[Rivet] Rivet question

[Andy Buckley]

Good! Phew, we don't need to re-engineer everything ;-)

Agreed that this is a bit of a special case, and to be honest I can't 
imagine an easier way, since adding a mechanism for all analyses to 
return arbitrary objects would *definitely* force some type-casting work 
on the user.

Let us know if you need any more suggestions, or if bits of the Rivet 
machinery are causing you a lot of trouble. Feedback is always good!

Cheers,
Andy



On 28/02/16 23:50, James John Lacey wrote:
> Hi Andy,
>
> I implemented a procedure similar to your suggestion and everything is working perfectly!
>
> This is probably the best solution for this special use case, as I don't imagine this type of functionality is required by the large majority of users.
>
> Thanks again!
>
> Cheers,
> Jim
> ________________________________________
> From: James John Lacey
> Sent: 28 February 2016 17:41
> To: Andy Buckley
> Cc: Rivet
> Subject: RE: Rivet question
>
> Hi Andy,
>
> Thanks for the quick response!
>
> I'm just on my way out for the afternoon,.. I'll try this approach out later today and let you know.
>
> Thanks and cheers,
> Jim
>
> ________________________________________
> From: Andy Buckley
> Sent: 28 February 2016 17:32
> To: James John Lacey
> Cc: Rivet
> Subject: Re: Rivet question
>
> Hi Jim,
>
> Hmm, good question.
>
> It's not something that we designed Rivet to do, since the only
> programmatic output after analysis is meant to be the centrally-managed
> histogram objects.
>
> I suppose we could add some general way of passing arbitrary objects
> about, but it's not obvious how it would help most users. I think it
> would be as easy to do a dynamic_cast to the exact analysis type and use
> a method specific to that analysis to do whatever you need. Something
> like this (code not at all tested) would work, I think... no
> dynamic_cast needed:
>
> AnalysisHandler ah;
> MY_SUPER_ANALYSIS myana;
> ah.addAnalysis(&myana);
> for (int i = 0; i< NEVENTS; ++i) {
>     HepMC::GenEvent* evt = get_next_event_from_somewhere();
>     ah.analyze(*evt);
>     const FourMomentum foo = myana.my_special_feature();
>     ...
> }
>
> Does that help?
>
> Andy
>
>
> On 28/02/16 15:53, James John Lacey wrote:
>> Hi Andy,
>>
>> I am developing a Rivet routine for use in standalone code using the C++
>> API. The routine is being developed in the context of the simplified
>> Higgs cross sections Truth Classifier that has been discussed...
>>
>> I have everything up and running and can implement a Rivet analysis in
>> my C++ code using the AnalysisHandler::analyze(*HepMC::GenEvent) method.
>>
>> Now given that I am using Rivet projections to compute the objects and
>> quantities used in the classification I would like to have Rivet return
>> these values so I can use/save them.
>>
>> My question is: is there a way to retrieve per event information (ex. a
>> 4-vector) from Rivet when running in standalone C++ ?,.. ie. for a given
>> HepMC event I want to analyze() the event in Rivet and retrieve the
>> relevant event information in the form of a double or 4-vector or ... etc.
>>
>> Thanks in advance for any insight/help you can provide!
>>
>> Cheers,
>> Jim
>>
>
>
> --
> Dr Andy Buckley, Lecturer / Royal Society University Research Fellow
> Particle Physics Expt Group, University of Glasgow
>


-- 
Dr Andy Buckley, Lecturer / Royal Society University Research Fellow
Particle Physics Expt Group, University of Glasgow