Posted by EvgeniyZh 10/23/2024
Once I could "see" the peak, without having to conduct statistical tests against expected background, it was "real" to me.
In these cynical times, it may be that everything is relative and "post-modern subjective p-hacking", but sufficient data usually ends these discussions. The real trouble is that we have a culture that is addicted to progress theater, and can't wait for the data to get in.
For the idiots in this post (me), could you please explain what that entails and why it helps confirm the discovery?
The invariant mass is the rest mass of the particle (i.e. it's "inherent" mass). You can calculate it by taking the final state decay products of the original particle (i.e. the particles that are actually observed by the detector) and summing up their four-vectors (squared).
You can plot the invariant mass calculated from any particular final state, and for a rare particle like the Higgs the majority of the contributions to your plot will be from background processes (i.e. not Higgs decays) that decay into the same final state.
If you have a lot of Higgs decays in your sample you should be able to see a clear peak in the distribution at the invariant mass of the Higgs boson, a clear sign that the Higgs (or something with the same mass) exists.
Often by the time the discovery has reached statistical significance, you might not really be able to see such a clear sign in the mass distribution. I.e. the calculations are telling you it's there but you can't see it that clearly.
I wouldn't really say this helps confirm the discovery in a scientific sense, just that it's reassuring that the signal is so strong that you can see it by eye.
It's really something when this happens. I worked on a big neutrino experiment searching for theta_13, where our goals were to (a) determine if theta_13 was dead zero or not (being truly zero would have a Seriously Major Effect in theories) and then (b) to measure its value if not.
Our experiment was big, expensive, and finely tuned to search for very, very small values of theta_13. We turned the thing on and... right there there was a dip. Just... there. On the plot. All the data blinding schemes needed to guarantee our best resolution kind of went out the window when anyone looking at the most basic status plot could see the dip immediately!
On the one hand, it was really great to know that everything worked, we'd recorded a major milestone in the field (along with our competition, all of whom were reading out at basically the same time), and the theorists would continue to have nothing to do with their lives because theta_13 was, in fact, nonzero. On the other hand... I wasted how many years of my life dialing this damned detector in for what now? (It wasn't wasted effort, not at all... but you get the feeling.)
I'm only an amateur, but wouldn't that give different results depending on choice of units? I.e, I usually use C=1.
m^2 c^4 = E^2 - p^2 c^2
where m is mass, E is the total energy in the decay products and p is the 3-vector sum of the momentum.
Those units should work out (they certainly do if you set c = 1).
[1]: https://en.wikipedia.org/wiki/Minkowski_space#Minkowski_metr...
Luckily, there's pretty simple statistics that one can throw at that once the third detector comes online. Hopefully that comes in before we spend too much money on LISA.
It's basically this, from the article, but from astro:
> Particle physics does have situations where the hypothesis are not so data driven and they rely much more heavily on the theoretical edifice of quantum field theory and our simulation of the complicated detectors. In these cases, the statistical models are implicitly defined by simulators is actually a very hot topic that blends classical statistics with modern deep learning. We often say that the simulators don't have a tractable likelihood function. This applies to frequentist hypothesis testing, confidence intervals, and Bayesian inference. Confronting these challenging situations is what motivated simulation-based inference, which is applicable to a host of scientific disciplines.
1 for 29 by 2019:
I don't think that's right. I think having an application is what ends the discussions.
If you have a group of people who think CD players work by using lasers, and a rival group who think they do something entirely different, and only the first group can actually make working CD players, people will accept that lasers do what group #1 says they do.
I certainly don't believe in black holes in the same manner that I believe in the breakfast I'm eating right now.
The information paradox is closer to us than we think!
Joking aside, another perspective on practical use is all of the technology and research advanced that have spun out of black hole research. Multi-messenger astronomy for example. We can point a telescope at the sky where two black holes merged.
Most people. Some fringe groups will believe it is all a front, and they are only pretending that so-called “lasers” are what make the CD player work when in fact it is alien tech from Area 51 or eldritch magics neither of which the public would be happy about. What else would CDDA stand for, if not Compliant Demon Derived Audio? And “Red Book”. Red. Book. Red is the colour of the fires of hell and book must be referring to the Necronomicon! Wake up sheeple!
