The Tevatron, Fermilab’s mighty but ancient (as these things go) particle accelerator, is scheduled to be shut down at the end of this year. But the old beast might have a trick or two left yet.
Way back in April we talked about a couple of lingering anomalies in the Tevatron data that had risen to the level where theorists were intrigued enough to start building models. One of these — a forward/backward asymmetry in top-quark interactions — had been around for a while, and was taken seriously by a number of people. The other — a tiny bump near 150 GeV in the total number of events that produce a W boson and two jets — was relatively new, and was greeted by a bit of scoffing. The bump credibility took another hit when it was pointed out that it could be explained away by a simple (although completely hypothetical) systematic error — a miscalibration of the jet energies. Bump-hunting is hard, and experiments near the end of their lifetimes are more willing to share their anomalies than they would be if they knew they were going to keep going, since there’s little hope that new data will solve the problem.
But there’s some hope. The real reason to be patient rather than excited by the bump at 150 GeV was that it was a 3-sigma effect, in a game where most 3-sigma effects go away. In particle physics, we generally take a solid 3-sigma result as “evidence for” something, and require 5 sigma — a much greater deviation from the expected numbers — to declare something a “discovery.”
More data are now in! This is from the CDF experiment at Fermilab, as reported in a conference talk by Giovanni Punzi (pdf), and shared worldwide by Jester at Résonaances. There’s a reason why I mentioned Résonaances among the physics blogs above — it’s unquestionably the go-to place for new results in particle physics.
And the anomaly is now — almost five sigma! It didn’t go away with more data, it became more prominent. It would be very hard at this point to simply attribute it to an energy miscalibration or something like that; if it is a systematic error, it’s a subtle one. But it doesn’t look like an error; it looks like a signal.
Of course, it’s still very possible that it will go away. These things usually do. But when an interesting result is pushing five sigma, it’s perfectly okay to get a bit excited and start wondering what’s going on. One of the nice things about this bump is that it’s not very hard to come up with models that can explain it — all you need is a neutral boson, similar to the well-known Z boson of the weak interactions, with a mass near 150 GeV. This kind of idea is so well-known in the trade that it already has a name — the Z’ boson, imaginatively enough.
Except it’s not that simple, of course — where would be the fun? When you start mindlessly adding new particles to the Standard Model, you have to check consistency with all sorts of experimental constraints. In particular, a naive Z’ boson would sometimes decay into leptons as well as quarks (the jets mentioned above). In that case, it would have been seen long ago in LEP, the electron-positron collider at CERN that previously lived in what is now the LHC’s tunnel. So what you really need is a “leptophobic” Z’, one that decays into quarks but not into leptons.
Or something along those lines, or something completely different. See Résonaances once again for the lay of the theoretical land. Yes, there are possible explanations within supersymmetry; and yes, there are explanations that have nothing to do with supersymmetry.
If this is real — still a very, very, big if — it’s the beginning of the “beyond the Standard Model era” in collider particle physics. Things aren’t going to snap into place overnight; there will be false starts, mysteries, and sudden epiphanies. That’s where the real fun is in science.
Update: Note that the very preliminary word from the LHC is that they don’t yet see the same bump that CDF does. But from a glance at the figure it doesn’t look like they have nearly as much data yet, so that’s probably not surprising. The LHC has seen incredible jumps in luminosity recently, however, so they should be able to do a proper check before too long.
May 30th, 2011 at 4:01 pm
The big question is whether D-zero sees the bump. I’m sure they’ve been analyzing it for a while now, but still no announcement. I do recall similar bumps proving that pentaquarks existed, they were 5 sigma and went away…
May 30th, 2011 at 4:06 pm
Shouldn’t we be asking Harold Camping to interpret this for us?
May 30th, 2011 at 4:52 pm
Very exciting.
I am as lay as lay gets, but one question I am wondering, is why we are seeing this bump now, since Femi has been operating for so long? Is this a new experiment not previously done?
Or is it even possible the “old” date might be mined to look for this signal?
May 30th, 2011 at 5:02 pm
Tom S—the Tevatron has collected much more data in the past year than previously. One needed several fb^-1 to detect a signal, and only in the last few months has this amount of data been collected. So there is no old data to be mined.
May 30th, 2011 at 6:30 pm
I don’t understand how much the upgrade to 5 sigma matters. It would if the “worry” with the previous result was that it was a statistical fluctuation, since you wouldn’t expect the new data to fluctuate the same way. If instead the suspected problem is something to do with jet scale, and its value and uncertainty not being well understood, then simply accumulating more data does nothing to solve that problem. The new analysis could reflect the same bias, statistical uncertainties go down, (and the scale uncertainties go down if they’re measured on a larger sample) so there is more bump and larger deviation from the SM prediction, and you can get larger sigma. For CDF to increase confidence in this result, they need to address the old questions about systematics, not just use more data, right?
