[Rivet] Rivet and Fragmentation off

[Holger Schulz]

Hi,

Frank pointed out that there are crashes in LEP analyses, when
running Sherpa in a configuration where Fragmentation is off.

I found that the culprit are events with exactly two charged
partons in the event with vertices as shown below.

Due to the mass of the quarks being zero, and thrust being
exactly 1 for this event, the calculation of e.g.

    const double rapidityT = 0.5 * std::log((energy + momT) / (energy -
    momT));

in ALEPH_1996_S3486095 leads to

  * division by zero as energy == momT for the quarks (energy being the
    energy of
    the particle and momT the projection of the momentum vector onto
    the thrust axis)
  * and subsequently to rapidityT being infinite (inf)

The crash occurs when rapidityT is filled into a histogram:

  * _histRapidityT->fill(fabs(rapidityT), weight);


So I wonder what to do in these pathological cases.

I was using the head versions of yoda and rivet to reproduce this problem.



Cheers,
Holger


I: 4     10005      -11 -0.00e+00,-0.00e+00,-4.56e+01,+4.56e+01  
11        -6
           10007       11 -0.00e+00,-0.00e+00,+4.56e+01,+4.56e+01  
11        -6
           10011        3 -3.22e+01,-4.78e+00,+3.20e+01,+4.56e+01   
3        -6
           10012       -3 +3.22e+01,+4.78e+00,-3.20e+01,+4.56e+01   
3        -6
  O: 7     10009       11 +0.00e+00,+0.00e+00,+4.56e+01,+4.56e+01   
3        -5
           10010      -11 +0.00e+00,+0.00e+00,-4.56e+01,+4.56e+01   
3        -5
           10013        3 -2.66e+01,-3.96e+00,+2.65e+01,+3.77e+01   1
           10014       -3 +1.80e+01,+8.08e+00,-1.87e+01,+2.72e+01   1
           10015       21 +1.19e+01,-1.85e+00,-1.34e+01,+1.80e+01   1
           10016       21 +1.90e-01,-1.28e+00,+4.15e+00,+4.35e+00   1
           10017       21 -3.47e+00,-9.85e-01,+1.51e+00,+3.91e+00   1
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://www.hepforge.org/lists-archive/rivet/attachments/20140807/4064e24b/attachment.html>