Re: distinguishing single end hits from non-shower hits

["Mark M. Ito" <marki@jlab.org>]

Mihajlo,

Kornicer, Mihajlo wrote:
>
> Mark is right, but dead channels will become an issue in a couple of 
> years.
You are right, for Monte Carlo events, a single-ended hit tells you that 
the hit came near the end in question. Also in Monte Carlo all hits are 
real.
> Right now reconstruction software assumes symmetric 
> upstream/downstream efficiencies, and if one later discovers 
> asymmetry, for any reason, the "middle" point could be adjusted 
> accordingly.
I'll mention the obvious that the reconstruction should not assume equal 
efficiencies in a fundamental way,  i. e., unless that is merely the 
equalization of independent parameters. The statement that a 
single-ended hit must come near the end is partially based on this 
assumption. For real data, you don't know that the efficiencies are 
equal, or that the thresholds are equalized or that the attenuation 
length is uniform in the counter or that electronic noise is not present.
> In order to make the physics case for saving shower outskirts there is 
> no reason to associate random upstream hits with downstream showers 
> and vice versa, regardless of the actual cost/benefit of salvaging 
> single-end hits.
Remember Blake's original question was in the context of whether a 
single-ended hit should be included in a cluster or not. I did not mean 
to say that we should assume that single-ended implies 
high-precision-z-measurement of z_BCAL = middle. I was thinking more 
about how to do the timing cut and the attendant correction for light 
propagation (which perhaps is not even important in this context). With 
an imperfect detector there will in fact be times (one end dead) where a 
big hit at the upstream end will be included in a downstream cluster 
when it's timing is perfect and it fits a reasonable shower profile. In 
that case the big upstream single-ender should have no influence on the 
z reconstruction at all, of course.

  -- Mark

>
>