0)In first attempt, try to replicate Will's changes.  Change so that takes neighboring sectors in addition to photons in the same sector.  I get no eta peak.

--In MixMaker:

---  remove if(fabs(u_1-u_2)< 20.0 || fabs(v_1-v_2)<20.0) continue - Note: this is actually a stupid thing to do.  Just need to modify the code so it looks for 20 strip differences across the sector boundaries properly.  But first re-checking in the same sector!

---  add condition that second point has <=2 points per plane

           if(psector[point1.sector()] > 2) continue;
           if(psector[point2.sector()] > 2) continue;

 1)Was able to verify I get a peak when I look in the same sector.  Now I need to try to fix up the 20 strip difference for neighboring sectors.

First try: used position of the cluster and looked at perp distance between them...right now set at 15 (20 strips should be about 10 cm).  I get a peak in the right place but it's very broad.  Ideas:

(1)increase distance - trying 30

(2) add a ratio or zgg cut or something....

2)Increase the dR cut.... got rid of low mass stuff but didn't help really pull out a better peak.  My mean is now kind of high, too (hmmm).

3)After thinking about it for a bit, decided problem must be clusters that have a large dR but a small angular separation.  I think what the 20 strip cut in the same sector was doing was to effectively impose an angular separation.

If look at this equation:

M = E sin(theta/2) sqrt(1-Zgg*Zgg)

I can come up with some reasonable bounds for the angle.  I assumed we would want etas with E between about 15 and 25 GeV and set boundaries assuming Zgg is less than 0.7.  Then I worked on some tweaks.  The peak is really starting to shape up.  More tweaking can problem make it narrower.  I plan to freeze what I have so far and then see what it does for etas in the same sector... would be need different cuts there?  Or will the same cuts give you something reasonable for clusters in the same sector as well?

4)Somewhat last resort here...I pulled Wills code again and made it compile with some effort.  I verified that I got a peak when looking in the same sector.  It looks very similar to Wills.  Next I'm trying to implement by angle cut.  I commented out the same sector requirement and added my energy-dependent angular cut.  If that still doesn't work I will need to add some additional plots to try to figure out what is going on.