 double-precision tflops -- half a QCDOC -- but will do it in the space of, well, a laptop, and 15 watts for around $2,000, versus the 100 square feet, 100 kilowatts and $5m that QCDOC took).

But as soon as someone starts referencing the tflops and precision needed, along with the timeframe, you might be able to guess where this ends up: in 2006, some researchers noticed too: the type of calculations needed for lattice QCD look *awfully* like the kinds of calculations videogame GPUs were starting to optimize for.

Doing the calculations for lattice QCD was one of the uses of parallel variant Cell Broadband Engine processors (the ones in the PlayStation 3!), on a system called QPACE, in the early 2010s. QPACE hit 200 Tflops in 2009, and was then followed by QPACE 2, which by then had switched to running on Xeon Phi (Knights Corner) processors and 310 tflops around 2015.

(Xeon Phi itself was interesting because it was Intel's attempt to produce something like a general purpose GPU but on x86-based cores. It didn't work out; Nvidia ate their lunch).

After that, in 2012, lattice QCD was run on supercomputers like Mira Mira (also known as IBM's Blue Gene/Q), in the class of petascale computing (which was a big deal at the time!) at 8.5 petaflops. Mira got retired in 2019, and now the calculations are being run on Aurora (an exascale supercomputer!). One nice thing that's come out of this progression is that high performance computing theoretical physicists are (rightly) annoyed at having to be locked into CUDA, so are looking into replacing it with oneAPI.
posted by danhon at 3:18 PM on August 10, 2023 [10 favorites]

I'm sure the PBS Space Time YouTube channel has done something on Lattice computational modeling at some point in the past.

This appears to be the one where they go into it most deeply:

What Happens Inside a Proton?

The lattice method:

1. Pixelate spacetime.
2. Mix in Monte Carlo random sampling.
3. Pretend that time is another dimension of space and pixelate it.
4. Run your calculations with a bunch of different pixel sizes.
posted by clawsoon at 3:21 PM on August 10, 2023 [4 favorites]

This is so far over my head, I have but one question. Is there any practical issues I need to concern myself with at this time?
posted by JohnnyGunn at 4:00 PM on August 10, 2023 [2 favorites]

Muons … that’s what cows are made of, right?

Mewons being similar, but for cats.
posted by GenjiandProust at 4:22 PM on August 10, 2023 [4 favorites]

Maybe this is the way we find dark matter, it's also being spit out of the the quantum foam, but the current calculations don't include it ....
posted by mbo at 4:40 PM on August 10, 2023 [4 favorites]

I took issue with the bit in the article mentioning gravity because I can't believe that anyone expects a TOE to arise from this. I see that a GUT, which really could result from this eventually, could be an important step toward a TOE, but to me this speculation reads like the underpants gnomes. I DM'd a close friend who's an (experimental) particle physicist, so I expect to be humbled soon.

Edited to add: Very much I think this could lead to an explanation for dark matter. Much less so dark energy or a QM explanation of gravity.

(For those not familiar, quantum chromodynamics — for which Feynman and Gell-Mann won a Nobel Prize — doesn't include gravity and, in fact, the fundamental problem is that General Relativity doesn't see gravity as a force mediated by a particle, but a fundamental property of the curvature of spacetime that's simply a function of mass. GR is rock-solid. QM including gravity would need to not conflict with GR, not the other way around.)
posted by Ivan Fyodorovich at 4:58 PM on August 10, 2023 [1 favorite]

Oh, look, a nit! Let's pick it: Feynman won the Nobel for QED, quantum electrodynamics, with Sin-Itiro Tomonaga and Julian Schwinger in 1965, not with Murray Gell-Mann in 1969 for QCD.
posted by indexy at 5:08 PM on August 10, 2023 [2 favorites]

Hadrons interact recursively with themselves such that they create what physicists call a “hadronic blob,”

It’s perfectly healthy and natural to interact with yourself, even if it makes a bit of a mess.
posted by dephlogisticated at 5:15 PM on August 10, 2023 [3 favorites]

“Oh, look, a nit! Let's pick it”

Wow, I've thought they shared the Nobel for QCD for thirty years. Thanks so much for picking this nit — I'm grateful for the correction. I guess I conflated the history of the development of QCD with their respective Nobels.
posted by Ivan Fyodorovich at 5:20 PM on August 10, 2023 [2 favorites]

I'm now enjoying thinking about how both Feynman and Gell-Mann would react at the thought of sharing their Nobels with the other ;)
posted by indexy at 5:30 PM on August 10, 2023 [4 favorites]

Is there any practical issues I need to concern myself with at this time?

