There still seem to be some issue when a pz selection is performed that appears to create a dip in the px, py distributions at low transverse momentum.

The plots diag_tkupx_comp.pdf and diag_tkupy_comp.pdf show that the dip only appears in the data and reconstructed MC. The truth underwent the same selection but does not exhibit the feature (all of the 2016-04_1.2 3mm data and simulation in those plots).

The plots diag_tkupx_data.pdf and diag_tkupy_data.pdf show that the dip only seem to appear when the pz = [135,145] cut is applied (all of the 2016-04_1.2 3mm data in those plots).

I loaded 08681 and recorded all the spillid/eventid of the TKU tracks that did not make cut and saved them in attachment:08681_pz_cut_failed.txt .