In response to RT ticket 2561, I wrote a simple Monte Carlo to test the effect of the TPC geometry on efficiency vs. p_T. It appears to me that I can qualitatively (almost quantitatively) match the embedding results that have a dip in efficiency around 1.0-1.5 GeV/c. One might be able to fiddle with the parameters of the Monte Carlo to make it match even better.

Assumptions:

• TPC active volume between radius in [60,200] cm, but only 20% acceptance below 123 cm to simulate spaced inner padrows
• TPC active volume gap between sectors is π/240 (i.e. each sector loses 1/80th of its π/6 width on each side)
• Hit reconstruction efficiency is 90%
• Track findind efficiency on top of that is 90%
• A minimum required fraction F of the possible hits must be found (this replicates cuts on the Nhits/Npossible)

I only simulated 1/4 of the full azimuthal acceptance, i.e. [0,π/2] instead of [0,2π], but this should suffice.

Running my Monte Carlo with different values of F give the following results:

F=0.5
F=0.6
F=0.7
F=0.75
F=0.8

I repeated this with everything the same except doubling the size of the gap between sectors to π/120. This makes it easier to lose tracks in the gap even with curvature, so the dip in efficiency moves to lower p_T:

F=0.5
F=0.6
F=0.7
F=0.75
F=0.8

The macro used to run the Monte Carlo and generate these plots is attached. Run it compiled as:   root -l -b 'rt2561.C+(0.7)'

-Gene