Live blogging of Tevatron Higgs Seminar

While I am not at Fermilab (or even in the US) right now, I will try to live-blog the Fermilab Tevatron Higgs seminars today at 9am US Central Time (4pm CET).

The link to the live stream is here:

Live blogging entries will appear below.

10:23: No more questions, and a wrap!

10:19: QUESTION: again on the model prediction and systematic uncertainty, do you have to improve expectations at each mass significantly in order to resolve the “tension with the Standard Model” issue? ANSWER: theory community has been pushing hard to make improvements on these issues. But, backgrounds to bb are still a dominant contributor to the uncertainty. Reducing them is still very critical to improving sensitivity.

10:17: QUESTION: has the uncertainty from the theory predictions of production and decay been improved? ANSWER: calculations have improved, but the computations are looking at higher-order in QCD for gluon fusion and this comes with bigger errors. When you look at Vector-Boson Fusion, then it’s dominated by weak production and there the uncertainties are more like 5% – much smaller.  The Tevatron experiments integrate over the uncertainty on the rate. It’s a small contribution to experimental sensitivities for H->bb. For the other searches, it’s more important because gluon fusion is assumed to be the dominant production mechanism.

10:15: QUESTION: What are the chances for improving the bb channel? Answer: there are ways to improve. QUESTION: what if you know the mass? Does that make improvements at a specific mass easier? ANSWER: well, we can avoid the look-elsewhere effect by knowing where to look. Depends on what the LHC has to say later this week.

10:14: QUESTION: is WW still looking suppressed in the data with all the improvements? Answer: it’s more compatible with the Standard Model than before.

10:13: QUESTION: for signal strength, can you separate H->bb and H->WW and show the differennce? Wade: the exact number is not in the talk. At 125 GeV, the bb searches play a slightly greater role than WW in the searches.

10:13: APPLAUSE.

10:12: Wade wrapping up.

10:11: H->bb only: 2.9 sigma from null hypothesis (all channels: 2.5 sigma from null hypothesis) for Higgs masses between 115 – 140 GeV.

10:10: Wade reviews the history of the log likelihood plots – shows that they’ve consistently seen an excess in these plots at the low masses, though it’s better resolved as time goes on.

10:08: Tevatron data disagrees most with the background-only hypothesis at around 120 GeV, but BE CAREFUL – a spike is not strong evidence of a signal. A BROAD excess is what is expected when signal is present, so don’t over-interpret spikes in this plot. This is because their mass resolution (how well they can spot the exact Higgs mass) is not great.

10:07: One final interpretation – the “log likelihood distribution.”

10:05: Wade notes that their significance is still “below 3 sigma” and so they are “not making any announcements today.” Awful lot of effort to go through to not make any announcements…! 😉

10:03: Measurements consistent with Standard Model Higgs boson, but more than 2-sigma away from a “No Higgs” hypothesis.

10:01: Bottom line: combination of CDF and Dzero sees a 2.5 sigma (3.0 sigma) excess in their data including (excluding) the look-elsewhere effect. Breaking out the H->bb and H->WW channels separately, the H->bb is slightly higher than expected from the SM and the H->WW channel is just about consistent with what’s expected from a SM Higgs boson.

10:00: They not only don’t exclude the SM down around 110-140 GeV, they see what looks like a >2 sigma excess.

09:59: They expect to exclude SM Higgs bosons in many places, but not between 120-140 GeV in mass. That’s their expectation. Their goal was to be able to reject SM (if it’s not right) all the up to the limit from the W and top masses – about 150 GeV)

09:58: They combine their channels and plot in terms of significance. They can clearly see evidence of an excess around 115 GeV in their data in the high-sensitivity channels.

09:56: Wade now talking about combining the Dzero and CDF results.

09:55: Now he goes to interpreting the data not as upper limits on a production and decay rate, but in terms of whether or not the null hypothsis – “There is no Higgs boson” – is excluded. Taking into account the look-elsewhere effect, the deviation from the null hypothesis is 1.3 sigma at around 130 GeV.

09:54: Wade is giving a VERY good, if short, review of statistics and how to avoid interpreting a fluke as a real signal.

09:52: Dzero’s limits on the Higgs production and decay are inconsistent with the background-only hypothesis around 130 GeV. About 2 sigma? (local significance . . . doesn’t take into account the fact that you are looking across a wider range when searching for the Higgs)

09:52: OK, time for Dzero Higgs search results!

09:51: Some improvements haven’t made it in time for summer – expect more improvements later.

09:50: 1203 viewers now.

09:49: They’ve also added new reconstructed channels of Higgs production and decay. These channels tend to be sensitive to larger Higgs masses.

09:48: Also improved the ZH -> (ll)(bb). This is another channel sensitive to low Higgs masses. Based on improvements, expect 10-15% improvements in sensitivity.

09:46: Now talking about WH -> W(bb), where the W decays to lepton and neutrino. They’ve added more data, improved background modeling, and separate background into finer categories. The latter is the biggest improvement. Expect 10-17% improvements in sensitivity.

09:44: Also made improvements in H->WW search. They added all of the available data (didn’t have that for the Winter). They’ve improved background categorizations and lepton identification. Expect 5-10% improvements just from improving techniques (not including increase in data sample size – that is in addition to these 5-10% improvements).

09:43: Dzero improvements since winter. First, in the two-photon channel. They improved the modeling of background using simulation. They are better combating systematic uncertainties – uncertainties caused by measurement effects, and not just by the size of your data sample. To do this, they split their events into categories to separate and attack specific backgrounds more effectively. This means 20-30% improvements in sensitivity just in this channel. They’re down to about 10 times the Standard Model in sensitivity.

09:42: Dzero and CDF Higgs search programs are similar. Wade mentions also that they search for Higgs decays to two tau leptons (the heaviest cousin of the electron), “Higgs to tau tau”

09:41: Wade’s turn to speak for the Dzero results.

09:40: Eric notes that all existing Higgs searches are completely consistent with Standard Model expectations. More data will help us understand whether or not, if it exists, it behaves as predicted in the Standard Model.

09:37: About to go to Wade’s talk, but he wants to set the stage for the results we’ll see this week. Lots of Higgs masses excluded (at 1-times the Standard Model expectation), but not around 120-130 GeV. CDF sees ~2.5 sigma excess there for H->bb.

09:36: He makes the point that every time CDF has added more data and re-done the Higgs search, they’ve improved BEYOND what was expected just from adding data. They make improvements, and achieve great sensitivity.

09:35: Eric reminds us that the Tevatron is sensitive to about twice the Standard Model but in 2011 the LHC experiments were sensitive to only about 7 times the Standard Model.

09:33: Eric reviews the CDF results in the search for H->bb. They DO see a little excess, he reminds us, which is about 1.5 standard deviations above the Standard Model expectation at a Higgs mass of 125 GeV, and more than 2 standard deviations (sigma) from the expectation of NO Higgs boson at that mass.  The Higgs cross-section determined from this is about twice that predicted by the Standard Model, but consistent with the Standard Model within uncertainty.

09:32: They can also validate their neural networks used for selecting H->bb in these control samples of WZ and ZZ events, which are normally produced and well-established. They have confidence that, if there is a Higgs signal, they will see it.

09:30: They validate their b-jet tagging using the well-known decay of a Z boson to two bottom quarks. This is clearly evident in their data and they appear to model it well using their simulation.

09:28: Eric shows the power of bottom-jet (“b-jet”) tagging – this is when you use the characteristics of the jet to “tag” it as being from bottom quark decay and hadronization. Being able to b-jet tag means being able to separate bottom jets from charm, up, down, strange jets. This levels the playing field and removes a lot of background. After making 2 b-jet tags, the backgrounds are greatly reduced.

09:26: Here we go – Eric gets to Higgs decays to bottom-quark pairs, H -> bb. This is where the Tevatron SHINES in comparison to LHC (for now). Tevatron cannot look directly for H->bb – background is overwhelming. Instead, look for events where the Higgs is produced along side a Z or W boson, and look for the W or Z boson to “tag” the event and look for the bottom quark jets from the Higgs decay.

09:25: Summary: Tevatron is sensitive to about 3 times the Standard Model expected rate of H->WW. LHC, by comparison, more sensitive to about 1-times the Standard Model with last year’s data.

09:24: The lower the real Higgs mass, the harder and harder it will be to distinguish H->WW from background. No real evidence of the decay in the plots he is showing.

09:22: Eric shows the way they visualize the many H->WW categories. They rank them by purity – ones that contain more signal, a priori, are ranked higher to the right in the plot. This lets us easily see which places in the plot should contain signal and which are plagued with background.

09:21: Eric is reviewing the advanced multi-variate techniques used to select H -> WW signal events. Also reminds us that they can look at the number of jets in each event and use the different kinds of backgrounds present for 0, 1, or 2 jets to train a neural network to better reject background in different jet categories.

09:19: about 1000 people watching the seminar now. Up from about 500 at 09:00.,

09:17: Reviews Higgs decays to a pair of W-bosons (“Higgs to W W”). The W-bosons then decay to a electrically charged lepton and a neutrino – you get charged particles and missing energy from neutrinos. Makes it hard to reconstruct the Higgs mass as well as in ZZ or two-photon, and the backgrounds are very large because you give up certain constraints (like being able to actually see all the Higgs decay products directly).

09:15: Reviews the decay of the Higgs to a pair of photons (“Higgs to gamma gamma”). The backgrounds are bigger in this channel, but so is the decay rate of the Higgs to this final state. At the Tevatron, expect only to be able to exclude rates that are TEN TIMES the Standard Model prediction; LHC can hope to do a lot better. So Eric is making the point that these channels are important, but Tevatron not expected to have sensitivity to the Standard Model predictions for the ZZ and two-photon decays. LHC can reach these much better.

09:14 Tevatron expects only 0.2 events from the “golden channel” in the detector, based only on the production rate and decay rate of the Higgs to the golden mode. LHC expects a factor of 10 more such events due to bigger energy at LHC.

09:12: The “golden mode” is discussed, Higgs decay to a pair of Z bosons. Give a clean four-lepton signature in the final state. Even in 10/fb of Tevatron data, there are very few events that pass this topological requirement, making it easier to sift signal (Higgs) from background (stuff faking a Higgs).

09:09: Dominant Higgs production mechanism at both Tevatron and LHC is gluon fusion, when two gluons meet and produce a Higgs boson, which can then decay. Other production channels contribute, but at lower levels. The other mechanisms matter when you have messy Higgs decays, letting you reject background more effectively.

09:08: Eric reminds us that this is a huge team effort, including the big community running the accelerator (the Tevatron) and the experiments.

09:07: Eric reminds us that the top and W mass measurements from the Tevatron constrain the Higgs mass to be less than 152 GeV in the Standard Model.

09:06: Eric will give overview of the Higgs and review the CDF results from the Winter. Wade will present the updated Dzero results, which include improvements not available in the Winter.

09:04: two 30-minute presentation from CDF and DZero. Eric James(FNAL)  starts us off, then Wade Fisher (Michigan State)

08:55: Stream is live (I just needed to click that little link that talked about popping it up in another window).  Waiting for seminar to start. All the SMU people in the CERN office are tuning in (Tingting, Aidan, and me).

Ready to watch the Tevatron seminar on the Higgs

08:46: OK, all setup with a laptop and a WIRED internet connection (to avoid problems duue to poor wireless). Just waiting for the Fermilab live stream to actually start…

About Stephen Sekula

This entry was posted in Higgs Hunting, Physics Analysis, Physics Event. Bookmark the permalink.

Leave a Reply

Your email address will not be published. Required fields are marked *