Tutorial - Part 1: Our First Model Point Test

The Result

If no error occured, CheckMATE should have produced an output similar to the following: Evaluating Results Test: Calculation of r = signal/(95%CL limit on signal) Result: Excluded Result for r: 2.2313438259 Analysis: atlas_1405_7875 SR: SR05_a.6jl

Let us understand these lines one by one:

As explained before, in the standard setup CheckMATE compares the model prediction to the upper limit by calculating r.
The most important information is told in the Result line, which tells us whether the input model point is excluded or allowed.
The r-value which was used to find the result is given here. If r is larger than 1 it means that one predicts more than is allowed.
The most sensitive signal region (according to the prescription explained on the previous page) that set the limit is given as well.

Check out the exclusion limit ATLAS quotes for the same analysis using the same model. Does our answer agree?

[Show Answer]

Note that due to the limited disk space on the virtual machine we could only create a rather small sample. Since CheckMATE uses the statistical error to weaken its exclusion statement (remember the numerator S-1.96 dS in r), it will produce weaker limits if the statistics is poor. Fortunately, the model point we considered was such that this effect still lead to excluded. However, if we tried to reproduce the precise exclusion line with r=1, we would need more simulated events to find agreement.

But is this really the best LHC8 could do?

In the above example run, we already told you which of the many analyses you should use for testing and we argued why this analysis should be sensitive to our model. However, maybe CheckMATE contains a different analysis which might have been even more sensitive. Often it might not be obvious at all for the user which analysis is actually the most sensitive. As a simple example, if a BSM decay chain produces W bosons it might not be clear at first if its hadronic or its leptonic decay modes prove to be easier to distinguish from Standard Model. Fortunately, the user does not have to set the Analyses keyword. If this parameter is not set, CheckMATE will automatically test against all implemented 8 TeV results from ATLAS and CMS.

Use your previous CheckMATE input parameter file and remove the Analyses keyword. Give this new run a new name cMSSM_2. Look at the long list of analyses CheckMATE will now test your input against. Guess which one might be the most sensitive one to our input model. Then, run CheckMATE and see whether you were right!
[Show Answer]

In principle there is no real disadvantage in always testing all analyses at once, even if one knew which analysis is the best. The computational time is increased by not much more than a factor of two which compared to the time required for event generation is completely negligible. Still, in some cases the users might want to restrict the testing to one (or few) of the implemented analyses. In thas case they are of course always free to specify the to-be-tested analysis list via the Analyses keyword as used before.

With this example, you know all the basics to test your model against LHC results using CheckMATE. The only thing you need to be able to do is using one of the many event generators on the market to produce .hep or .hepmc files plus the corresponding cross sections for the most relevant production times decay scenarios. Then you can straightforwardly put these into CheckMATE to identify excluded and allowed parameter regions.
Click to finish exercise 1.