[Rivet] BUG fix!! HEpData -> Yoda problem: Crash for HepData with large (asymmetric) pT bin

[Kristin Lohwasser]

Dear all,

I have imported HepData and initialized a Yoda histogram in rivet with 
this. I however get now a crash for a histogram, with large pT bin, 
where I believe, the export algorithm from the HEPdata does have a small 
bug:

I want to plot this histogram
http://hepdata.cedar.ac.uk/view/ins1282447/d6

This is what is written in the *yoda file:

# BEGIN YODA_SCATTER2D /REF/ATLAS_2014_I1282447/d06-x01-y04
Path=/REF/ATLAS_2014_I1282447/d06-x01-y04
Type=Scatter2D
# xval   xerr-   xerr+   yval   yerr-   yerr+
10.0	2.0	2.0			0.38	0.03605551275463989	0.03605551275463989
16.0	4.0	4.0			0.35	0.0282842712474619	0.0282842712474619
30.0	10.0	10.0			0.23	0.01414213562373095	0.01414213562373095
60.0	6940.0	20.0			0.085	0.009433981132056603	0.009433981132056603
# END YODA_SCATTER2D

And when I initialize the histogram with this setting, I get an error 
message and a crash (namely:
Rivet_i             FATAL  Standard std::exception is caught
Rivet_i             ERROR Bin edges overlap: 80 -> 8
--> the full output is pasted below after the signature).

This is solved by reversing the xerr- and xerr+, so I believe, this 
should be somehow fixed in the calculation of the bin borders in the HEP 
data (?) scripts.

30.0    10.0    10.0            0.26    0.03162277660168379 
0.03162277660168379
60.0    20.0    6940.0            0.094    0.01676305461424021 
0.01676305461424021
# END YODA_SCATTER2D

Thanks and best
Kristin

Rivet_i             FATAL  Standard std::exception is caught
Rivet_i             ERROR Bin edges overlap: 80 -> 8
AthAlgSeq           FATAL  Standard std::exception is caught
AthAlgSeq           ERROR Bin edges overlap: 80 -> 8
AthMasterSeq        FATAL  Standard std::exception is caught
AthMasterSeq        ERROR Bin edges overlap: 80 -> 8
Traceback (most recent call last):
   File 
"/afs/cern.ch/atlas/software/builds/nightlies/devval/AtlasCore/rel_2/InstallArea/jobOptions/AthenaCommon/runbatch.py", 
line 11, in <module>
     theApp.run()     # runs until theApp.EvtMax events reached
   File 
"/afs/cern.ch/atlas/software/builds/nightlies/devval/AtlasCore/rel_2/InstallArea/python/AthenaCommon/AppMgr.py", 
line 627, in run
     sc = self.getHandle()._evtpro.executeRun( nEvt )
RuntimeError: (file "", line 0) Bin edges overlap: 80 -> 8 (C++ exception)
StoreGateSvc         INFO Stop StoreGateSvc