If you have some theory as to why that test is inadequate, it's on you to lay your cards on the table and state it so there's something substantive to discuss. Until then you're just trolling.
Perhaps because cowardice prevented you from stating your point clearly, so instead you resorted to vague implications in the form of a question.
A fine tactic which gives the coward a plausible deniability defense when their bullshit is called out: I didn't say that, you said that.
Coupled with other similar techniques, e.g.:
* You know what I mean
* Questions that have actual answers, but are asked rhetorically to make an implication: How many cats are eaten alive each year by immigrants?
* Both-sidesing/false balance (giving equal weight to contradicting points of view, regardless of what the reality has to say)
* Ambiguous implication that can be taken in many ways, depending on how the asker defines the terms: Do you really think the immigrants are helping the economy?
* Non-sequiturs and non-answers
* False dichotomies
* Etc
...this gives the coward a superpower to start a heated discussion, where the coward never actually says anything directly.
Instead, the opponents exert an enormous effort to debunk each leaf on the tree of possible interpretations of the coward's incomplete thought, effectively doing the thinking for the coward in a futile attempt to nail down what point the coward intended to make.
(There is no such clear point, other than I am right and you are wrong).
By the time it becomes clear that no possible interpretation is supported by reality, the coward silently leaves the discussion, and says the same exact things elsewhere, feeling empowered by having made others frustrated.
That frustration counts as a victory in an argument in coward's view, with the assumption that it comes from inability to counter the coward's (unstated, vague, implied) points.
In reality, it's infuriating to have to be made guessing what the other person wants to say, doubly so when they are the one forcing to play that game, which, miraculously, you always seem to lose, because the outcome is invariably "that's not quite what I'm saying" (and what the coward is saying is never directly stated by the coward).
At the same time, this protects the coward from being riduculed for any of the views they're promulgating in the discussion.
This is why the coward never states the views directly, and instead e.g. "presents questions side by side".
If the coward were to say outright "Unlike CD players, there's actually no clear way to say whether vaccines work or not", the coward would be laughed out of the room, and they know it — which is why the resort to implications and rhetorical questions about coward's motivation.
I hope this fully answers the question asked.
Sometimes I wonder whether it's worth to write anything of this sort, or am I screaming into the void.
Your comment makes it absolutely worth it.
Way too many confounding factors here, and nobody will risk and experiment with a control group in this situation.
We can't tell whether a CD player has a listening device in it, for instance.
COVID on the other hand doesn't have such a mechanism, and just relies on being really contagious. So if everyone would stay up to date in their boosters and continue masking in public places, we may be able to get rid of it in a couple of years.
By that logic we'd have gotten rid of the flu. Vaccines for rapidly mutating viruses like flu and COVID can't keep up and remain an epidemic. The only disease we've actually been able to eliminate worldwide due to vaccines is smallpox. We'd have gotten rid of measles too if crazies hadn't decided the MMR vaccine causes autism due to criminally fraudulent research.
https://www.npr.org/2024/10/18/nx-s1-5155997/influenza-strai...
Don't underestimate the impact of stock viral interference - flu & COVID are both respiratory infections and COVID was much more infectious. Some flu strains probably just couldn't remain competitive with the combined set of other flu and COVID strains.
While masking and social distancing have a beneficial impact on limiting the spread of respiratory diseases, there are practical reasons why it doesn't work to eliminate it altogether and ignores the possibility and likelihood of other resevoirs to reintroduce the disease. For example, if North America remains masked & socially isolated by the virus persists in Europe, then as soon as North America opens up you'll get the virus in North America again. And imaging a simultaneous world wide lock down is a laugh - even during COVID governments were not globally coordinated and even within national governments there was mixed local coordination.
Aside from all that, let's say it was purely a result of masking and social distancing. The consequences of that were quite sever & catastrophic, not to mention that no one actually stayed away vs limited their normal contacts & there were plenty of practical reasons it wasn't possible (e.g. getting groceries). Life involves death & risk and it's pretty clear that even before the vaccines became available many people were not OK with the tradeoff COVID entailed (e.g. Florida).
The statistical interpretation showing a 5 sigma signal was certainly essential, but I suspect it would have taken the collaborations much longer to publish if there wasn't a massive bump staring them in the face.