May 30th, 2011 at 8:30 pm
Good news indeed! And a well written article.
Cheers
May 30th, 2011 at 8:43 pm
@prasad, you’re right, this is only evidence against one of the three possibilities — (1) statistical fluke, (2) systematic / modeling error, (3) new physics.
May 30th, 2011 at 9:28 pm
For TGD based explanation of the 150 GeV bump in terms of scaled up hadron physics predicted by TGD see this. Note that the leptophobic character of bump is the tell tale signature in this respect.
May 31st, 2011 at 12:40 am
What prasad and Peter Davis said. Is there some subtle reason, Sean, why 5-sigma makes us more confident than 3-sigma that this is a legitimate bump rather than a systematic error? The Résonaances post doesn’t suggest that there is.
I guess I can imagine that the shape of the bump (some higher-order moment beyond the height and position)—which becomes more visible with more data—could discriminate against plausible systematic sources.
May 31st, 2011 at 12:44 am
I’m a little curious about the 5-sigma limit. If it is still very possibly (likey, even?) for a 4.8-sigma bump to go away, how often do “discovery” signals greater or equal to 5-sigma go away?
Very interesting result! Will be looking forward to seeing how this turns out!
May 31st, 2011 at 12:56 am
[...] Artículo publicado por Sean Carroll el 30 de mayo de 2011 en Cosmic Variance [...]
May 31st, 2011 at 2:43 am
[...] Giovanni Punzi delivered a talk that has already caused a stir around the blogosphere, see here, here, here. First take a look at some of his collected plots for Higgs [...]
May 31st, 2011 at 4:27 am
Is this the HIGGS-BOSON ?
May 31st, 2011 at 4:59 am
Simplicity, nope, as other blogs explain, this couldn’t be just the Higgs.
May 31st, 2011 at 8:02 am
Ok, thank you very much Damjan
May 31st, 2011 at 8:49 am
Anyone from D0 reading this care to comment???
May 31st, 2011 at 11:17 am
Chris, basically there’s a fiendishly complex calculation that ends up saying “the number we should measure is X±Y”, where Y is the standard deviation. It should be within X±Y 67% of the time, within X±2Y 95% of the time, within X±3Y 99% of the time, etc.
They’re saying that the number they’re actually measuring is X+4.8Y. which is a long enough way from X to warrant attention.
But this all depends on X and Y being computed accurately, and that is, as I mentioned, fiendishly complicated. A slight error in either number and the 4.8 vanishes.
For example, look at the recent proposed solution to the “Pioneer anomaly”. The spacecraft is accelerating back toward the sun ever so slightly faster than gravity predicted. All kinds of explanations have been proposed including modified theories of gravity. In March this year someone proposed that the models of radiation pressure from the heat photons given by the radioisotope thermal generator failed to account for reflections off the back of the main radio dish. Which always points toward earth, so the anomalous solar sail effect is always toward earth. No change to the law of gravity required, just an accurate model of non-gravitational forces.
The number “5″ is arbitrary like the size of a gift an elected official is allowed to accept. Normally, when you have 5 sigma, you make an announcement, then everyone else goes looking, and with more data the Y value shrinks and the significance climbs until it’s way above 5 sigma and everyone accepts it.
May 31st, 2011 at 12:28 pm
Thanks for the description Aaron, though I do understand the concept alright.
My question (which I probably should have worded better) is, what percentage of these “discovery” signals (~5-sigma) turn out to be nothing? Is it common for 5-sigma signals to go away? I assume that the arbitrary level was put at 5-sigma because it is a relatively reliable indicator that they’re actually seeing something interesting – more often than not the signal turns out to be real, for example. But it also sounds like (from this blog post, and others) this 4.8-sigma signal is likely to fizzle into nothing, leading me to assume that quite a large number of ~5-sigma discovery announcements also turn out to be nothing.
I just wanted to get a feel for what this number might be. Is the assumption that this 4.8-sigma signal will turn out to be nothing simply restrained skepticism, or is it just common that ~discovery level events end up that way?
May 31st, 2011 at 12:35 pm
I second Tom #16. Come on, Dzero, just do it! You know you want to.
May 31st, 2011 at 2:17 pm
Question for Experimenters: Can D0 see the Standard Model WW/Z dijet peak in l\nu jj ??
May 31st, 2011 at 2:47 pm
[...] dauern, bis er das Tevatron-Resultat klar bestätigen oder widerlegen kann. (Quantum Diaries 31., Cosmic Variance 30., Francis Science News 28., Nature News 27., AFP 24., CERN Release, Symmetry Breaking 23., New [...]
May 31st, 2011 at 2:50 pm
It’s condensate glueballs.
May 31st, 2011 at 2:51 pm
[...] lo cuentan aquí, aquí (traducido aquí al castellano por Kanijo) o aquí, pero todavía no hay que lanzar las campanas al [...]
May 31st, 2011 at 3:09 pm
[...] mismos métodos de Montecarlo para estimar el fondo). También puedes leer a Sean Carroll, “Anomaly at the Tevatron Might Be Something Real?,” Cosmic Variance, May 30th, 2011; Philip Gibbs, “Rencontres de Blois,” [...]
May 31st, 2011 at 3:40 pm
Google “pentaquark” for information on a particle which was observed with similar significance, 4.6 sigma, but vanished when the statistics improved – in the same detector at Jefferson Lab.
May 31st, 2011 at 4:07 pm
I’m struck by how wide the mass range of the bump seems to be. The elevation is about the same in the data points from 120 GeV to 160 GeV. It almost looks like a composite of a double or triplet of similar particles decay sequences with similar but not identifical masses (a bit like the W+, W- and Z at ca. 80 Gev and 90 Gev).
May 31st, 2011 at 6:21 pm
[...] http://blogs.discovermagazine.com/cosmicvariance/2011/05/30/anomaly-at-the-tevatron-might-be-somethi... [...]
May 31st, 2011 at 11:40 pm
[...] a post on Discover‘s Cosmic Variance blog, Caltech theoretical physicist Sean M. Carroll nicely contextualized the tantalizing but still unconfirmed finding: If this is real—still a very, very, big if—it’s the beginning of the “beyond the [...]
June 1st, 2011 at 2:38 am
[...] more relevant posts see here, here and [...]
June 1st, 2011 at 5:15 am
Most 3 sigma results “go away” and “turn out to be nothing” and now, Sean, you are essentially saying the same thing about ~5 sigma results. Not long ago I thought 5 sigma was enough to claim a discovery?
All of this talk of sigma this or sigma that seems almost meaningless at this point since you guys are saying there could always be some confounding factor that could creep into an experiment at any level of sigma. If some subtle systematic effect can always creep in when can you ever really claim a discovery in science: SOME SKEPTIC COULD SAY THAT WHAT YOU ARE SEEING IS DUE TO SOME OTHER EFFECT!
June 1st, 2011 at 7:38 am
The importance of a 5 sigma result is just that at that point other experimenters take you very seriously and look hard at debunking or confirming your claims. Until your “discovery” has been independently verified in different places by different people with different techniques then, yes, skeptics can say what they will.
So a 5 sigma signal is just a chance to put other experiments on notice that there is something very interesting worth looking at. In contrast, a 3 sigma result is something that you as an experimentalist need to work harder to resolve. Theorists might get excited, but that’s a low bar. Discovery takes a while; 3 sigma and 5 sigma are important milestones on that road, but by no means the end of the process.
June 1st, 2011 at 8:51 am
@forester: You need 5 sigma to rule out “it’s due to chance”. You need independent confirmation to rule out “it’s due to systematic errors.”
In other words 5 sigma is necessary for discovery, but not sufficient.
June 2nd, 2011 at 3:26 pm
147(5)GeV – Excellent! This shifts the center of the data to my 10+ year old prediction (148 GeV) in eq. 18 of http://theoryofeverything.org/TOE/JGM/ToE.pdf.
June 2nd, 2011 at 3:58 pm
Eccezionale
June 2nd, 2011 at 4:06 pm
Well, this is still same old wrong Jet Energy Corrections. If you will add more data with same crappy corrections, you will get larger bump – garbage in, garbage out.
I have a conspiracy theory here – FNAL looking for ways to get back the funding which is lost when Tevatron was scheduled to stop at the end of this year. I remember same thing at LEP when it was less than a year before closing. Everybody looked for Higgs and found it with 3 sigma at 115-118 GeV – remember Wu?
So, relax and don’t hold your breath. This is a battle for the money. It has nothing to do with physics or science.
June 2nd, 2011 at 5:59 pm
invariant mass released today : http://www-cdf.fnal.gov/physics/ewk/2011/wjj/7_3.html
for relaxed #35 : see Fig 5
June 4th, 2011 at 8:35 am
[...] for “W-jj excess”). One of the attendees wrote a blog post on it; that, in turn, got picked up by Cosmic Variance, on the popular Discover Blogs site, which is where I found it. Following the [...]
June 5th, 2011 at 4:26 pm
[...] that; if it is a systematic error, it’s a pointed one,” Caltech’s Sean Carroll writes over during Cosmic Variance. “But it doesn’t demeanour like an error; it looks like a [...]
June 10th, 2011 at 9:40 am
[...] here we have science, marching on. Just last week we heard that CDF, one of the big experiments at the Tevatron at Fermilab, had collected more data relevant [...]
June 11th, 2011 at 11:07 pm
[...] about the hunt for new particles at Discovery News.So here we have science, marching on. Just last week we heard that CDF, one of the big experiments at theTevatron at Fermilab, had collected more data relevant [...]