As long as your muons are securely fastened and not wobbling excessively, you should be fine.
posted by paper chromatographologist at 5:33 PM on August 10, 2023 [8 favorites]

So, it sounds like the experimental result is in really precise agreement with a theoretically-calculated value based on the Standard Model. The theoretical number is different from the previously-accepted value which is partly based on experimental values, which has people confused.

But it seems like the value with which there is such good agreement is one based on the Standard Model with no extra kinds of particles? Which sounds like it tends to confirm that the Standard Model accounts for all of the kinds of matter, at least ones that can interact with muons?
posted by Aardvark Cheeselog at 7:21 PM on August 10, 2023 [2 favorites]

More evidence for a new physics
posted by blue shadows at 11:13 PM on August 10, 2023 [1 favorite]

I think I'm physically incapable of not reading hadron as hardon.
posted by Pyrogenesis at 11:59 PM on August 10, 2023 [2 favorites]

I'm happy to serve as the Friendly Neighborhood Particle Physicist for this thread. I'll write a bunch of short comments instead of one massive one, just to keep myself from getting too deep in the weeds. Questions welcome.
posted by fantabulous timewaster at 5:14 AM on August 11, 2023 [4 favorites]

> Every few weeks someone gets all press-releasey on *OMG the Standard Model is toast!* and then nothing comes of it.

This is the ghost of Thomas Kuhn's mental picture of the "paradigm shift": the idea that progress in science is made in little fits and starts, leading to a growing, gnawing confusion and discomfort, then suddenly an overwhelming burst of devastating creativity. People who are science spectators and who also like to tell stories, such as science journalists, are desperate to get in on the ground floor of the next revolution.

It's like if you left a city council meeting where your community came to a compromise on some thorny question about garbage collection, and the headline in your next morning's local paper was "Our District Seven Alderwoman Might Become America's First Female President." Which might be true — Katie from D7 is great, and she clearly has a goal of parlaying her local leadership experience and community-organizing skills into something broader, and she'd be a welcome breath of fresh air in national politics. But it's not really what happened at the city council meeting.

> Of course the Standard Model is incomplete -- if it were complete it would be compatible with Relativity.

I had to read this about three times, because the standard model of particle physics (a name which is less magical if you don't capitalize it, reminding the reader that there are other possible models, one of which just happens to work the best) is what happens when you combine the tools of quantum mechanics with special relativity. This was first done by Dirac, who discovered that special relativity brings with it a hidden symmetry which permits/requires the algebra of "spinors." Spinor algebra describes electron spin and also predicts antimatter. Headline Physics gets all in a knot about the conflict between (special) relativistic quantum mechanics versus general relativity, which describes gravity.

For my money, results like the muon g−2 are best equipped to give hints about dark energy. The standard model is all about the symmetries of the vacuum, and which particle-like excitations of the vacuum are allowed. The standard answer (note how "Standard Answer" sounds much more pretentious than "Standard Model") is that the symmetries of the vacuum are U(1)×SU(2)×SU(3). This is an infuriatingly glib sentence that took me about ten years to understand, but it does turn out to be closely related to the more common pop-physics diagram of the different flavors of quarks, leptons, and force-carrying bosons.

The conflict between the standard particle physics model versus general relativity is something that only becomes interesting in regions where gravity is very, very strong — that is, where there's a lot of spacetime curvature. But dark energy is most important in the empty regions between galaxies, where gravity is weak and spacetime is very flat. The muons in the g−2 experiment are interacting with that mostly-flat spacetime here on Earth.

[short comments, my eye]
posted by fantabulous timewaster at 5:56 AM on August 11, 2023 [14 favorites]

MetaFilter: This is so far over my head, I have but one question. Is there any practical issues I need to concern myself with at this time?

(Maybe put that on any new t-shirts for fund-raising?)
posted by wenestvedt at 6:59 AM on August 11, 2023 [1 favorite]

Alternate takes in the popular-but-literate press today. Nature: Dreams of new physics fade with latest muon magnetism result, leaning heavily on the "this new measurement confirms our new simulation-derived result". The Guardian: Scientists may be on brink of discovering fifth force of nature, lots of quotes from scientists hyping the idea that there's something novel here but never quite explaining what it might be. Perhaps the ghost of Thomas Kuhn visited them recently.

