Result from LHCb experiment challenges the Standard Model
bbc.co.uk
bbc.co.uk
https://theconversation.com/amp/evidence-of-brand-new-physic...
The non AMP page is 1.4MB and 45 requests, the AMP page is 400KB and 16 requests - normally this means AMP just blows everything out of the water (despite HN hate). Something is messed up here that AMP is so poor, initial server response is terrible.
Meat:
> The LHCb produces sub-atomic particles called "beauty quarks", which are not usually found in nature but are produced at the LHC. Sub-atomic particles undergo a process known as decay, where one particle transforms into several, less massive ones. According to the Standard Model, beauty quarks should decay into equal numbers of electron and muon particles. Instead, the process yields more electrons than muons. One possible explanation is that an as-yet undiscovered particle known as a leptoquark was involved in the decay process and made it easier to produce electrons.
I only posted because it was the first thing I wanted to know, and I had to dig around in the article for a bit before I found it at the end.
This is not a bad thing, by the way - just an inevitable consequence of doing many searches.
As in 3-sigma ~= 1/1000
I guess that only applies if there are no mistakes or biases at all?
Video News release : : https://videos.cern.ch/record/2758757
LHCb article : https://lhcb-public.web.cern.ch/Welcome.html#RK2021
LHCb paper : https://arxiv.org/abs/2103.11769
https://www.ukri.org/news/result-from-lhcb-experiment-challe...
Where the universities involved are
Bristol
University of Cambridge
Imperial College London.And here is a post by some of the main authors describing their findings: https://theconversation.com/evidence-of-brand-new-physics-at...
Particle physics, the field where you can make a difference... ;-)
To this non-physicist, it's not clear at all how this potential discovery might do any of that. From here, it looks like a mininally consequential addendum to the standard model, almost like a single epicycle to a geocentric cosmology. The effect only shifts by a few percent the chance of some obscure and rare interaction. Is it really plausible that it could have such grand implications?
The problem in modern physics is that the epicycles work too well. There are some big obvious mysteries and an incredibly-successful epicycle-based theory. Essentially everyone expects that the epicycles are not the truly fundamental theory of things.
So, physicists keep comparing the predictions of the "epicycle-theory" with reality in new ways, ways where a deviation from the prediction would be an unambiguous signpost pointing a direction toward which one might be able to address a big mystery. As those tests become established, they are then performed with higher and higher precision.
The signal potentially seen with LHCb is, if it persists, important. In a nutshell, electrons are, in this particular way, predicted to act exactly like muons. No ifs, ands, or buts. What they see is that electrons and muons might be acting slightly differently. If so, some new physics (or a major error in our understanding of the epicycles -- I'm looking at you, loops) must be at work. If this result persists, it is guaranteed that we will learn something.
Too convoluted? Here is a different analogy:
You're pretty sure someone is embezzling from your company.
You look at your bank account, and you have ~95% less money than you thought you should. Really. Oh, no.
You check the summary budgets, they're balanced. So you do this for all the company's divisions. Yep. Still balanced. You get paranoid. You do this for decades -- the books balance, but at the end of the day? 5%.
Then an employee runs up to you and says, "Shmageggy! We're not sure yet, but it looks like one of our most-trusted partners, whom the firm has worked with since 1936, transferred thirty-seven cents to an account in the Cayman Islands!"
Until we find the embezzler, that's what every one of these precision physics tests that makes the news is all about.
There are well-known places where we think the armor might be weak and we're trying all of them roughly equally. The moment someone finds an actual hole in the armor, a lot of people will jump in and hammer on that one in particular.
The key properties of the weak points are:
a) We have to be able to hit it there in a meaningful way
b) Anywhere that we expend millions of dollars and oodles of people-years of work to look for holes, we have to be capable of verifying that a possible-hole is an actual-hole.
This particular anomaly scores pretty well on b). If they have a clean signal there, then it is a pretty big deal.
Edit: https://arxiv.org/pdf/2103.11769.pdf - references 54 to 90 are about potential new physics models.
Does top quark have its antecedent too?
Also, Some physicists call the fourth, fifth, and sixth derivative of position, "snap, crackle, and pop" respectively. Physicists are a fun lot aren't they?
Not defending BBC's using non-prescribed terms, but quarks were always named whimsically. The word quark itself having been taken from Ulysses.
It's hard to argue with snap following jerk, then acceleration then velocity. The next two follow snap, if not physically then at least historically.
https://en.wikipedia.org/wiki/Fourth,_fifth,_and_sixth_deriv...
I thought were all useless terms but apparently snap is quite commonly used in quadrotor motion control.
I've also heard this is important for self-driving convoys. You want the cars in the convoy to all accelerate and brake smoothly rather than jostle all the things inside them.
That's third derivative of position and called jerk.
Atomic cross section is measured in barns. As in "you couldn't hit a barn door".
https://home.cern/science/experiments/lhcb
Moreover, the paper that describes the result that this article is discussing has "beauty quark" in the name: https://arxiv.org/abs/2103.11769
Right there with developers but our terrible naming conventions don't really make it to publications.
On the other hand, everyone else who discovers a hominin still gives it a new species name.
The various quantum numbers of the quarks are real physical quantities, but not ones for which we have a good way of conceptualizing in classical, macroscopic life. The up and down quarks aren't actually spinning, but they do kind of behave as if they were spinning, so their names make sense, but for the others there isn't even an analogue. Luckily no one expects to be able to visualize strangeness, and likewise charm and beauty convey quite well that the naming is arbitrary. Top and bottom on the other hand sound like they could refer to a "real" property, or perhaps represent two ends of a spectrum. While I've seen no evidence that confusion is rampant, it would be easy to see how confusion could arise. If anything, the Top quark should be renamed to something more whimsical, or both should be renamed to something more clearly arbitrary like heads and tails.
The fact that "up" is associated with positivity is totally arbitrary, and even the fact that positive charges are called positive is totally arbitrary.
I’ve seen “sideways” sub for strange, which actually seems to fit the mnemonic better anyway. Why not “curly” or something instead of charm?
In particular, collider experiments were giving rise to a 'particle zoo' of pions, kaons, lambdas, etc. all of which were assumed to be fundamental. Some of these lasted much longer than expected, and were called "strange".
It turned out that these could be modelled by giving fundamental particles a new property, which got the name "strangeness". Similar to how particles can have "charge", which is zero for neutral particles; non-strange particles have zero strangeness. Their long lifetime was explained due to the strong force conserving strangeness: collisions between non-strange particles can make pairs of particles with positive/negative strangeness (as long as it sums to zero); but once those individual particles have separated they can no longer decay via the strong force; they have to wait for the weak force, which gives them a longer half-life.
It was only later, once the quark model became more accepted, that these particles were no longer considered fundamental, and "strangeness" could be explained as "containing strange quarks".
On the other hand charm, beauty/bottom and truth/top seem to all be built on the quark model from the start; which makes the whimsy in their names more deliberate.
---
FYI the BBC's Horizon did an episode in the 60s called "Strangeness Minus Three" ( https://www.bbc.co.uk/programmes/p01z4p1j ) which is interesting to watch in hindsight, now that we have quarks and the standard model; and makes a nice comparison to the more recent search for the Higgs boson.
---
Tangentially this idea, of a new property whose conservation prevents massive particles from decaying, also arises in supersymmetric models of dark matter. There, supersymmetric particles have a different "R parity" to normal particles, which prevents them decaying even if they're very massive; such particles are a candidate for dark matter.
They still teach it for some reason though.
The particle does not have a name in the official (PDG) listing, only a symbol "b". The "bottom" quantum number is universally called as such so one talks of "bottom hadrons" and so on. On the other hand the general area of research is called "b-physics" or "beauty physics", LHCb is the LHC-beauty experiment and so on. It makes perfect sense that this article uses "beauty" throughout for consistency.
At the end of the day its just a name and this particle happens to have more than one (as does the J/psi and others) and neither is more whimsical than the other. Physicists tend to be quite whimsical anyway, particle physicists perhaps especially so.
