Re: distinguishing single end hits from non-shower hits

[Zisis Papandreou <zisis@uregina.ca>]

Hi Blake:

Matt had referred to a procedure for single-ended hits in his Dec 30  
email.  Indeed, the procedure would involve:
- searching for cluster(s) near the single-ended segment hit
- using the z and t of the cluster to correct the energy for  
attenuation at the single end
  -using the same z and t,  see what the energy should have been at  
the other end and whether it indeed was too low for the thresholds, as  
a cross check
- adding the corrected energy to the cluster and looking at the  
invariant mass plots, etc, to see if there is an improvement (in  
sigma, chi-squared or confidence level).

If there are two or more clusters in the vicinity, loop and pick the  
best one, etc..  The process is laborious, and this is why such  
analysis is done in a second pass through the data, with the first  
pass involving only good two-ended hits.

The most important issue, however, is not the procedure but whether it  
is necessary at this time to pursue it and Matt discussed this as  
well.  We need to do some quick estimates on how critical it is for  
this stage of your work, both from the GlueX perspective but also (and  
most importantly) from your thesis requirements and timetables.  The  
latter is an internal UofR discussion, of course.

I agree with Matt and I recommend you play with the threshold as a  
quick and dirty way to estimate the magnitude/severity of the single- 
ended hits, as a function of distance from the end.  Basically, run  
different thresholds and then bin the results in z and do your  
beautiful reconstruction/thrown plots so that we see what comes out.

Cheers, Zisis...


On 6-Jan-09, at 2:15 PM, Blake Leverington wrote:

> Hey all,
>
> Ok, so I was thinking of ways to go about adding the single-ended  
> hits back into the cluster, but how does one distinguish which  
> cluster a hit would belong to since we now have no time information  
> to determine if it belongs to a cluster or not, or which cluster if  
> there is more than one nearby. We have x,y and E, but no t or z to  
> put it with a cluster.
>
> Any thoughts?
>
> -Blake

---
Dr. Zisis Papandreou		|  email: zisis@uregina.ca
Department of Physics	|  tel. : (306) 585-5379
University of Regina		|  tel. : (306) 585-4149
Regina, SK  S4S 0A2		|  fax. : (306) 585-5659
World Wide Web:    http://www.phys.uregina.ca/sparro/zisis/