[Halld-pid-upgrade] Data Challenge Skims

[Justin Stevens]

Hi Paul,

While testing other values for the tracking FOM cut I realized the efficiency I gave before of ~90% for the cut at 0.1% was using the subset of events which already satisfied the kinematic fit CL cut in my analysis, so that overestimated the efficiency.  

The real efficiency of the Tracking FOM > 0.1% cut on the eta'(2300) signal events is more like ~70%, and even moving the cut to 0.001% the efficiency only goes up to ~77%.  The increase in the bggen sample size scales roughly the same with reducing the cut value, so unless most analyzers are planning to use a similar cut on tracking FOM it may not be a net win to apply the tracking FOM cut in the skim.

-Justin

On Feb 12, 2013, at 2:08 PM, Paul Mattione wrote:

> What I listed is per track; I can change it to combined though.  Can you test the effect of other cut values on the signal yield to see where the ideal cutoff would be?  It would be great if we could increase the MC yield to 95% or maybe even 99% without increasing the background much.  
> 
> - Paul
> 
> On Feb 12, 2013, at 1:58 PM, Justin Stevens wrote:
> 
>> Hi Paul,
>> 
>> The 2+, 2- option would be useful for me, and reduces the data volume significantly.  Is the tracking FOM listed on your webpage combined for all particles, or cut on a track-by-track basis?  I checked the eta'(2300) signal MC and this has a ~90% efficiency for a combined FOM, and ~85% efficiency when applied for each track, so it may be worth it to reduce the data volume by almost a factor of 2.
>> 
>> -Justin
>> 
>> On Feb 11, 2013, at 3:53 PM, Paul Mattione wrote:
>> 
>>> I would like to get input from members of the physics and pid-upgrade groups on what kind of community-wide skims we want to have for the data challenge data.  At the physics meeting today I presented some off-the-top-of-my-head ideas for these skims, and how large they would be:
>>> 
>>> https://halldweb1.jlab.org/wiki/index.php/Mattione_Update_02112013
>>> 
>>> Some of these are obviously untenable (e.g. 2+ 1- is too large).  Others need to be tweaked (e.g. the cut on vertex-z may cut low-theta tracks that don't have enough resolution).  
>>> 
>>> I'm hoping this email starts a discussion on what data sets would be interesting to have.  Please let me know what you guys think.  
>>> 
>>> - Paul
>>> 
>>> 
>>> _______________________________________________
>>> Halld-pid-upgrade mailing list
>>> Halld-pid-upgrade at jlab.org
>>> https://mailman.jlab.org/mailman/listinfo/halld-pid-upgrade
>> 
>