...of which the story itself is an overly neat simplification that has spread by better capturing the imagination of descriptive-minded physics students, in the teachers' own science communicating :)
There is a huge demand for CS skills of all kinds. CERN has one of the largest SCADA installations in the world, runs its own internet exchange point, operates a network of computing centers with a million cpu and hundreds of PB of data, and maintains many many millions of lines of code of specialized software, just to give you an idea.
It is a fantastic place to work, the level of expertise and dedication among the people there (scientists and engineers alike) is very very high.
I realize this isn't a big deal for quarks, but it's been on my mind a lot. The application to politics is left as an exercise for the reader.
No, that’s not what this means. But this is about one in 1,000 BBC Science articles that have repeated this confusion.
Edited to add:
What this actually means is that, conditional on the null hypothesis (that there is nothing there to find), there is a one out of 1,000 chance of seeing an observation such as this by chance.
But the BBC article has inverted this. Rather than giving you the probability of the data given the hypothesis, they're purporting to give you the probability of the hypothesis given the data (which in general cannot be objectively stated; this is a standard difficulty in the philosophy of science and statistics): "there is roughly a one in 1,000 chance that the measurement is a statistical coincidence [i.e. the null hypothesis is true and there's nothing to see here]". They make it sound like the scientists are virtually certain that they've found something here, when in reality this is probably nothing (although that's a matter of opinion).
I'm not statistician however so presumably I'm just as wrong
I’m thinking it would be more accurate (by being less precise) for BBC to describe the 1:1000 possibility as just coincidence rather than “statistical” coincidence.
Re “statistical”, to me, it sort of implies all sources of error or variability were provably not present (prove a negative) and the only reason for a coincidental result was a statistical anomaly. What’s got to be far more likely is some unaccounted-for variable produced the coincidental result.
Here's an old article from the time of the Higgs Boson discovery that had a great explanation:
https://blogs.scientificamerican.com/observations/five-sigma...
> Chances are, you heard this month about the discovery of a tiny fundamental physics particle that may be the long-sought Higgs boson. The phrase five-sigma was tossed about by scientists to describe the strength of the discovery. So, what does five-sigma mean?
> In short, five-sigma corresponds to a p-value, or probability, of 3x10-7, or about 1 in 3.5 million. This is not the probability that the Higgs boson does or doesn't exist; rather, it is the probability that if the particle does not exist, the data that CERN scientists collected in Geneva, Switzerland, would be at least as extreme as what they observed. "The reason that it's so annoying is that people want to hear declarative statements, like 'The probability that there's a Higgs is 99.9 percent,' but the real statement has an 'if' in there. There's a conditional. There's no way to remove the conditional," says Kyle Cranmer, a physicist at New York University and member of the ATLAS team, one of the two groups that announced the new particle results in Geneva on July 4.
The BBC article says:
> The measurement from LHCb is three-sigma - meaning there is roughly a one in 1,000 chance that the measurement is a statistical coincidence
My interpretation of this is "If there was no new physics, then there is a 1/1000 chance the same result could be measured by coincidence."
I understand why some people may interpret this as "There is a 1/1000 chance that there is no new physics" though.
"There is a one in 1,000 chance that the measurement would arise by statistical coincidence" uses the conditional, (implicitly) presuming a hypothesis and saying something about the chance that, under that hypothesis, such a measurement as occurred would occur. It's P(data | hypothesis).
I now see your criticism and I agree the subjunctive is much clearer. I guess my brain just converts the other wording into yours whenever I read text like that.
Actual subjunctives are only rarely used in English and often sound clunky. If you did use it it would be "There is a one in a 1,000 chance that the measurement arise by statistical coincidence". Note that it must be "arise", not "arises".
I'll correct it in my comment (you're replying to someone quoting me).
"I don't think he'd appreciate you stealing his spotlight on grammatical minutia."
"I wouldn't bet on it."
You can read more about it here: https://en.wikipedia.org/wiki/English_subjunctive
Also note that I'm specifically referring to the plain subjunctive. There is also something sometimes referred to as the "past/imperfect subjunctive" which is something different entirely.
I think, however, that I prefer your rigorous distinctions. Moods seem to be important to human language, even if they're atrophied in English. So keeping the gate on the subjunctive vs conditional distinction points people toward a deeper understanding of how language works.
The "experiment" (colliding protons together) has been repeated quadrillions of times at the LHC. But there are only a few parameters you can change about it, namely the energy and how focused the beams are. Altering the detectors to change/improve how data is collected is an expensive and slow process.
LHC looks at MANY possibilities. For a while, they tried to keep track to quantify the look-elsewhere effect. I think they have given up on that.
Now all that doesn't really matter as this is already factored into the standard deviation of their result. They have enough data to say it's a 3 sigma (being the standard deviation) deviation from the standard model. Now this "1 in 1000" really just means that the null hypothesis (the Standard Model) says there's a 0.1% chance for that result to happen. Of course it could also be that they forgot to factor in some uncertainty of their measurements and their sigma is actually much bigger. But it surely won't be the size of their dataset as this is quite obvious and easy to take into account.
To be more specific, to the average BBC reader, the phrases
The measurement from LHCb is three-sigma - meaning there is roughly a one in 1,000 chance that the measurement is a statistical coincidence.
and
The measurement from LHCb is three-sigma - meaning there is roughly a one in 1,000 chance to have seen this data even if the particle doesn't exist.
convey the exact same message, the first one just uses fewer words.
I mean, we might as well say that the BBC should just use the words "million" and "billion" interchangeably because most readers have no sense of scale.
Bottom line: thanks to this line, the reader is likely to think that new physics has very likely been discovered, when the opposite is the case.
> The measurement from LHCb is three-sigma - meaning there is roughly a one in 1,000 chance that the measurement is a statistical coincidence. So people should not get carried away by these findings, according to team leader Prof Chris Parkes, from the University of Manchester.
The difference in these two formulations is that for b) it's explicit wrong, for a) this is arguably what the article says (which is not what the scientists say)
The correct interpretation is "there is a 999/1000 chance that you wouldn't see this data if the particle didn't exist" / "there is a 1/1000 chance that you would see this data if the particle didn't exist".
The wrong interpretation that still agrees with the statement is "there is a 999/1000 chance that the particle exists given that we have seen this data" / "there is a 1/1000 chance that the particle doesn't exist given that we have seen this data".
P(data | no particle) = P (no particle | data) * P (data) / P (no particle). As a layman, I have no idea what P(data) is, and I think no one can give a convincing estimate for P(no particle).
Since P(data | no particle) and P(no particle | data) are in fact proportional, the only thing I should really care about as a layman is the value of any one of them. I any way can't evaluate myself what would be a convincing probability, since I don't have any proper priors.
The probabilities 1/1000 (or 999/1000) are only meaningul probabilites for something in the world where in fact it was a coincidence. In any other world, they are not. So saying "there is a 999/1000 chance that this is not a coincidence" isn't true, as the calculation of 999/1000 must assume that it fact it IS a coincidence.
Seriously, I completely glanced over this. Slightly embarrassing.
In any case, if the LHC runs thousands of such experiments, we do expect such signals, right?
Consider this simpler example: "Scientists always assumed their coin was fair. Well, they flipped it 10 times, and it came up heads every time. This is a 3-sigmal result. There's a 1 in 1024 chance of this measurement being a statistical coincidence."
The reader would take this "1 in 1024 chance of coincidence" to mean "if the coin was fair, they'd have a 1 in 1024 chance of seeing this event occur". That's P(data|fair) = P(10 heads|fair) = 1/1024.
I don't see how you read that and translate it into P(fair|10 heads).
In this case, grammatically, they're talking about the chance that the measurement (data) is a statistical coincidence, as opposed to not being a statistical coincidence. They're not talking about the chance that the data exists, as opposed to not existing.
Yes, and if I say "there's a 0.001% chance that these coin flip observations were coincidental", I'm taking it as a given that there were coin flips observed (indeed, there were!) I'm talking about the chance that the observations, which we (indeed) know to exist, were coincidental.
That is... literally P(observations | null hypothesis) = P(10 heads | fair). It's clearly not P(null hypothesis | observations) = P(fair | 10 heads)...
> In this case, grammatically, they're talking about the chance that the measurement (data) is a statistical coincidence, as opposed to not being a statistical coincidence. They're not talking about the chance that the data exists, as opposed to not existing.
Which is... perfectly fine. The data clearly does exist (as would the coin flips in my example), so the chance of the data existing is 100%... there's nothing interesting to talk about there. But the chance of it having been coincidental is 0.001%.
This all seems correct to me...
The difference between the data being a statistical coincidence and the data not being a statistical coincidence is precisely the difference between the null hypothesis being true and the null hypothesis not being true. So they're making a claim about the probability of a hypothesis.
But that's exactly what I'm saying too? When I flip 10 coins and see 10 heads, that is the data, and it very much exists. Nobody is talking about any data except that one. Not me, not you, not BBC.
> Grammatically, they've worded it so that the question is whether this particular data that we have seen (as opposed to other data we could have seen) was in fact produced by chance or not.
Yes, and as I see it this is 100% correct. Like in my example where I saw 10 heads (that is the one and only dataset we know exists), and I'm wondering if they were produced by chance or not. That is precisely P(10 heads | fair coin) = P(observations | H0).
> The difference between the data being a statistical coincidence and the data not being a statistical coincidence is precisely the difference between the null hypothesis being true and the null hypothesis not being true.
Maybe your English analogy is the source of the confusion here? I don't know what rigorous definition you might have for "the difference" here(?), but whatever it is, "10 heads have a 0.001% chance with a fair coin" doesn't imply "10 heads have a 99.999% chance under an unfair coin"... right? Even though the former assumes H0 and the latter assumes !H0.
Or to put it another way, just because H0 is true in one statement and false in another, that doesn't mean we're talking about the probability of H0 being true in one statement and the probability of H0 being false in another.
By definition, "the measurement is a statistical coincidence" means the same thing as "the null hypothesis is true". That's literally just how we define "statistical coincidence".
So the sentence might as well be worded: "there is roughly a one in 1,000 chance that the null hypothesis is true".
(edit: I realize that you mean this. I just wanted to make it clearer. The article should have said: It's a 1/1000 chance that a measured difference of this or larger magnitude occurs as a statistical fluctuation assuming the SM)
OK, now I see what you're saying. I'm not sure that's how people interpret it in plain English though. I feel like people would interpret "is a coincidence" to mean "occurred coincidentally", aka "occurred naturally [by chance]". It might be kind of like arguing that "ladies and gentlemen" refers to their intersection rather than union. Mathematically it does, but English is another matter...
Maybe the only way to settle this is to actually go do an experiment on people. I could be wrong, but I feel like if you go up to people and say "I think these 2 dice are fair, and I got two 6's when I tossed them; what are the chances this is a coincidence?" you'd get back "1/36" from most people. (Assuming they remember basic probability at all. You can filter against that by first asking them something more obvious, like maybe "what are the chances of getting 2 heads in a row with a fair coin" and making sure they tell you 1/4 before you proceed.)
If my dice are loaded then the two 6’s weren’t a coincidence. They were guaranteed. So to say they happened by coincidence is to say the dice are fair (the null hypothesis is true).
Consider a typical English sentence: “was it just a coincidence that I saw Frank in town, or was he following me?” They are two alternative hypotheses to explain the same observation: In the first scenario, we assume that Frank was going about his business and just happened by chance to be in the same place as I. In the second, we assume Frank was following me and that’s why he was in the same place. The question is about the hypothesis invoked to explain the observation. “Coincidence” is a (null) hypothesis.
1/36 is the wrong answer for the question "what are chances this is a coincidence?". That probability is unknown, because it depends on whether the dice are actually fair, and what other dice are there. For example, if you have to discriminate between fair dice and unfair dice with only 2s on them, a 12 would indicate that with 100%, it was a coincidence.
You might ask differently though: "What WAS the probability to throw a 12 with two fair dice?". But that probability collapsed to 100% when you actually throw them and got the result.
