More study is done on RHRS sieved run with positron cuts. Still no luck... Tried: * full dp, small dp (+-0.5%), smaller dp (+-0.2%) * use sieve out run to select good vz region then repeat study * shift central of small dp window to +- 1% * optimize binning. An example is shown below, which is cut on |dp|<0.2%. Besides, black line on bottom right plot is cut on vz:tg_ph correlation, extracted from sieve out runs. As you can see, vz has expanded significantly around this region (due to multiple scattering and conversion?). Expected sieve holes diameter is couple pixels.