|
[Rivet] two ATLAS analyses to be included in RivetFrank Siegert frank.siegert at cern.chWed Sep 26 20:48:02 BST 2012
D'oh, of course. Thanks for spotting this and confirming that it still works correctly for your unweighted samples. I have committed this into Rivet's SVN now. Cheers, Frank On 26 September 2012 14:38, Roman Lysak <lysak at fzu.cz> wrote: > > Hi Frank, > > you are right, your changes made the code more general and clear. > However, there was a bug in your version of code, I fixed this (see > attachment), tested and the results are the same as before. > > Thanks, > Roman > > > > On 09/26/2012 12:14 PM, Frank Siegert wrote: >> >> Dear Roman, >> >> Thank you for these analyses. I have one comment on ATLAS_2011_I954993 >> (the WZ one). >> >> When you calculate the statistical error on the fiducial cross >> section, you are using >> fidXS_err = fidXS/std::sqrt(AllPassed); >> This isn't going to work in the general case with event weights, even >> though AllPassed is correctly built using the weights. >> In fact, it is going to be much simpler to use a Rivet histogram right >> away for keeping track of "AllPassed", and just normalise it correctly >> in the end. >> >> I'm attaching a tarball where I have made this modification, which >> also makes the code clearer and shorter. Note, that we can use a >> histogram here without correcting for the unwanted bin width division >> because the bin width is chosen as 1. >> >> Let me know if you can confirm that these changes still reproduce your >> result correctly (I have only tested that it compiles so far), and >> I'll commit the analysis into Rivet's SVN. >> >> Cheers, >> Frank >> >> On 26 September 2012 11:24, Roman Lysak <lysak at fzu.cz> wrote: >>> >>> Dear Rivet authors, >>> >>> we have prepared two analyses from Atlas experiment which we would like >>> to >>> include into Rivet. >>> We validated that they produce consistent results comparing to results in >>> the papers. >>> All necessary information can be found in the attachment. >>> >>> Please let me know if anything else is required. >>> >>> Thanks, >>> Roman Lysak >>> >>> >>> _______________________________________________ >>> Rivet mailing list >>> Rivet at projects.hepforge.org >>> http://www.hepforge.org/lists/listinfo/rivet > >
More information about the Rivet mailing list |