This does not mean that, if you get 10 heads, it's a fair coin with 1/1000 probability. In the strongest sense, without further information, one can not make any statement about that probability.
But that's wrong, the statement must be: If it is a statistical coincidence, it had a probability of 1:1024 of occurring. But then the probability that it was cause by a statistical fluctuation is 100% -- It's assumed to be true.
Additionally, integrated luminosity is luminosity integrated over time, and is just the number of events produced per unit area. It is the main metric used when CERN releases data and shares just how many events are in a given data set; this is done such that you can take the cross section of a given event (like Higgs production), multiply it by this value, and get how many events of that type happen (how many Higgs were produced).
Have worked in the dark basements for a news company, it's A-Z testing for some online pieces to see what gets more fish on the hook. They'll use wildly different headlines without changing a single word in the article.
When I posted, I believe it was “LHC machine finds tantalising hints of new physics”. I remember that seemed oddly redundant, like an “ATM machine”. Current title for the article is showing as “Machine finds tantalising hints of new physics”, but a search is also coming up “LHC machine challenges leading theory of physics”.
Unfortunately in LHC jargon, "the machine" is the collider itself rather than the detectors.
I'd encourage people to read up on the Look-elsewhere Effect when you see something like this:
https://en.wikipedia.org/wiki/Look-elsewhere_effect
I'm not saying this isn't potentially an interesting result, but 3sigma deviations when your parameter space is so large is actually not a big deal.
The super-luminous neutrinos had 6 sigma significance and still no one took it seriously because it was simply such a ridiculous claim. Obviously the number of zeros in your p-value don't mean a damn if you haven't plugged in your equipment correctly!
We are pretty confident of this, because the theory describing them is so precise -- QED is the best tested theory of all, to about 14 digits IIRC. In fact, a big factor in the discovery that a proton is a composite system was that it could not be described by QED, it has a anomalous magnetic moment.
You have got to see the implicit, unspoken qualifiers that are on all statements like that. "It would be counter to all known patterns if...", "it would have to be extraordinarily small, beyond unimaginable sensitivities, if..." and "we can't imagine any practical distinction between what it seems like and what it might be," are all reasonable interpretations of the statement "it isn't." The only unreasonable interpretation would be the literal one.
(P.S. It is not actually fundamental, it is made up of right-electrons and left-electrons, coupled together.)
Let's use the bad model that an electron is a small ball made of some magic material. And that the material inside it is even, there are not more dense parts. Also the charge is distributed evenly, it's not concentrated in some parts.
The electron is spinning, so you imagine that the ball is spinning. So you can calculate the angular momentum of the electron, assuming it's an even ball of a magical material. The angular momentum can be measured experimentally, so you can calculate how fast the electron is spinning, assuming it's an even ball of a magical material.
It has charge, and the charge is moving, so it is like a small magnet. You have calculated how fast it is spinning, and you can calculate the magnetic moment of the electron, assuming it's an even ball of a magical material.
Now you go to the lab and measure the real magnetic moment of the electron and it is twice the value of the number you got assuming it's an even ball of a magical material. This number is called g, so for an electron g=2 (actually almost 2).
You can repeat the same calculation for protons, and the g of the proton is g~=1.410606...
For an elementary particle, there are theoretical reasons to be sure that g=2. So the conclusion is that the electron is an elementary particle and the proton is a composite particle.
---
Bonus: Actually, you can get a perfectly isolated electron alone, because there are nasty virtual particles floating around. You can't see the virtual particles, but they cause small corrections in the experiments with high energy particles. In particular, the g of the electron is not exactly 2 because these virtual particles cause a small correction. The value is approximately g=2.00231930436182(52) https://en.wikipedia.org/wiki/Electron_magnetic_moment
If it smell to much hand waving, don't worry. There are very good models for all the nasty virtual particles, and you can make a long calculation of the effect of them, and calculate these correction. You can make the theoretical calculation and the experimental measure and they agree with 10 significant figures. They prefer to publish a=(g-2)/2 and
a_thoretical = 0.001159652181643(764)
a_experimental = 0.00115965218073(28)