jump to navigation

CDF beats theory on the top pair cross section August 18, 2008

Posted by dorigo in news, physics, science.
Tags: , ,

Among the huge amount of beautiful new measurements produced at the Tevatron by the CDF and D0 experiments last month, just in time for showing at ICHEP 2008, the international conference in High-Energy Physics, there is one which does not make headlines, but it deserves one. It is the measurement of the top pair production cross section, a number which is by itself not terribly informative – it is basically only a check that perturbative calculations with Quantum Chromodynamics work well when they deal with an energy scale where the strong coupling constant is small enough. That is: the above is the only thing one gets from a precise measurement of the top cross section provided one is convinced that there is no other process, so far undiscovered, hiding in top production or top decay.

It is absolutely fair to ask oneself whether top pairs are produced at the Tevatron energy solely by quark-antiquark annihilation and gluon-gluon fusion, the two leading order QCD processes, or whether there is a heavy object X which decays to top quarks, thus enhancing the observed rate of top quarks over what QCD predicts. It is also perfectly legitimate to investigate whether the cross section is in line with predictions regardless of the final state in which one searches for top quarks: some non-standard decays of top could modify the mix. Further, one could hypothesize that the top quark dataset -the data enriched with top events which are used by the experiments to measure cross sections- contains some other process which messes up some of the measurements.

The above ideas are to me the most important reason for being interested, 14 years after I first got to know that the top quark existed, in the very precise new determinations of top quark pair cross section obtained by CDF. So let us look at the graph on the right, which details some of the recent determinations, which have been averaged into a result which carries a 8% total uncertainty, beating by 1% the most precise theoretical estimates (9% relative error).

One interesting thing to note is that the cross sections measured with SLT are higher than the average. SLT is the soft-lepton tagging algorithm, which tags b-quark jets coming from top decay through the identification of a muon or an electron embedded in the jet. In Run I, CDF measured a top cross section which was 9 picobarns when using SLT, while about 6 picobarns when using SVX tags -secondary vertices in the jets. Back then, the disagreement was the source of a huge controversy on the hypothesized presence of new physics in the sample of events containing SLT tags. The data did lend itself to some exotic interpretations, but things petered out after years of review and internal diatribas. Now, it does not look like there will be a reprise of that controversy, but the fact remains that SLT cross sections are still there: higher than they should be!

In any case, I salute this new, important result by the CDF top group, and by dozens of dedicated physicists who put their time and efforts into obtaining a very precise measurement. Now the ball is in the theorists’ court, to improve the precision on the theoretical estimate.

UPDATE – ok, a moment after posting the above piece, I looked back at the picture, and I realized that it is not true that the CDF determination is more accurate than theory. It is the theory band which has an 8% uncertainty if I am not mistaken, while CDF has the 9% measurement. That does not change much of the discussion, however, since once the result found by D0 is added to the above one, experiment does get the better hand.

UPDATE II: I also forgot to point interested readers to the public note describing the result!



1. Odysseus - August 18, 2008

It is the theory band which has an 8% uncertainty if I am not mistaken, while CDF has the 9% measurement. That does not change much of the discussion

While that may be true, it changes the validity of the post title. Anyway, good to know that theory still seems to hold here.

2. Fred - August 19, 2008

Odysseus, your assumption based on theory-over-CDF was correct until Glory chimed in and perhaps proved everyone wrong with her very own implied measurement of her top pair production cross section. And Glory… change your name to go along with your persona as it doesn’t quite evoke those visions of loveliness.

3. Guess Who - August 19, 2008

Dear Glory, TD is a happily married man, but since you seem to find physicists so irresistible, Esmeralda may have another prospect for you. šŸ˜€

4. dorigo - August 19, 2008

Fred, sorry – had to dump glory’s comment. WordPress is serious about bugging blogs which do not take care of spam.


5. Fred Altieri - August 19, 2008

Good move, T. I would do the same thing with those irritable incursions. I was just having a little early morning fun. On a totally unrelated note, have you been able to watch any of the Olympic events on television and are the Italians particularly interested?

