[Rivet] New AGILe and Rivet versions to test

[Andy Buckley]

On 18/03/12 19:19, Anton Karneyeu wrote:
> Hi Andy,
> thanks for the update.
>
> 1. Patches are still necessary for 1.8.0beta0:
>
> - fix pyext build:
>
> https://svnweb.cern.ch/trac/GENSER/browser/GENSER3/pkgsrc/MCGenerators/rivet/trunk/patches/patch-pe?rev=7895

Oops! Sorry...

> - for some reason the rivet-nopy do compile on mac10.6 but do not
> compile on slc5 (see attached rivet-nopy.log), following patch fixed the
> problem:
>
> https://svnweb.cern.ch/trac/GENSER/browser/GENSER3/pkgsrc/MCGenerators/rivet/trunk/patches/patch-np?rev=7895

Hmm, odd. The .la file should contain the HepMC lib path metadata, 
AFAIK, but I've made the -L modification. I've not included the 
-R/-rpath, as we want the binaries to be relocatable and are hence 
avoiding use of RPATH in Rivet builds.

> 2. In fact the "Could not determine lifetime" error message comes from
> LHCB_2010_S8758301 (not from the LHCB_2011_I917009 which you updated
> recently). Please, see attached two replies from Frank and David. As I
> understand they proposed to complete internal particle table of
> LHCB_2010_S8758301:
>
> http://rivet.hepforge.org/trac/browser/trunk/src/Analyses/LHCB_2010_S8758301.cc#L201

Thanks: I've added the two genuine states and added an explicit 
exclusion of Pomeron/Reggeon/Odderon PDG codes from UnstableFinalState. 
I have left the warning message active for the unknown PYTHIA6 state 
pointed at by Frank, as I'd need a more complete diagnosis to justify an 
explicit blacklisting (there are many invalid PDG codes, why pick on 
this one?!) Hope that's not a big problem.

> 3. The log axis settings in aida2flat output is not a big problem. Let
> it be a feature request which would be nice to have one day :).

:-)

I've made another tarball, Rivet-1.8.0beta1: can you give it a quick 
test and let me know if I got it right this time? ;) If so, this will 
become out new release asap... at last!

Thanks again! This pre-release testing *really* helps :)

Andy



> 16.03.2012 19:40, Andy Buckley:
>> Hi again Anton,
>>
>> Apologies for the late response: I finally got time today to act on your
>> findings.
>>
>> The patches were very helpful and I've added (hopefully) equivalent
>> fixes to the code.
>>
>> For the LHCb analysis, I didn't find anything in my mail archive
>> agreeing to hide the lifetime-lookup error messages, but I've added a
>> protective check to the analysis which means that it'll only print an
>> error if the offending PID is a hadron: this should eliminate the
>> warnings about reggeons from herwig++ (although I would still be
>> concerned that you had status=2 Reggeons in the event record! David?)
>>
>> On the last issue (log axis settings in aida2flat output), there is no
>> support in aida2flat for histogram properties other than the title and
>> axis labels: they are only used in the output for make-plots produced by
>> compare-histos, although I appreciate that the two flat formats are very
>> similar. This can be improved when we finally release a version of Rivet
>> with our new histogramming package, where the data objects can hold
>> arbitrary annotations as in the make-plots format. Right now I would
>> have to significantly change the lighthisto module to add LogX/Y
>> annotation support, and it would just open the floodgates for requests
>> for all the other make-plots features that we don't support in
>> aida2flat. Hope you can work around it.
>>
>> I've made a new tarball and uploaded it to the usual place. It should
>> work fine at runtime, but if you could just test the compilation
>> (without patches) and let us know if the fixes have worked, that would
>> be great. Cheers!
>>
>> Andy
>>
>>
>> On 06/03/12 23:44, Anton Karneyeu wrote:
>>> Hi Andy,
>>>
>>> tests are finished fine, there are no any serious problems. Just want to
>>> mention a few issues:
>>>
>>> - Two patches are necessary to build Rivet on macosx 10.6 platform:
>>>
>>> Patch to build LHCB_2011_I917009.cc:
>>> https://svnweb.cern.ch/trac/GENSER/browser/GENSER3/pkgsrc/MCGenerators/rivet/trunk/patches/patch-ba?rev=7878
>>>
>>>
>>>
>>>
>>> Patch to build pyext:
>>> https://svnweb.cern.ch/trac/GENSER/browser/GENSER3/pkgsrc/MCGenerators/rivet/trunk/patches/patch-pe?rev=7878
>>>
>>>
>>>
>>>
>>>
>>> - There are still "Could not determine lifetime" error messages from
>>> LHCB_2010_S8758301 analysis. Please, see attached email for original
>>> issue report. It was agreed to mask particles which leads to error
>>> messages, but probably forgotten to be implemented in the analysis code.
>>>
>>>
>>> - There is mismatch in the field LogY of PLOT section of .dat files
>>> produced from reference .aida file and corresponding .plot file. For
>>> example for ATLAS_2010_S8894728:
>>>
>>> contents of
>>>
>>> /afs/cern.ch/sw/lcg/external/MCGenerators_hepmc2.06.05/rivet/1.8.0alpha0/x86_64-slc5-gcc43-opt/share/Rivet/ATLAS_2010_S8894728.plot
>>>
>>>
>>>
>>>
>>> and contents of *.dat files from
>>> aida2flat -s
>>> /afs/cern.ch/sw/lcg/external/MCGenerators_hepmc2.06.05/rivet/1.8.0alpha0/x86_64-slc5-gcc43-opt/share/Rivet/ATLAS_2010_S8894728.aida
>>>
>>>
>>>
>>>
>>> See also attached script logy.sh which demonstrate the issue on lxplus.
>>> The last issue is not specific for Rivet 1.8.0a, was also present in
>>> previous versions.
>>>
>>> Cheers,
>>> Anton
>>>
>>>
>>> 28.02.2012 0:43, Andy Buckley:
>>>> Hi Anton & Peter,
>>>>
>>>> We have new alpha/beta tarballs of AGILe 1.4.0a and Rivet 1.8.0a to
>>>> test
>>>> -- can you please work your customary and much appreciated mcplots
>>>> magic
>>>> to make sure we didn't foul anything up?
>>>>
>>>> The tarballs are in the usual places:
>>>> http://www.hepforge.org/downloads/agile
>>>> http://www.hepforge.org/downloads/rivet
>>>>
>>>> Thanks!
>>>> Andy
>>
>>
>> -- Dr Andy Buckley SUPA Advanced Research Fellow Particle Physics
>> Experiment Group, University of Edinburgh


-- Dr Andy Buckley SUPA Advanced Research Fellow Particle Physics 
Experiment Group, University of Edinburgh



-- 
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.