StoreGateSvc         INFO stop: setting service priority to 101 so that 
event stores get finalized and cleared before other stores
TagMetaDataStore     INFO Stop TagMetaDataStore
../../EVGEN/nom/amcatnlo_nom_340.pool.root Info Database being retired...
Domain[ROOT_All] Info ->  Deaccess DbDatabase   READ [ROOT_All] 
EB46D5BB-C889-814A-AA59-490518DA41E1
MetaDataStore        INFO Stop MetaDataStore
InputMetaDataStore   INFO Stop InputMetaDataStore
ApplicationMgr       INFO Application Manager Stopped successfully
Rivet_i              INFO Rivet_i finalizing
THistSvc          WARNING no stream "Rivet" associated with id: 
"/Rivet/ATLAS_2012_I1282447/d01_x01_y01"
Rivet_i              INFO TH1D /ATLAS_2012_I1282447/d01_x01_y01 created 
from YODA::Histo1D
THistSvc          WARNING no stream "Rivet" associated with id: 
"/Rivet/ATLAS_2012_I1282447/d01_x01_y02"
Rivet_i              INFO TH1D /ATLAS_2012_I1282447/d01_x01_y02 created 
from YODA::Histo1D
THistSvc          WARNING no stream "Rivet" associated with id: 
"/Rivet/ATLAS_2012_I1282447/d01_x01_y03"
Rivet_i              INFO TH1D /ATLAS_2012_I1282447/d01_x01_y03 created 
from YODA::Histo1D
THistSvc          WARNING no stream "Rivet" associated with id: 
"/Rivet/ATLAS_2012_I1282447/d03_x01_y01"
Rivet_i              INFO TH1D /ATLAS_2012_I1282447/d03_x01_y01 created 
from YODA::Histo1D
THistSvc          WARNING no stream "Rivet" associated with id: 
"/Rivet/ATLAS_2012_I1282447/d03_x01_y02"
Rivet_i              INFO TH1D /ATLAS_2012_I1282447/d03_x01_y02 created 
from YODA::Histo1D
THistSvc          WARNING no stream "Rivet" associated with id: 
"/Rivet/ATLAS_2012_I1282447/d04_x01_y01"
Rivet_i              INFO TH1D /ATLAS_2012_I1282447/d04_x01_y01 created 
from YODA::Histo1D
THistSvc          WARNING no stream "Rivet" associated with id: 
"/Rivet/ATLAS_2012_I1282447/d04_x01_y02"
Rivet_i              INFO TH1D /ATLAS_2012_I1282447/d04_x01_y02 created 
from YODA::Histo1D
THistSvc          WARNING no stream "Rivet" associated with id: 
"/Rivet/ATLAS_2012_I1282447/d04_x01_y03"
Rivet_i              INFO TH1D /ATLAS_2012_I1282447/d04_x01_y03 created 
from YODA::Histo1D
THistSvc          WARNING no stream "Rivet" associated with id: 
"/Rivet/ATLAS_2012_I1282447/d04_x01_y04"
Rivet_i              INFO TH1D /ATLAS_2012_I1282447/d04_x01_y04 created 
from YODA::Histo1D
THistSvc          WARNING no stream "Rivet" associated with id: 
"/Rivet/ATLAS_2012_I1282447/d05_x01_y01"
R

