[Rivet] bug in ATLAS_2014_I1325553 (inclusive jets)

[RICHARDSON P.]

Hi,

  So here's the plots with the head of rivet and the first version in the repository

http://herwig.hepforge.org/images/rivet/

  there's also the early ATLAS measurements from the same runs of Herwig++. The stats are still a bit low as I was still checking the new analyses looked reasonable before running the high stats but I don't see how the 2014 one can be consistent without the factor of 0.5 in the r4728 line and not the r4665 line. Also have I missed something in the paper as unless there's some fine print I've missed something it is measuring dsigma/dy not dsigma/d|y|.

  Also I think the point David was making is that if the validation was not with the final version of the code including all the factors in the code and not adding them by hand afterwards then its not a true validation.

Peter

________________________________
From: Holger Schulz [hschulz at physik.hu-berlin.de]
Sent: 21 May 2015 22:18
To: Christian Gütschow
Cc: RICHARDSON P.; Rivet; Vojtech Pleskot
Subject: Re: [Rivet] bug in ATLAS_2014_I1325553 (inclusive jets)

On 21/05/15 22:06, Christian Gütschow wrote:
That is the normalisation factor ( = cross section / sum of weights), i.e.

AMI xsec = 2.9675E+07 * 1000.
SumOfWeights = 1.5358320748e+19
--> scale factor = 1.9321773836416558e-09

which I calculated manually at the time, without using the finalise method.

That makes sense. I wonder nothing really has changed in the code since then.

Peter, would you be able to run H++ with an earlier version of rivet to
see if there is the same discrepancy or not?

Holger


Cheers,
Chris

On 21 May 2015 at 21:58, Holger Schulz <hschulz at physik.hu-berlin.de<mailto:hschulz at physik.hu-berlin.de>> wrote:
Hmm, there is a scale attribute in the source files of the plots:

E.g. in http://cgutscho.web.cern.ch/cgutscho/ATLAS_2014_I1325553/d01-x01-y02.dat:

# BEGIN HISTOGRAM /Users/geon/Documents/rivetArea/stdm2013_11_inclusiveJets/mc.yoda/ATLAS_2014_I1325553/d01-x01-y02
LineColor={[HTML]{EE3311}}
LineStyle=solid
Path=/Users/geon/Documents/rivetArea/stdm2013_11_inclusiveJets/mc.yoda/ATLAS_2014_I1325553/d01-x01-y02
Scale=1.9321773836416558e-09
Title=MC
XLabel=
YLabel=

Can you find out what that Scale thing was for/from?

Thanks,
Holger



On 21/05/15 21:53, Christian Gütschow wrote:
Yep, they live here:

http://cgutscho.web.cern.ch/cgutscho/ATLAS_2014_I1325553/

Cheers,
Chris


On 21 May 2015 at 21:11, Holger Schulz <hschulz at physik.hu-berlin.de<mailto:hschulz at physik.hu-berlin.de>> wrote:
Christian,

do you think you can still find the validation plots from back then?

Holger


On 21/05/15 20:20, RICHARDSON P. wrote:
Dear Christian,

  So I was comparing the results with the latest version of Heriwg++ as part of our testing before the next release and we were much further off than I expected. After looking at the analysis and the paper I believed the paper was plotting d sigma /dy/dpT and the analysis d sigma/d|y|/dpT which differ by the factor of 1/2. I contacted Vojtech and he agreed it was wrong so I made the change.

  If it is wrong we can change it back

Best Wishes

Peter
________________________________
From: Christian Gütschow [chris.g at cern.ch<mailto:chris.g at cern.ch>]
Sent: 21 May 2015 18:44
To: Rivet; RICHARDSON P.
Subject: bug in ATLAS_2014_I1325553 (inclusive jets)

Dear Rivet authors,

it looks like a bug has been introduced in the current trunk version of the ATLAS inclusive jets measurement (the version from the last release is fine). Specifically, the normalisation factor in the finalise routine has been modified:

https://rivet.hepforge.org/trac/browser/src/Analyses/ATLAS_2014_I1325553.cc#L83

to include an additional factor of 0.5 -- apparently to fix an issue with the normalisation.

There should be no such factor as absolute rapidity has been used consistently throughout the analysis and such a factor is only really necessary if absolute and signed rapidities are being mixed (axis label vs cross section differential).

Note that the original version (without 0.5) has been validated and no issues with the normalisation have been observed, so I would recommend to unfix this again. :-)


Many thanks,
Chris

--

 Dr. Christian Gütschow

 TU Dresden
 Institut für Kern- und Teilchenphysik
 Zellescher Weg 19
 01069 Dresden

 > E17, Andreas-Schubert-Bau
 > chris.g at cern.ch<mailto:chris.g at cern.ch>




_______________________________________________
Rivet mailing list
Rivet at projects.hepforge.org<mailto:Rivet at projects.hepforge.org>
https://www.hepforge.org/lists/listinfo/rivet




--

 Dr. Christian Gütschow

 TU Dresden
 Institut für Kern- und Teilchenphysik
 Zellescher Weg 19
 01069 Dresden

 > E17, Andreas-Schubert-Bau
 > chris.g at cern.ch<mailto:chris.g at cern.ch>





--

 Dr. Christian Gütschow

 TU Dresden
 Institut für Kern- und Teilchenphysik
 Zellescher Weg 19
 01069 Dresden

 > E17, Andreas-Schubert-Bau
 > chris.g at cern.ch<mailto:chris.g at cern.ch>


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://www.hepforge.org/lists-archive/rivet/attachments/20150522/9c23ff12/attachment.html>