The Wikipedia article is giving me a clearer picture of the stage of g-2 measurements than any of the articles I read. At least it has a clear "one section per measurement" structure in the Timeline part.
posted by Nelson at 7:07 AM on August 11, 2023 [1 favorite]

The Scientific American article is a little confusing to the non-specialist. It's presented as a news story, but it contains one "new" and two "olds."

• The Fermilab g−2 experiment has released their new measurement of the muon's magnetic moment. The difference from their old measurement is that the new measurement includes more data. But the result isn't different; it's the precision that's different.

Measurements that get better with more data are confusing to non-scientists, because in clothesmaking or construction you make a measurement and then you're done. But physics measurements are statistical, like opinion polls. Imagine if Rasmussen or Gallup or whoever released poll results in pieces, as they talked to more people. You'd have an initial "commissioning" result, "We asked 100 people this question, and got 61% yes and 39% no." Then later, a final result: "We asked 10 000 people this question, and got 61.1% yes, 38.9% no, and 0.02% undecided." These two polls are consistent with each other, but you can have more confidence in the second one. If there were an independent poll that also asked a different 10 000 people, you would be unsurprised to get 59.8% to 40.2%. But if your independent 10 000-person poll got a split like 75-25 instead of like 60-40, you would sit up, scribble on a napkin about the width of the Poisson distribution, and decide that the 75-25 was almost certainly not measuring the same thing.

In the muon case, Fermilab injected muons into a storage ring with their spins "polarized," so that all of their north poles were pointed the same way. In the magnetic field of the storage ring, the muon's spins precess, like the spin of a top precesses before the top falls over. If you know the magnetic field of the storage ring, then the rate of the precession tells you the muon's magnetic moment. The statistical component comes because you can't measure the muon's polarization directly; you have to let it decay, and look at the directions of the decay products.

• The experiment was only interesting because we already had a prediction for the result. When the preliminary g−2 result came out a few years ago, that prediction turned out to have been wrong.

This is the exciting piece, but it's olds, not news. The g−2 is interesting because "a muon" and "a magnetic field" are both approximations. When we say that a muon interacts with a magnetic field by exchanging photons, that's true enough to be useful. Most people are relatively comfortable with the idea that the electromagnetic field is a continuous property of empty space, and also comfortable with the idea that "a photon" is somehow a discrete excitation of this field. But the "muon field" is also a continuous property of empty space, and "a muon" is likewise a discrete excitation of the muon field. In the standard U(1)×SU(2)×SU(3) model, you're able to talk about "a muon" only if you're prepared to admit that all of the other fields are still involved.

Muon precession in a magnetic field is mostly an interaction between the muon field and the electromagnetic field. But it also includes the electron field, which gives the first-order result g−2 ≈ α/π ≈ 0.002. And it also includes the proton field, the top-quark field, and presumably whatever field corresponds to dark energy. When people say that the muon g−2 is sensitive to "new physics," this is what they mean.

• After the preliminary g−2 experimental result was released, a lattice QCD group published a new calculation that "fixed" the disagreement between prediction and experiment.

Also olds. As someone who knows a lot about the field but not a lot about the details of lattice QCD, I'm deeply skeptical about this result, which was a post-diction rather than a pre-diction. The trouble with doing a calculation where you think you know the answer is that, once you've gotten the answer "right," you can stop there.

Experimentalists in physics have developed a strong caution against this bias. The standard in a big experiment is to include some unknown "blinding factor" in your analysis, so that you know all of your initial analysis is not going to give you the final answer. Only when you have decided based on other reasons that all of your analyses and corrections are in place do you "un-blind" your result — and then you publish it, whether you like it or not. That's the ceremony described in the NYT story.

My experience has been that this culture is not as strong in the theoretical/calculational community, compared to the experimental community. So if the theorists are coming to a consensus on the BMW post-diction, then this particular story is a dead end for now. Someone upthread put it nicely. Hadronic physics — the details of the fields corresponding to virtual protons and neutrons and heavy quarks — is the "turbulence" of particle physics. It's very messy, and there is a risk of being able to "predict" anything you like. The next challenge will be to take whatever innovations went into the "fixed prediction" and use them to make a precision prediction about some other system. But we are at the point where the precision predictions usually require ten-year experiments to test.
posted by fantabulous timewaster at 7:25 AM on August 11, 2023 [10 favorites]

I think I'm physically incapable of not reading hadron as hardon.

