[Halld-physics] ROOT TTree Format Overhaul

[Paul Mattione]

Sorry, a clarification (see below).  

On Sep 9, 2015, at 12:10 PM, Paul Mattione <pmatt at jlab.org> wrote:

> I’ve overhauled the output ROOT TTree format for the analysis library, making it event-based (one entry per event) instead of combo-based (which it was before).  This overhaul is so that all of the information necessary for an analysis is available within the same event, which enables usage of the multithreading capabilities of PROOF.  
> 
> Quick Format Summary: 
> 1) Combo-Independent particles stored in arrays (one index per particle).  E.g. Throwns, ChargedHypos, NeutralShowers, Beam (not all: only those used in combos)
> 

I should clarify: All Throwns, ChargedHypos, and NeutralShowers are stored, whether they are used in the combos or not.  Only unused beam particles are excluded.  

> 2) Combo-Dependent particle info stored in arrays (one index per combo).  E.g.: Proton__P4_Kinfit, PiMinus2__ChiSq_Timing, etc.  These each have a variable that points you to the array-index storing the combo-independent information.  E.g. ChargedHypo__P4 (pre-kinfit), dE/dx, etc.
> 
> When you cut combos, just set the branch variable “IsComboCut” to kTRUE for that combo, before writing out a copy to a new tree.  However, we’ll still have to do some kind of tree-flattening to make things combo-based prior to input to the BDT, so more scripts are needed.  
> 
> Documentation: https://halldweb1.jlab.org/wiki/index.php/Analysis_TTreeFormat
> 
> An example TSelector on how to use the new format will be available soon-ish.  
> 
> - Paul
> 
> 
> _______________________________________________
> Halld-physics mailing list
> Halld-physics at jlab.org
> https://mailman.jlab.org/mailman/listinfo/halld-physics