The Particle Physics Results You Didn't Hear About
HTML-код
- Опубликовано: 11 сен 2024
- Experimental results in particle physics tend to only get the attention of the press if they disagree with the predictions of the Standard Model of particle physics. This presents a skewed view of particle physics results--it picks out those results that are genuine discoveries, but also those where experimentalists got really unlucky.
Here, we try to give a more complete picture of the body of experimental particle physics results. We discuss how physicists quantify disagreement of an experimental result with a theoretical prediction. We then talk about the look-elsewhere effect, also known as the multiple-comparisons problem. Lastly, we try to give a reasonable lower bound on the number of experimental results that particle physicists produce in a year. We look at the experiments at LHC (ATLAS, CMS, LHCb, and ALICE) and then briefly mention the multitudes of other particle physics experiments actively reporting results.
Deviations from the SM prediction mentioned in the video:
Muon g-2:
• Miniseries on g-2 of t...
CDF W boson mass measurement:
• That New Fermilab Resu...
• Just How Important is ...
Flavor anomalies:
• All Those Flavor Anoma...
Other relevant videos:
Brief Introduction to Gaussian Errors:
• Brief Introduction to ...
Look-Elsewhere Effect:
• Why the Look-Elsewhere...
Recent Atlas results mentioned in video:
Search for new phenomena in multi-body invariant masses in events with at least one isolated lepton and two jets using s√=13 TeV proton-proton collision data collected by the ATLAS detector:
arxiv.org/abs/...
Search for boosted diphoton resonances in the 10 to 70 GeV mass range using 138 fb−1 of 13 TeV pp collisions with the ATLAS detector:
arxiv.org/abs/...
Any questions?
Out of topic: is this channel specifically for particle physics, or we may see some QFT, GR, etc.?
@@rohanshah7960 Check Eigenchris here on youtube, and also his recommendations
@@Kraflyn alright, will do. Thanks!
@@rohanshah7960 Hi!
Well, the hope is that it's not about particle physics per se but the methods that physicists use to extract conclusions from experimental data. But, yeah, I'm a particle physicist, so there's a strong bias in that direction!
Nice and clear. Appreciated, thanks!
[the rest of my comment is super ramble-y and doesn’t really express a clear question. As such, I suggest not reading it without first explicitly considering whether you actually want to]
I recognize that the P(measurement deviating from standard model by at least x | standard model is true) < p does not by itself give a probability for “the measurement differs from SM because of a way in which SM is incorrect/incomplete”, but, if we had a prior probability distribution for P(standard model makes correct predictions with regards to the thing being measured in this experiment) and P(the standard model is incorrect/incomplete in ways this experiment could reveal), uh, these, or something like them, could be combined to produce a probability for whether the difference from prediction is due to a way SM is incomplete, right?
But maybe there isn’t really a principled way to collectively assign such prior probabilities?
I guess uh, if a non-physicist who is hearing about stuff on the news is trying to assign a probability for “this result I heard about is due to a newly discovered real deviation from the SM” they should be like, P(it is a real deviation, and I heard about it)/(P(I heard about it and it is a real deviation)+P(I heard about it and it is a fluctuation or mistake)),
and...
Well, if we assume that they hear about anything above n sigma or something,
then,
hm,
I’ve confused myself, probably by conflating some distinct ideas. Like, if one is evaluating “what is the probability that *this specific measurement result* is due to physics beyond the SM ?” seems confusing to me to think about the prior probability for it, because before hearing about it they would not have conceptualized that possible experimental result (not having been aware of what experiments were even in progress).
But... somewhere in the right answer to the question that I am trying and failing to ask,
I imagine there is a term for “how many experiments were done which, if they turned up with a deviation from SM of at least x size, would I have heard about”?
Ok, uh, if we assume that the news-hearer hears about all 3+ sigma events, and the only thing they hear about each event is how many sigmas,
then, if assuming (I think this is a bad assumption but I don’t see a better way to simplify) that whether each experiment is measuring something where the true value differs from the SM value by a given amount, is independent (again, this seems a bad assumption)
and, also assuming identically distributed,
then, I guess you would look at...
Ok no, even in this way oversimplified model this is too complicated for me to reason about while reclining in bed on my phone.
Because like, there’s a distribution over lists of varying lengths of numbers, where this list comes from taking the entries in another list which are above some threshold.
Well, I don't know if I'll say anything useful, but, well, let's see!
I think you're asking how a person, outside of physics, hearing about a result that appears to disagree with the SM, would assign a probability to the hypothesis that the result, does, in fact, indicate a problem with the SM. Is that right?
If one assigns prior probabilities (as you've indicated), in principle, we can do this, in at least a rudimentary way. But assigning those priors is.....um....definitely not straightforward! Within physics, people would disagree strongly about what those priors should be. I, uh, definitely don't have a concrete suggestion for how to do that. That's hard.
@@ThinkLikeaPhysicist Yeah, that’s basically what I was asking. Thanks!
I tried to come up with a simplified model which assumes away the difficulty of coming up with a prior for the standard model having different possible problems with it, to try to focus on just the part of the problem that comes from the selection effect from what results a non-physicist hears about, but even that part confused me too much, and even if I wasn’t confused, I got the feeling that the calculations would still be hard.
@@drdca8263 Yeah, this is very hard problem. As physicists, we typically quote our results in frequentist language (instead of Bayesian language, appropriate for the question you're posing here), because, well, it's just too hard to answer questions like these.
Very interesting analysis and clarification that explains unusual physics measurements to be statistically normal and why the 5σ convention limit is necessary for a discovery to be declared and even then this has to be verified from more than one independent sources in case in the original experiment the experimenters did a mistake (have got unlucky ).
If you're getting your information from the "News" you are not about real science 😂thank you for this channel ❤
You're welcome!
Thans for your videos. Plese keep on producing new ones.
Are all results shared between labs (Cern, Fermilab, China, …)?
Hi! Typically, when the experimentalists are confident of their results, they are made public. (Occasionally they are hard to find, but they are typically available.)
I heard a rumor that LHCB's results on RK and RK* are just their mistake in their analysis and soon this anomaly will dissapear.
Looks like you were on to something. You might like today's video.
Can you give an update on LHCb's now result on the flavor anomaly?
I will have to get back to you. I think they did just put out a paper--I will need to read it so that I can tell you the answer! ;-)
Thanks!
@@ThinkLikeaPhysicist I did my undergrad senior thesis relating to this topic...
I can't help but feel a little let down 😭