I think the author is using the original motivation of musing on null hypotheses to derive the title "The Higgs Discovery Did Not Take Place", and he has successfully triggered the controversy the subtitle ironically denies and the inevitable surface reading condemnations that we see in some of the comments here.
[1] https://www.argmin.net/p/the-higgs-discovery-did-not-take
Also he makes many factual claims that are just incorrect.
Just seems like an extremely arrogant guy who hasn't done his homework
There was an entire class of engineers at google- SREs- many of whom were previously physicists (or experts in some other quantitative field). A fraction of them (myself included) were "cluster whisperers"- able to take a collection of vague observations and build a testable hypothesis of why things were Fucked At Scale In Prod. Then come up with a way to fix it that didn't mess up the rest of the complete system.
Nothing- not even computers are truly built on formal logic. They are fundamentally physics-driven machines with statistical failure rates, etc. There's nothing quite like coming across a very expensive computer which occasionally calculates the equivalent of 1*1 = inf, simply because some physical gates have slightly more electrical charge on them due to RF from a power supply that's 2 feet away.
Even massive distributed systems, while complex, still follow explicit rules for how they change state. Every bit of information exists in a measurable form somewhere. Sure, at Google scale we might not have tools to capture everything at once, and no single person could follow every step from electrical signal to final output. But it's theoretically possible - which is fundamentally different from natural systems.
You could argue the universe itself is deterministic (and philosophically, I agree), but in practice, the emergent systems we deal with - like biology or economics - follow rules we can't fully describe, using information we can't fully measure, where complete state capture isn't just impractical, it's impossible.
If you see a particle accelerator occasionally make an observation that breaks the standard model, depending on what it is breaking you can be very confident that the observation is wrong, but you cannot know that with absolute certainty.
I'm not in HEP, but my graduate work had overlap with condensed matter physics. I worked with physics professors/students in a top 10 physics school (which had Nobel laureates, although I didn't work with them).
Things may have changed since then, but the majority of them had no idea what pre-registration meant, and none had taken a course on statistics. In most US universities, statistics is not required for a physics degree (although it is for an engineering one). When I probed them, the response was "Why should we take a whole course on it? We study what we need in quantum mechanics courses."
No, my friend. You studied probability. Not statistics.
Whatever you can say about reproducibility in the social sciences, a typical professor in those fields knew and understood an order of magnitude more statistics than physicists.
For pre-registration, this might be debatable, but what I meant was that we have teams of people looking for specific signals (SUSY, etc). Each of those teams would have generated monte carlo simulations of their signals and compared those with backgrounds. Generally speaking, analysis teams were looking for something specific in the data.
However, there are sometimes more general "bump hunts", which you could argue didn't have preregistration. But on the other hand, they are generally looking for bumps with a specific signature (say, two leptons).
So yes, people in HEP generally are knowledgeable about stats... and yes, this field is extremely strict compared to psychology for example.
But... modern particle physics is one of the simplest things around. (Ex-physicist here, see username.) It only looks complicated because it is so simple that we can actually write down every single detail of the entire thing and analyze it! How many other systems can you say that about?
Whereas to do physics well you need only mathematics. Well, at least, to do the theories well. To actually execute the experiments is, ah, more challenging.
So I would argue the Standard Model is pretty much the only thing in all of human knowledge that depends on no other physical theories. It's the bottom. Shame it's pretty useless (intractable) as soon as you have three or more particles to calculate with, though....
It's probably a reference to "The Gulf War Did Not Take Place" by Jean Baudrillard, which took a similar critical view of the Gulf War as TFA takes of the Higgs discovery.
"This isn't music, back in my day we had Credence"
A Berkeley academic invoking "it's actually your fault for believing the words that I wrote" and following it up with a "I'm not mad, I actually find this amusing" ... it's just disappointing.
> In any event, I use irreverence (i.e., shitposting) to engage with tricky philosophical questions. I know that people unfamiliar with my schtick might read me as just being an asshole. That’s fair.
People are piling the hate on Ben Recht here. I appreciate that he's calling his post what it is rather than doubling down.
It's also a great chance to lecture people on 4-momentum, thanks everyone!
Even though I'm a theoretical physicist I've gone into the lab and spent the time to learn how to conduct experiments and what I've learned is that a lot of theoretical wrangling is not relevant to actually getting a useful result that you can be confident in.
Looking at Recht's publication history, it looks like few of his papers ever do real-world experiments; mostly, they use simulations to "verify" the results. It may very well be that his gaps in experimental physics lead him to his conclusion.
We(particle physicsts) have been performing similar, and in a lot of ways much more complex analyses using ML tools for decades in production.
Please stop shrouding your new 'golden goose' of AI/ML modelling in mystery it's 'just' massively multi-dimensional regression analyses with all of the problems, advantages and improvements that brings...
Why is there some beef that nature is complex, if you had the same vitriol toward certain other fields we'd be worrying about big-pharma's reproducibility crisis just at the top of the ice-berg of problems in modern science, not that most people are illiterate when it comes to algebra...
Hadn't realised Higgs' boson denialism was really a thing.
In its pages you could find EE professors and chartered engineers arguing that Einstein was so, so wrong, decades after relativity was accepted.
I'd trust an EE to build me a radio, but I wouldn't let an EE anywhere near fundamental physics.
[1] https://www.nytimes.com/2010/09/12/magazine/12FOB-IdeaLab-t....
If it's the same thing I'm thinking of, it was kinda flawed, IMO, in that it was a comparison of such beliefs amongst various types of scientists, with, for some reason, engineers thrown in, too. And yeah, it's kind of unsurprising that engineers are more into unscientific nonsense than various types of scientists, because engineers aren't scientists. It would be more surprising if they were significantly worse than the _general population_, but I don't think that it showed that.
Lots of failed theorists with that personality type/flaw as well.
Keeping EEs and any E for that matter away from fundamental physics is a shortcut to producing a whole lot of smoke and melted plastic.
I mean, I'm as prone to the "I'm a smart guy, so I understand everything" delusion as the next person, but I usually only show it in the comments here. (And in private conversations, of course...)
Worse, the author of the original FUD is a professor of EE at Berkeley [1] with a focus in ML. It almost goes without saying, but EE and ML would not exist without the benefit a lot of fundamental physics research over the years on things that, according to him, "no one understands".
You know, when I first read this thread and the 3 posts involved, I find the original post Ben wrote arrogant and hard to swallow. But once I searched who he is, and recognized him, knowing his character I immediately “get his point”. While not an expert in the fields, I have both graduate level educations in HEP and ML. My point is that my conclusion is unlikely due to the lack of understanding of these fields, but more because of my understanding of who he is…
Admittedly, he should not assume people read it as he intended how people would perceive. It took a lot of contextualizations, including the expectation from the title he explained in the later posts, to really take his posts seriously.
Quoth the AI researcher.
So what’s to be done? There’s only one option:
throw most of that data away in the smartest way
possible, and ensure that the data retained is
processed and stored efficiently.
[1] https://profmattstrassler.com/2024/10/21/innovations-in-data...
Most of the experiments cannot because of the data acquisition problems.
To give some numbers:
- The LHC has 40M "events" (bunch of collisions) a second.
- The experiments can afford to save around 2000 of them.
This is a factor of 20k between what they collide and what they can afford to analyze. There is just no conceivable way to expand the LHC computing and storage by a factor of 20k.
Valid question would be why they don't just collide fewer protons. The problem is that when you study processes on a length scale smaller than a proton, you really can't control when they happen. You just have to smash a lot and catch the interesting collisions.
So yeah, it's a lot of "throwing away data" in the smartest way possible.
-------------------
All that said, it might be a stretch to say the data is "thrown away", since that implies that it was ever acquired. The data that doesn't get saved generally doesn't make it off a memory buffer on a sensor deep within the detector. It's never piped through an actual CPU or assembled into any meaningful unit with the millions of other readouts.
If keeping the data was one more trivial step, the experiments would keep it. As it is they need to be smart about where the attention goes. And they are! The data is "thrown away" in the sense that an astronomy experiment throws away data by turning off during the day.
They never came close to what they said they needed.
But they now claim they succeeded in finding the Highs Boson.
And the paper setting out the criteria has been memory holed.
I call BS in the Highs Bozo team.