[Rivet] Fwd: [Fastjet] Bug in CDFJetClu::_insert_unique

James Monk jmonk at hep.ucl.ac.uk
Fri Aug 15 10:12:12 BST 2008


This explains why this bug suddenly turned up when we started using  
the user added information to the jets

Begin forwarded message:

> From: Gavin Salam <salam at lpthe.jussieu.fr>
> Date: 15 August 2008 10:03:26 GMT+02:00
> To: James Monk <jmonk at hep.ucl.ac.uk>
> Cc: fastjet at projects.hepforge.org
> Subject: Re: [Fastjet] Bug in CDFJetClu::_insert_unique
>
> Hi James,
>
> Yes, that was a nasty hack -- we needed to know which particle was  
> in what jet, but the jetclu code didn't provide any easy way of  
> accessing a particle index to retain that information. (And we  
> didn't want to touch
> jetclu). Of course, it had always worked for us!
>
> We'll look into a better solution...
>
> Ciao,
> Gavin
>
>
>
>
>> Date: Thu, 14 Aug 2008 21:07:53 +0200
>> From: James Monk <jmonk at hep.ucl.ac.uk>
>> Reply-To: "Email the developers (subscription not required)"
>>    <fastjet at projects.hepforge.org>
>> To: fastjet at projects.hepforge.org
>> Subject: [Fastjet] Bug in CDFJetClu::_insert_unique
>> Hi Gavin,
>>
>> We found an infinite loop bug in the CDFJetClu plugin when run on the
>> system used for the Debrecen tutorial.  It looks like there is a map
>> of the jets with the energy as the key.  Using find on a map keyed by
>> a double is not safe, and somehow this causes the while loop on line
>> 152 of CDFJetCluPlugin.cc to continue forever.  We temporarily solved
>> the problem by reducing the value of the increment on line 156 to
>> 1e-9, but the proper fix would be to use upper_bound and lower_bound
>> to safely find the entry in the map instead of find.
>>
>> cheers,
>>
>> James
>>
>> _______________________________________________
>> Fastjet mailing list
>> Fastjet at projects.hepforge.org
>> http://www.hepforge.org/lists/listinfo/fastjet
>>
>
> - Gavin Salam  
> ------------------------------------------------------------
>  http://www.lpthe.jussieu.fr/~salam/          |  tel: +33 1 44 27 74  
> 31
>  LPTHE, Univ. Paris 6 et 7, Paris 05, France  |  fax: +33 1 44 27 70  
> 88
> ------------------------------------------------- salam at lpthe.jussieu.fr 
>  -

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.hepforge.org/lists-archive/rivet/attachments/20080815/44dc4df5/attachment.htm 


More information about the Rivet mailing list