At the last spin meeting, the 10 keV thresholds defined by LOW_EM were questioned.  In order to address this concern I repeated the previous sampling fraction study using modified tunes.  

    Three tunes were carried over from the previous study: nominal, LOW_EM, and Tune Three.  Tune Three lowers the delta ray production thresholds along with the electron transport threshold in absorber materials to 10 keV (the propagation threshold in sensitive volumes is already at 10 keV in the nominal tune) and is able to reproduce the LOW_EM physics.

    Two additional tunes have been introduced for this study, both initialized to the values in Tune Three.  Tune Four increases both delta ray production and the absorberelectron transport to 100 keV, while Tune Five increases the delta ray production and electron transport thresholds to 100 keV in the absorber only.

As evident in the plot, neither of the new tunes recreates the LOW_EM physics.  In fact, both are indistinguishable from the nominal GEANT tune.

    What's going on?  As I tried to mention in the spin meeting, the important cuts in LOW_EM have nothing to do with the propagation of particles in the shower and the arguments questioning the 10 keV thresholds are not quite relevant.  The necessity for lowing the delta ray thresholds suggests that, somewhere between 10 keV and 100 keV, delta ray production dramatically increases.  Most of these delta rays are immediately reabsorbed, but near the lead/plastic boundary they would dump nontrivial energy into the plastic.  Integrated over the shower profile and each layer of the BEMC, these small depositions would lead to the small but significant effects seen in the sampling fraction.  In order to see this effect the electron transport thresholds have to be lowered to 10 keV or the delta rays wouldn't have the chance to propagate into the plastic.  Unfortunately, this also effects other electrons in the reconstruction, tracking them to unneeded precision and likely causing the increased computation time.