I would like to understand how experimentalists search for new physics at the LHC. Lets imagine I want to use the LHC data to put a bound on the coupling of some new physics effective operator, say, for example a flavor changing Yukawa interaction of the form $t_R c_LH$ (top-charm-Higgs).

What I have understood up to now is the following (correct me if I'm wrong): one must first look at the interesting channels that involve these new interactions e.g. $p p \to t \bar t$ followed by the new decay $t\to Hc$. Next, assume the new effective coupling is such that the Branching ratio lies within the current experimental uncertaintes $Br(t\to Hc)\sim 1\%$. Then you have to simulate (with Monte Carlo) the relevant processes, count the number of expected events (after subtracting the SM background) and finally compare it to the real Data. This should give an upper bound on the effective coupling.

Can someone give me a detailed description of these type of LHC searches? I am particularly interested in understanding the role of the MC simulations (MadGraph, Pythia, etc). Thanks!


You pretty much have the right ideas here. The MC is necessary because the functional dependence of the detector response and the underlying physics is too complex to calculate in any other way. And therein lies the actual analysis problem: it's not enough to simulate the physics based on the standard model (or some other hypothetical interaction). In order to compare to the real world, those event patters have to be sent trough a calibrated beam/detector model. I know that the software tools are public, but I doubt that enough of the detector calibration data is available to do non-standard physics analyses outside of the collaborations. It seems unlikely that one could reliably discover some new physics that the folks in the collaboration haven't seen, already. If it's obvious, everybody and their grandma has seen it already, and if it is hidden in the data, then it takes an enormous effort and all the available data to coax it out.

I may be wrong, but I think your best bet to do this kind of work is to actually become part of a collaboration and to be shown the ropes (and being given access to ALL the pieces that you need) there.

This, by the way, is not just true for LHC. The very same problem applies for other experiments like the x-ray and gamma satellites, neutrino observatories etc.. People are frequently attempting to do third party data analysis on the public data sets, but it's always very questionable how good the results of that are without knowing the quirks of the detector errors.

  • $\begingroup$ Ok, so the role of the MC simulations is to model how the detector will respond to your new interaction, i.e. it tells you the value of the efficiency $\epsilon$. I am not sure if I agree with what you say after. I don't see why 3rd parties can't use the CMS and ATLAS searches to give improved bounds on new physics. We can also use the same detector simulations they use. Sometimes they base their analysis on a given number of processes and leave out other possible competing processes that, if included, could give better bounds on effective operators. $\endgroup$ – Dar Dec 20 '14 at 10:52
  • $\begingroup$ I am not talking about discovering a new particle hidden in the data (that I agree is very unlikely), I am just talking about given improved bounds on new physics. $\endgroup$ – Dar Dec 20 '14 at 11:14
  • $\begingroup$ MCs are used to model physical hypotheses, the detector, the trigger system, the filters, pretty much everything that is between the physical vacuum and the publication. Of course you can use the public data to perform searches... but they won't be as sensitive as a search by members of the collaborations will be. Moreover, I personally, won't pay much attention to your results unless the collaborations (in this case both of them!) verify them independently. Strictly speaking, I won't care, at all. $\endgroup$ – CuriousOne Dec 20 '14 at 11:16
  • $\begingroup$ Are your telling me that 3rd parties cannot recast the known searches of CMS and Atlas to improve the bounds they give on a specific effective operator? $\endgroup$ – Dar Dec 20 '14 at 12:56
  • 1
    $\begingroup$ I told you that you had all the right ideas. It seems that the only thing that you don't like is that I don't give much credence to analyses not done by the groups who have the actual manpower to build the experiments. Having said that, what, exactly, stops you from joining? If you have the funding to do the analytic work, all you need is to ask to become part of the collaboration. That's how science works and it's not the actual rocket science part of the business. $\endgroup$ – CuriousOne Dec 20 '14 at 14:50

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.