The plots show that the clustering works.

(I tried to post this on friday, but was thwarted by the drupal editor.  Hopefully this one will fare better)

At this point, I can apply two cuts through my macro:

  The minimum cluster energy is a threshold for a cluster to be considered a 'good' cluster in the plots below.  For diagnostic purposes, I still generate all clusters with E>0, but only do further processing on ones above the threshold.

  This allows only the clusters with weighted centers falling within 1.0 and 2.0 in detector eta.  Currently, this is at least as wide as the entire active surface of the calorimeter, but in the future this may be tightened so that showers that are not completely contained within the calorimeter are not included (first and last eta bin)

 

 

  At the current values, the cuts are extremely wide and only chop out the very low-energy clusters.  (Still, this has an effect on the number of successful clusters reconstructed per event, even though most MIPs are energetic enough to pass.)  When I start to consider 'physics', I will change the minimum cluster energy to 5.0GeV instead of 0.5.

 

To demonstrate that this is working, I ran 500 events each of muons and gammas from our single particle MC samples.  It is important to note that these MC samples were generated between 0.8 and 2.2 to ensure full coverage of the detector.  In the 500 event samples, then, we expect ~140 events to reconstruct no towers whatsoever, which agrees with the number of entries we see on the following plots:

 

gammas:

Unfortunately, the above plot does not show up well at this resolution -- the upepr right is meant to demonstrate that the cluster centers are falling along eta bins with some small bleed-over when the shower falls near a boundary.  The lower left shows that the difference between the detector-space eta and phi of the actual particle and the reconstructed cluster are nearly always within 1/2 a tower width in eta and in phi, which is expected for well-collimated showers.

 

The number of raw clusters in a gamma event peaks sharply at 1, but there is also a significant high-end tail.  Since the tail is drastically reduced even with a very small clsuter threshold, we can see that the additional clusters had exceedingly low energy on average.

 

 

 

 

 

muons: