[Rivet] aidamerge

[Andy Buckley]

On 20/01/12 09:10, Hendrik Hoeth wrote:
> Hi Holger,
>
> Thus spake Holger Schulz (holger.schulz at physik.hu-berlin.de):
>
>> could someone please tell me where aidamerge can be found on svn?
>> I got it from somewhere but cannot find it in the trunk.
>
> svn+ssh://svn.hepforge.org/hepforge/svn/rivet/contrib/
>
>> I would like to commit a switch I have been using quite heavily recently.
>> Also, can we promote it so the same level as aida2flat such that it
>> becomes available when sourcing rivet from /afs/cern.sh/... ?
>
> I think I'll let Andy comment on this ...   ;-)

;)

No, we aren't going to do that, because unlike aida2flat it can't be 
guarantee to get the right answer: AIDA files don't store enough info to 
merge exactly, so we have to make a guess about the scaling of the 
errors -- sqrt(n) or 1/sqrt(n). For the wrong type of plot, the 
1/sqrt(n) convergence used by that script won't work -- basically, it 
doesn't work without hacking for unnormalised histograms. Also, it has 
to assume that each input file represents equal stats.

All very hacky, hence it's not an "official" script, but more a template 
for users to hack for specific purposes.

So what is the flag that you wanted to add? We'd like to keep this 
script as simple as possible, because of its role as a template, but 
cool features are still welcome as long as they're of general usefulness 
and don't make the script HUGE :)

Andy

-- 
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.