6. Andrea Giammanco - August 19, 2008

Since a major part of the “theory uncertainty” is due to uncertainty on the Parton Distribution Functions, it’s not so fair to say that experiments are beating theory… The PDFs themselves are constrained by other experimental data (some of them coming from Tevatron itself).
Indeed, one can use in principle this same cross section measurement as an additional constraint to the PDFs (although it’s probably more effective to use the indirect measurement of the gg->tt/qq->tt ratio that you already mentioned).
In this table the top mass uncertainty is not considered part of the theory error (it is fixed to 175 GeV, as written in the bottom-right corner), but if considered as part of the “theory uncertainty” it would be a major part of it too. And indeed both CDF and D0, as you know, published papers where they use the cross section measurement as an indirect mass measurement.

It’s interesting to note how the situation for cross section measurements at hadron colliders is very different with respect to LEP, where small experimental errors were a challenge to theory in the sense that higher and higher order diagrams had to be calculated in a theory in order to keep the pace. Here, instead, if you had found a significant disagreement between theory and data, blaming the lack of a NN…LO QCD calculation would have not been necessarily the explanation, since a possible alternative explanation could have been an inadequacy of PDF modeling as well.

7. dorigo - August 19, 2008

Hi Fred,

sure, I know you like to indulge like that sometimes.

I do follow the olympic games, which are a major event in Italy. Federica Pellegrini, the swimming star from Spinea (Venice), was one of our favorites. I have followed live the 100m men’s final, with Bolt showing his name is not an accident. I would love to see some of the discus throw, but in Italian TV they don’t show it… We do not have a tradition on that specialy, although I used to practice it in my heydays.


8. dorigo - August 19, 2008

Hi Andrea,

while I concur that part of the theory uncertainty comes from the PDF -of the order of 0.3 picobarns as evaluated by Kidonakis and Vogt-, the experimental uncertainty also carries significant PDF errors…

In fact, it would be good to compare apples to apples, something which is usually not done in these kinds of comparisons. A NNLO theory prediction using a particular PDF set could be compared to an experimental determination using the same set, and a part of the common uncertainty could be removed.

Your point on theory carrying uncertainties because PDF “are constrained by experiment” is kind of a point in favor of experimental determinations, in my opinion. Wouldn’t we all love it if theory could determine PDF from first principles, rather than requiring a ton and a half of experimental input ?


9. dorigo - August 22, 2008

GW your comment had been caught by the spam filter. I recovered it, but maybe I shouldn’t have – it makes reference to another one which I myself removed because it is a sex advertisement…


10. Guess Who - August 22, 2008

No problem TD, but that new spamfilter is a real killjoy. It did not only eat the comment, it also removed the URL from the anchor tag. Considering that it pointed to another comment on this blog, that’s pretty draconian!

Unless that filter becomes a little more discerning, my happy days of posting silly two-line comments + funny URL are over. šŸ˜¦

11. Afron - August 26, 2008

I’ve been getting the sense the CDF does better physics then D0. Which experiment do you think does better at the LHC, CMS or Atlas?

12. dorigo - August 26, 2008

Hi Afron,

CDF has a history of outperforming D0 that dates back to Run I -when there was really no competition: D0 did not have a inner silicon tracker, lacked an axial magnetic field, and had problems from main ring backgrounds. In Run II the two detectors are more similar, but still CDF does better in all but a few measurements. I think it is a combination of several factors.

Concerning LHC, ATLAS on paper looks the better detector, but things are not so clear if we look at some details. I am concerned with the hadron calorimeter of CMS, but its electromagnetic calorimeter rocks. ATLAS has a more balanced system there. The CMS muon system is a strong point.
All in all, I think the two detectors should be providing a similar sensitivity. The challenge is really on which will be the collaboration with more skills, a better organization, and less internal conflicts.

One thing to note is that ATLAS is really the “CERN experiment”, and retains some slight but visible advantages. One of them being its location (CMS is miles away from the lab, ATLAS is across the road). I discussed another one here.

Another thing: be wary with projections.

Sorry comments are closed for this entry

%d bloggers like this: