The New Scientist has picked up my story of the Bump Hunt in their issue this week. The author, Anil Anathaswamy, interviewed me for an hour a couple weeks ago on the phone, wrote a draft, let me look it over for fact-checking. By and large he got the story correct, I’d say. It’s an odd feeling, though, I have to say, to se it out there like that.
The article was also provocative about the possible bump in the Z to bb analysis from CDF, which, now that it has been approved, is public. Tommaso Dorigo’s blog has a more in-depth look at that. But I think that is even more of a stretch than our two-sigma excess. The reason is simple: suppose our excess is in fact the beginning of a Higgs signal. You can ask the following obvious question: if we have a certain rate in the tau lepton pair channel in our analysis, what would the rate be like in the bb mode that Tommaso studied? Can they even see such a thing?
We can answer the question ourselves without doing a lot of work, by making a simple assumption. In both analyses, we are looking for a signal which is a small bump centered at 160 GeV or so, on top of the falling spectrum for events with masses larger than that of the Z boson, which appears in both analyses as a much larger bump centered at the Z mass, 91.2 GeV. In the tau pair case the excess is about 15 events, compared with about 250 Z events. So what does that imply for the bb case? We can assume that if there is a signal, the ratio of Z events to Higgs events is the same, but then also you need to take into account the different branching ratio to the bb final state. For the Z, the branching ratio to bb is 5.6 times bigger than the tau pair ratio, and for the Higgs it’s about 10 times bigger. So in the bb analysis, the ratio of Higgs to Z events is 1.8 times bigger than that in the tau pair case. You still with me?
Now, the bb analysis (which is quite a nice piece of work by the way) has about 5600 Z events, or 200 times more than the tau pair case. (It’s much easier to see these sorts of events than tau pairs, in case you are wondering…) So, if the tau pair excess is real, they should expect 1.8 x 5600 x 15/256 events, or on the order of 400-600 events, spread out around 160 GeV in mass. It would be about 10% as big as the Z peak.
It’s pretty clear that the data don’t really show such a peak…
The bottom line is that the New Scientist article, though provocative, overplays this non-signal, and so does Dorigo himself, I think. Sorry Anil! I should have looked into this earlier.
Otherwise, they’d have supersymmetric Higgs limits much like the tau pair ones, right? We’ve already excluded anything they can see in the bb analysis…but they should do the limit!
March 3rd, 2007 at 6:03 pm
Hi John,
it’s nice to meet virtually here to discuss your signal and my non-signal
You did a clear back-of-the-envelope calculation, and I have little to object to it. Of course, one could try to make it a bit more precise (say, the efficiency at 160 GeV is larger than at 90, blah). The bottomline, though, I think is the one you already drew.
Indeed, one should see a signal a tenth of the Z->bb or so in our plot. The signal would have a width of 20 GeV or so, though, that is almost twice as large as the Z peak, because our resolution scales with energy, as you well know (for by-standers here, John and I have teamed together in the Higgs Sensitivity Working Group four years ago: and I have to say he made me really blush once, when he spent words of praise on one bit of result I had gotten within that group - it was indeed about increasing the jet energy resolution!).
Anyway. The fact that the H->bb at 160 would be almost twice as fat as the one at 90, means that a tenth of the rate of Z->bb decays in my plot, spread in a bump twice as wide, would be 20 times less tall.
And the Z signal in my plot is “400 events” tall.
So you get that you need to look for a 20-25ish event tall gaussian excess, spread over the region 130 to 190 GeV.
Now, can we rule that out in my plot ? No. Because there are several points above the fit in the 160 GeV region.
More importantly, the data-fit distribution shown in the inset of the plot shown in the post does not account for background uncertainties… All in all, John, I think you are being too nice by saying the MSSM H search should be done in our sample! But we will give it a look!
Cheers,
T.
March 3rd, 2007 at 6:28 pm
[…] Now John does a nice back-of-the-envelope calculation , showing that if the H->tau tau signal is real, there should be a more evident bump in the CDF Z->bb signal plot. And then I counter in the comments section of his post, arguing that indeed, my plot does not rule out a signal the size he sees in his tau-tau analysis. […]
March 9th, 2007 at 11:49 am
[…] CDF has seen a beautiful (no pun intended) signal for Z -> bb, and in that mass spectrum they could in principle see the Higgs but in my opinion it seems like they aren’t quite as sensitive as the tau pair analysis, and would not have been expected to see anything yet. […]
March 11th, 2007 at 6:21 pm
[…] Tommaso (Dorigo) fa qualche alluzione un po’ azzardata: se nei dati di Conway ci fosse veramente il bosone di Higgs, allora dovrebbe fare capolino anche da altre parti, per esempio nello spettro di massa delle coppie di quark b, e lui quasi quasi suggerisce che in effetti sembrerebbe esserci. La discussione va avanti, John (Conway) fa due conti semplici semplici che smentiscono Tommaso e mostrano come nello spettro delle coppie di b non ci sia traccia di alcuna particella a 160 GeV, e via così. Tutto lecito e pure bello, peccato che, siccome queste chiacchiere viaggiano in chiaro sulla rete, qualche giornalista un po’ avido di scoop le intercetta, fa un paio di telefonate, equivoca più o meno di proposito due o tre cose, ed ecco che escono prima un articolo su New Scientist, e poi persino sul numero dell’Economist che ho tolto dal cellophane ieri. Risultato: smentite a destra e manca, disclaimer vari, e qualcuno di CDF si arrabbia pure. Accidenti. […]
May 13th, 2007 at 10:38 am
What’s this about a 5-sigma resonance at 180GeV from this past week … ?