[Rivet] query on rivet-analysis

[Andy Buckley]

Hi Mukesh,

Rivet doesn't have any *built-in* support for ntupling -- it's primarily 
designed as an events-to-plots system for MC validation, tuning and 
prototyping.

The file format is not the only reason that the files would be much 
bigger -- although YODA's text format is not primarily designed for 
space efficiency. In the mode that you want, the output files grow with 
the number of input events, while histograms have a fixed size 
determined by the numbers of bins. You can of course already make cut 
flow histograms etc. within the event loop, but you won't have access to 
event-by-event correlations and other things that are useful for cut 
optimisation.

But if you want you can create a ROOT TTree and TFile in the init() 
method and then fill entries into the tree with each event: that is 
pretty easy to do, and will result in an output ROOT file like you 
want... but there aren't any built-in tools to help you do it.

Andy


On 29/11/15 12:43, Mukesh Kumar wrote:
> Hi Andy,
>
> Thanks  for the answer, and in fact you answered by saying "ntuple" is
> much larger file but is there any possibility
> to save that in lighter version may be in root files so that one could
> do analysis for rates, cut-flows etc?
>
> *Mukesh Kumar
> **School of Physics & MITP
> University of the Witwatersrand
> Private Bag 3
> Wits 2050
> South Africa
> Mob. No. +27 829 624 780
>
> *
> *emails: mukesh.kumar at cern.ch <mailto:mukesh.kumar at cern.ch>
> *
> *mukesh.kumar at wits.ac.za <mailto:mukesh.kumar at wits.ac.za>
> *
>
> On Thu, Nov 26, 2015 at 10:11 AM, Andy Buckley <andy.buckley at cern.ch
> <mailto:andy.buckley at cern.ch>> wrote:
>
>     On 25/11/15 18:34, Mukesh Kumar wrote:
>
>         Hi Andy,
>
>         I have a simple query on rivet-analysis:
>         After piping with HepMC files we get .yoda files following the
>         analysis
>         routine where we filled
>         all the histograms and as a result get the needed plots.
>         But is it possible somehow to use those .yoda files for the
>         purpose of
>         statistical analysis,
>         I mean further if I just want to look for significance after
>         some cut
>         choices etc. In this way
>         always we don't need to run the codes again and again which
>         really takes
>         longer.
>
>         Hope you understand what I meant to ask.
>
>
>     Hi Mukesh,
>
>     I'm not sure I do fully understand -- the .yoda file contains
>     histograms, i.e. the "per event" observables and their correlations
>     have been lost. Otherwise it would be an ntuple file and would be
>     *much* larger.
>
>     You can of course make transformations on the histograms, i.e.
>     rescale, divide, integrate, ... but I think that's not what you're
>     asking for?
>
>     Andy
>
>     --
>     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