[Halld-pid] [EXTERNAL] PID systematic uncertainties

[Sean Dobbs]

Hi all,

To follow up on the discussion about PID systematics at today's
meeting, the current ideas and information are summarized in this wiki
page:
https://halldweb.jlab.org/wiki/index.php/PID_study_proposal

You can see the suggestion for how to evaluate the uncertainty do to
data / MC modeling for these selections under "Systematic studies",
along with an example in pi0 -> gg by Justin and other studies by
Mahmoud and Justin.

In a general sense, I think that the procedures and recommendations
need input and guidance from the regular attendees of the PID working
groups, but the actual analyses are best done by the people analyzing
the particular final states.  If we do see differences in efficiency
determined between data and MC, then there are several possible
options, including:  (1) widen the cuts used until data and MC agree;
(2) change the smearing in MC to between match the data; (3) just
quote some sufficiently large uncertainty to cover the difference.  I
think at this point it would be best to have some consistent
recommendations that we can give to analyzers.


So in terms of moving forward, I think that the plan we discussed of:
1. Reviewing this information
2. Discussing it via email before the next meeting
3. Getting more analyzers to contribute to these studies
4. Discussing the current set of results and the plan forward at the
next meeting
sounds fairly reasonable.


One final bit of information is a reminder of the charges to be
addressed by the October Collaboration Meeting (hopefully):

* Determine a new set of common loose PID criteria to be used when
creating analysis skims
* Determine if stricter event selections are required for any of the
current cross section analyses
* Evaluate the efficiency of these criteria in data and simulation for
the 3 GlueX-I run periods
* Provide a prescription for determining a systematic uncertainty due
to data/MC modeling of the efficiency of these criteria

Cheers,
Sean