[Rivet] Python, Rivet 1.20, CERN AFS, LXBATCH

[Andy Buckley]

On 25/03/10 09:18, Frank Siegert wrote:
> Hi Terrance,
> 
> Terrance Figy, Thursday 25 March 2010:
>> Hello Frank,
>> I was wondering if there is a rivet script to combine several mc runs.
>> For example, I want to generate 5 million events for top pair
>>  production by running 5 event generation jobs with 1 million events
>>  each. Each job would produce an rivet.aida file. I would like to
>>  combine the 5 rivet.aida files into one to gain statistics.
> 
> I have written such a script (or rather modified one originally from Andy) 
> which combines multiple runs. Due to the limited information stored in 
> aida files it has to assume though that all of your runs have the same 
> number of events in them. Also I am not sure whether it does a proper job 
> at calculating the combined errors in profile histograms (as opposed to 
> normal histograms).
> 
> The syntax is very simple:
> 
>   aidamerge -o out.aida in1.aida in2.aida in3.aida [...]
> 
> Riveters: Do we want to distribute this with Rivet (currently it lives in 
> Sherpa's benchmark SVN)?

I think the one it was modified from lived in Rivet's bin directory in
uninstalled (and maybe untarred) form... it may still be there! So I
think this could be added in the same form. But I wouldn't be happy
about making it an official, installed, etc. script since the error
treatment is (by necessity) a bit of a guess.

At some point we need to start some YODA work again, and decide what we
consider essential in the first instance. I think we have a pretty clear
idea of what we should be doing re. (negative) weighted events now,
thanks to that long email discussion a while back, and personally I
would like a nice little stats package for other work, too! I'll see if
I can get back on to working on this, and then beg for help in more
specific areas!

Andy

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