One wonders what you think the CERN physicists are getting up to over at the LHC.
posted by The Bellman at 8:03 AM on August 11, 2023 [2 favorites]

Fundamental Physics is in a lot of trouble. They're pulling gob smacking amounts of money but have very little to show for it.

One of the reasons the (probably not real) superconductor news got so much attention, IMO, is that we're really desperate for evidence that there's transformative discoveries to make.
posted by constraint at 9:34 AM on August 11, 2023

fantabulous timewaster: Muon precession in a magnetic field is mostly an interaction between the muon field and the electromagnetic field. But it also includes the electron field, which gives the first-order result g−2 ≈ α/π ≈ 0.002. And it also includes the proton field, the top-quark field, and presumably whatever field corresponds to dark energy.

I'm still waiting for the standard model to include a potato field.
posted by clawsoon at 2:10 PM on August 11, 2023 [1 favorite]

fantabulous timewaster, with all of these fields interacting with each other, is there some sense (mathematical or otherwise) in which they're all parts/aspects of the same underlying thing? I.e. is the separation into separate fields mostly a mathematical convenience?
posted by clawsoon at 5:22 PM on August 11, 2023

As someone who knows a lot about the field but not a lot about the details of lattice QCD, I'm deeply skeptical about this result, which was a post-diction rather than a pre-diction. The trouble with doing a calculation where you think you know the answer is that, once you've gotten the answer "right," you can stop there.

I'm familiar with lattice QCD so I'll jump in. The BMW result was NOT a postdiction: the result was made public a year before the fermilab results were available, and had been expected to agree with the previous SM result/dispersive calculation using experimental data. The result was far enough from expectation that they gave time for the community to make comments and suggestions before formally publishing in the week before the fermilab result was made public. The Scientific American article muddles the timeline, in part because the BMW result was incredibly political and often dismissed by anyone with a stake in the status quo. It's gained broader public support more recently (including grudgingly from the 2020 g-2 white paper team) as other groups and methods have had a chance to weigh in and corroborate the correctness of the result. (this took a long time because lattice is so computationally expensive)

It's possible that the reported BMW result is in fact lower than it should be, because people were worried its deviation from the previous SM value was a mistake and thus some changes were made in the yearlong review. You'll be pleased to know that a lot of the lattice community is now using blinding in their process to remove these kinds of "expectation biases" in future work.

Long story short: BMW was snubbed and downplayed last Fermilab press conference to push the "possible new physics discovered at Fermilab" angle, I am glad to see the collab actually got a fair mention this time around.
posted by irrediated at 8:09 PM on August 11, 2023 [3 favorites]

> is there some sense (mathematical or otherwise) in which they're all parts/aspects of the same underlying thing?

Yes. The underlying thing is reality, of which we have only one. The separation into different mostly-independent fields is entirely for the convenience of human comprehension.

For an analogy, perhaps consider that ice water is made of protons, neutrons, and electrons, and that a hot dog is also made of protons, neutrons, and electrons. But there are several intermediate stages of interaction — nuclei, atoms, molecules — which make it really inconvenient to suppose that water and hot dogs are "parts of the same thing." If there is some deeper interaction which means that electrons and charm quarks are "parts of the same thing," we don't have any evidence for it.

> The BMW result was NOT a postdiction:

I was not aware that I had misunderstood this timeline, and I apologize if I have helped promulgate an untruth.
posted by fantabulous timewaster at 9:18 PM on August 11, 2023 [2 favorites]

fantabulous timewaster, irrediated, and any other physicist MeFites

As a non-physicist (& non-mathematician) I follow Sean Carroll, PBS Spacetime (Matt O'Dowd), and Sabine Hossenfelder on YT to try gaining some feeble grasp of the state of basic physics and where it is going. Also Huygens Optics for the more optics based stuff.

Are they good value? Any others you would recommend?
posted by Pouteria at 9:36 PM on August 11, 2023

Becky Smethurst ("Dr. Becky") runs an excellent astronomy-centric YouTube channel.
posted by fantabulous timewaster at 9:34 PM on August 15, 2023 [1 favorite]