The goal of this excercise is to use the weights files generated in the BDT training process to select events for amplitude analysis.  The weights files written in the previous training step are used in the TMVA::Reader class to compute the BDT classifier value for each combo, which can then be cut on to give the desired purity/efficiency.

Steps:

1) First setup the working directory by creating soft links to the relevant data and the weights files from the previous training:

ln -s $GLUEX_DATA/session6b/ data
ln -s ../../session6b/weights/ weights	

2) Use a TSelector to process the analysis TTree and output a smaller analysis TTree containing only events which satisfy the user defined BDT cut with the following macro (not the different options passed to the TSelector from the previous exercise in session6b):

root -l postTrain_n3pi.C

3) Goal is to plot signal and background 3 pi invariant mass distributions using a BDT cut and some other "simple" cut combination 

   a) Add histograms to selector_tree_n3pi for 3pi mass after BDT cut and "simple" cuts on KinFit CL, etc (in EvalTMVA()). Then write them to the output TFile.


Additional exercises:
* From the histograms described above compute mass dependent purity: S/(S+B)
* Add histograms filled before cuts to allow you to compute mass dependent efficiency