The particle identification is done by the deconvolution methods developed here at Davis.  The Winter workshop talk by Jenn Klay explains much of this process.

The dE/dx vs. rigidity plots are first parmeterized by the Bethe-Bloch equation in the regions where the bands are clearly separated. Combining the information of more than one particle allows one to determine the centroid of the dE/dx bands in locations where the bands are not separated.

Next the particles are selected from a cut of theta=90+-15 in the center of mass.  This is done before particle identification, so the particle mass has to be assumed to boost to the center of mass  and make this cut.  To do this, the particle-of-interest mass is used to boost all of the particles to the center of mass, and the cut is made. The particle of interest is selected from the bands , and the other particles are ignored.  This is done for all of the particles of interest.  The particles that are obtainable with this method are the protons, negative pions, and the helions.

The positive pions cross the protons and are removed in a more complicated way.  The ratio of the pi+ and pi- are sutdied where the bands are separated.  From this ratio the pi+/pi- ratio is extrapolated into the region where the pions and protons cross.  Using this ratio and the yield of pi-, the pi+ are predicted and subtracted from the protons.