b) e/h isolation based on Geant Record (Mike, April 3)

This is geant-track based analysis


Bottom plots are S/B ratios.  

These are extremely loose cuts and I don't think you can base the entire FGT analysis on them.
First of all, the e/p cut isn't quite right.  I use the actual energy of the track whereas most hadrons would just drop MIPs into the calorimeter making the e/p cut very helpful.  I did try to do this in my code but the e/p cut became TOO good and I didn't want to have to deal with trying to convince people of that.
Second of all, no neutral energy isolation cut is made to veto against neutral energy depositions.
Third of all, no shower shape (transverse and longitudinal) information is used at all.
I think it's fair to say that this is a worst case scenario analysis.


I try to understand better this last two S/B plots.
The message from the lower figure is:
* isolation cut alone results with background by a factor 2 to 3 for W+ and factor 4 to 10 for W- with reverse PT dependence for W+ & W-
* away said jet veto helps almost nothing in e/h discrimination.
* if only isolation & away side jet cuts are applied background dominates over signal by a factor of ~300 at PT of 20 GeV and improves to a factor ~2 at PT=40 GeV.

Taken at the face value  Enddcap information add discrimination power of ~1000 for PT=20 GeV  changing toward factor of 6  for PT of 40 GeV/c. (Assuming we want S/B=3)

My comment:
The value of 6 is in reach but value of 1000 may be not trivial.

 e/p cut  - I agree with Mike, 
Looks like the e/p cut applied to h-  reduces it up to a factor of 2  
for pT of 40 GeV.
This cut needs to be dropped for real data- we will not measure PT in  
FGT with reasonable accuracy at large track PT - this reduces overall  
e/h discrimination power for W- at  for PT>30 GeV, so more realistic  
estimate of discrimination power of this (geant based) algo for W- at  
PT of 40 GeV is rather  S/B=0.35 and we need additional factor of 8  
from the endcap.