jump to navigation

New D0 Higgs limit combination for 1.7/fb December 18, 2007

Posted by dorigo in internet, news, physics, science.

Just a few days ago D0 released their new results from searches of the H \to WW  decay process. That result has now been combined with all their other searches, in particular those for a lighter Higgs boson, which mostly decays to a b-quark pair if M_H<135 GeV as the following graph explains.


In the graph, the line marked “bb” shows the fraction of times a H boson disintegrates into a pair of b-quarks, which subsequently – given the large energy each is endowed with, equal to M_H/2 – produces two hadronic jets in the D0 detector. The line marked “WW” shows instead the dominance of the WW final state if the Higgs mass is larger, an effect due to the fact that the elusive particle decays to the heaviest bodies available: as soon as two W bosons are energetically possible (and even a little before that, when one W is produced with less energy than its rest mass) they become the dominant decay. Note also the three regions with background color yellow, pale blue, and bluer blue: yellow marks the region where LEP II has excluded a Higgs boson, while pale blue marks the region where the bb final state is dominant, the so-called “light-higgs” region. 

To find a H \to b \bar b decay one needs some extra handle, because two b-jets are a quite common occurrence at the Tevatron – one event in a hundred thousand, as opposed to one in a hundred billion. The extra handle is provided by requiring that the H comes together with a W or Z boson, in a process called associated production, or Higgs-strahlung from vector bosons – the latter adapted from the german word bremsstrahlung, which means braking radiation, the one produced by a charged particle forced to slow down. Electrons (as well as any other charged body)produce real photons by bremsstrahlung when they are curved or decelerated, while W and Z bosons “emit” real H bosons when they have enough energy to do so.

D0’s combination is a pleasure to look at, because for once I do not need to wrestle too much into gory details, and can concentrate on describing the results. And you, dear reader, are presented with a series of very nice and clear plots, where everything is tidy and whose message is clear.

The D0 plots are made all with the same style: a neural network classifier is used to discriminate signal from background, and you get to see the data (black points) in a histogram of the NN output, compared to the expected sum of backgrounds (all together in a single distribution, marked with a red line).

The agreement of data and background speaks by itself, but in addition you get to see how the Higgs signal would show up in the distribution. To see it better, it is multiplied by a factor of x15. Thank god, a common factor across the board (for all WH/ZH searches, while the higher S/N of H \to WW searches forces the removal of the enhance factor), so that one can easily compare visually the statistical power of different searches. Kudos to D0 for their excellence in data display!

So here are the individual search plots. They are based on data samples ranging from 0.6 to 1.7 inverse femtobarns.


Above, the WH \rightarrow e \nu bb search results. The expected signal is in blue, multiplied by a factor of 15.


Above, the WH \rightarrow \mu \nu b b  search results. The expected signal is in blue, multiplied by a factor of 15.


Above, the ZH \rightarrow \nu \nu bb search results. The expected signal is in blue, multiplied by a factor of 15.


Above, the ZH \rightarrow l l b b search results (l stands for a charged lepton, either electron or muon). The expected signal is in blue, multiplied by a factor of 15. This plot is made with events containing two b-tagged jets, for maximum signal purity.


Above, the ZH \rightarrow l l b b search results (l stands for a charged lepton, either electron or muon). The expected signal is in blue, multiplied by a factor of 15. This plot is made with events containing only one b-tagged jet, for a larger signal efficiency.


Above, the WH \rightarrow W W W search results. The expected signal is in blue, multiplied by a factor of 15. This analysis searches for the spectacular signal of three W bosons, which is difficult to see because of the price to pay in branching fraction of W bosons to leptons. Backgrounds, however, are smaller than in most other searches. Here, instead than a NN output, a discriminant is plot on the x axis. Its meaning, however, is exactly the same: a value close to 1 is more likely for signal-like events.

Above, the H \to WW search result from the latest chunk of 0.6/fb data analyzed by D0. Here, the signal (still in blue) is not multiplied by 15, but the y axis is logarithmic.

Finally, all searches are combined to extract a global 95% confidence level limit. You can see it in the plot below. The limit, as usual with Higgs searches at the Tevatron, is on the number of times the Higgs production could exceed the predicted Standard Model rate without having been seen by D0.

The black curve is the limit found by D0, while the hatched red line is the limit expected given the sensitivity of the searches and the data available. You can see that a big progress has been made in the searches at low Higgs mass, where the limit is sitting at only a few times the SM. This leaves room to hopes that the Tevatron may say something meaningful also in the low mass region before LHC becomes the only game in town.

For more information, please visit the D0 site of this analysis.


1. Nikita Nikolaev - December 27, 2007

A couple of questions, may I?

1. Why exactly is it harder to detect the WWW signal, rather than a WW signal? That is, why a three-pair is more likely to branch to leptons, than a two-pair?

2. In the first graph on top, why the bb-curve and the ZZ-curve so neatly touch in the darker-blue region, as if their inflections are equal in some vicinity there?

3. You didn’t explain what dark-blue region means. Could you drop another word or two on what it means? May be the answer would both explain questions this one and number 2.

4. In the first graph again, all the curves \tau^\ast \tau^-, gg, c\overline{c}, behave as if they would like to tend to some sort of a similar limit there in the dark blue region again. Why?

Sorry, perhaps this is a bunch of useless questions. =) Tell me if they are such

Thank you,

2. dorigo - December 27, 2007

Ciao Nikita,

my internet connection is lousy so I only answer this one comment today… And it too quite quickly.
1. H->WW is direct production, and has a cross section 4 times larger than WH->WWW. So detecting WW is better, the more so because every time you ask for a W->l nu decay you pay a branching ratio price of 2/9ths in terms of total rate.
2. they both fall down quickly at 160 because it is where the H->WW decay resonates and the Higgs decays dominantly so.
3. the dark blue region is called “high mass region”, in the sense that it is the mass region where the Higgs is best searched in WW or ZZ.
4. because when high mass decays open up, the Higgs wants to use them. The Higgs couples to particles in measure proportional to their masses, so it preferentially decays to the heaviest ones. As its mass is high enough, the other decays become suppressed.


3. Nikita Nikolaev - December 28, 2007

Thank you very much for your answers! All of this is really very interesting. Sometimes I can’t believe I am (and can’t wait to be) going to go deeper and deeper into all of this! 😀

And don’t worry about responding quickly or frequently — you better enjoy skiing. I don’t know about you, but here in Toronto we finally have some nice weather in terms of the snow, and especially up in highlands (~100-140 km north of Toronto) the trail have been nothing but lovely in the past month or so, so it is always really nice to get up there and ski (cross-country, that is). And all of this is a abrupt contrast with last year, when most of the ski courses had to close for the season, because of the lack of snow!


4. dorigo - December 29, 2007

The area around Toronto is certainly good for skiing… I guess you have a good time. I never tried biathlon, but I think I would like to try Alan Turing’s version, the “move and run around the block” version of chess. You play a move, run around the house, and when you’re back there your opponent has to move. Should be very interesting to try.


Sorry comments are closed for this entry

%d bloggers like this: