Transverse Shower Leakage in Run 9 W analysis

Simulation Used:

25K single thrown electrons flat in eta=[-1,1] and phi=[0,2pi] at fixed energy = 40 GeV.  The full geometry was used in this simulation 

Figure 1:  2x2 Cluster Energy / Thrown Energy

From the figure above it appears that a small amount (~2%) of the electron energy is not measured in the reconstructed 2x2 cluster energy.  This is similar to what Ilya saw in some QA of the W->e simulation.

Figure 2: 2x2 Cluster Energy / BTOW Summed Energy

The figure above shows that the 2x2 cluster energy contains essentially 100% of the energy deposited in the BTOW.  So the shower appears to be well contained in the 2x2 cluster we use to measure the electron energy.  However figure 1 indicates that this reconstructed 2x2 cluster energy is ~2% less than the thrown electron energy.  A previous study showed that this missing energy is not coming from longitudinal shower leakage, so where is this missing energy going?

Some possible explanations why the reconstructed energy could be systematically low:

-sampling fraction is slightly off at the ~2% level for these high energies

-some subtle effect in barrel slow simulator (photoelectron response?)

-others?