On 25.07.2014 11:39, Kristin Lohwasser wrote:
> Hi all,
>
> thanks for the help, it seems to be working now.
>
> One thing, that I was not aware of... --> you need to change the 
> *.plot file *EVEN* if you are using the pre-defined histograms from 
> HEP Data. (Not sure, whether I overlooked this or whether it should be 
> mentioned somewhere).
>
> I have another, slightly odd request / question (mainly to RIVET):
>
> Is there any possibility to select the leptons (virtual W bosons) from 
> semileptonic heavy flavour decays? And then feed these as veto 
> particles into the Jet Finding?
>
> Thanks & Cheers
> Kristin
>
>
> On 24.07.2014 11:51, Andy Buckley wrote:
>> D'oh, my mail client malfunctioned and forwarded your own message back
>> to you. Damn thing. Here is the one I meant to re-send:
>>
>> Andy
>>
>>
>> -------- Original Message --------
>> Subject: Re: [Yoda] Plotting with Yoda?
>> Date: Sat, 19 Jul 2014 22:38:30 +0100
>> From: Andy Buckley <andy.buckley at cern.ch>
>> Reply-To: andy.buckley at cern.ch
>> To: Email the developers (subscription not required)
>> <yoda at projects.hepforge.org>
>>
>> On 18/07/14 16:57, Kristin Lohwasser wrote:
>>> Dear all,
>>>
>>> I am a newcomer to Rivet/Yoda and I am all alone in my quest....
>> Hi Kristin,
>>
>> Hope we can make it less lonely :-)
>>
>>> I am trying to implement an ATLAS analysis for rivet and I have 
>>> achieved
>>> to have a full routine because of some features of the jet routine, I
>>> thought I'd test the plotting side of things.
>>>
>>> Basically, what I have done is:
>>>
>>> 1) Download / create analysis
>>> rivet-mkanalysis ATLAS_2012_I1282447
>>>
>>> 2) tried to run yoda on it:
>>>
>>> yoda2flat ATLAS_2012_I1282447.yoda
>>> (there are some problems, output: failing to parse this line...)
>> Hmm, interesting... can you send use those error messages?
>>
>>> make-plots ATLAS_2010_S8919674.dat yields:
>>>
>>> %make-plots ATLAS_2012_I1282447.dat
>>> Making 1 plot
>>> Plotting ATLAS_2012_I1282447.dat (0 remaining)
>>> Error: max() arg is an empty sequence
>>>
>>> it seems there are some problems with the data definitions / how the
>>> data is defined in HEP Data (and I guess, yes, there are some
>>> exceptional tables, e.g. uncertainty matrices etc.), so I thought maybe
>>> I'll try it just for the first 2 or 3 histos to get a feel, however not
>>> even that works:
>>>
>>> yoda2flat test.yoda (as attached)
>>>
>>> % make-plots test.dat
>>> Making 1 plot
>>> Plotting test.dat (0 remaining)
>>> dvips: ! DVI file can't be opened.
>> I think the problem here is that you're trying to use yoda2flat which is
>> *not* intended to produce files for plotting via make-plots, although
>> the two "flat" formats are similar.
>>
>> Have you tried running rivet-mkhtml or rivet-cmphistos + make-plots on
>> your .yoda files?
>>
>>> I am running rivet/yoda as implemented/setup in Athena relsease
>>> asetup devval,64,rel2
>> That should by Rivet 2.1.2 and the latest YODA release, so fully up to
>> date. No problem there.
>>
>>> It would be great to get any hints, because at this stage, I am not
>>> sure, whether:
>>> - 2Dscatter histograms should not work (not sure even, why it is
>>> exported at this)
>> What is the final type of your histogram in the code? Maybe attach your
>> .yoda file so we can look at it. 1D histograms should write out as
>> HISTO1D type in the .yoda file, even if they have been scaled, but if
>> you do something like divide two histos then you get a scatter because
>> there's no longer a meaningful sum-of-weights, etc..
>>
>>> - whether yoda genuinely does not work (yet) with the rivet version 
>>> from
>>> Athena I am using (would I have to install it myself)?
>> It should do (see above). I think the problem is confusion about the two
>> types of .dat file.
>>
>>> I am also a bit confused as to whether I should implement my own
>>> histograms or take the hepdata definition (to take that one would be
>>> great, cause as far as I understand, this way one would be easily
>>> capable of doing data/MC plots).
>> If you can take what's in HepData, that's usually easiest. If you have
>> problems reading the exported format then please let us / the HepData
>> managers know.
>>
>>> Any pointer, also to a tutorial younger than November 2013 at CERN 
>>> would
>>> be useful. I would really be very very grateful!
>> I am going to write a manual for YODA as soon as possible -- many other
>> things needing to be done first, unfortunately! The Nov 2011 tutorial
>> was the most recent one that I've given on the whole Rivet system...
>> feels like it's about time to give another one. But I don't think it is
>> very out of date...
>>
>>> And then a link maybe useful to you:
>>> http://www.everydayhowto.net/recreation-and-sports/origami-yoda-how-to-make-a-jedi-master-from-paper/ 
>>>
>>>
>>> (production of merchandising)
>> Ha, awesome! Now I'm even more pleased at our violation of Lucas'
>> copyright ;-) That will be a great graphic to use if we ever add
>> "folding" machinery to Rivet...
>>
>> Let us know if rivet-cmphistos or rivet-mkhtml help. And if you have any
>> other requirements, queries, etc. -- YODA is a side-project, but I hope
>> we can make it a really nice tool and provide some half-decent 
>> support :-)
>>
>> Andy
>>
>

-- 
****************************************************
Dr. Kristin Lohwasser

DESY
Zeuthen
Raum: 3L-022
Telefon +49 33762 7-7647
Platanenallee 6
D-15738 Zeuthen
Deutschland

Kristin.Lohwasser at cern.ch
****************************************************