Jetnet  checks (19/11/18)

 Fig 1 jetnet shows results from a 2 10 1 neural net.
 Plot 5 shows the input signal (H0)  and background (H1) - top row.
 And on the  second row the selected signal (red) and the signal (red) + background).
 The performance is similar to that found using TMVA.

 Fig 2 jetnet shows results from a 3 10 10 1 neural net.
 The third input variable (0 or 1)  results in perfect separation of the two distributions
 ( the additional hidden layer is unimportant in this respect). 

 Fig 3 jetnet shows results from a 2 10 10 1 neural net 
 with a test sample of three Gaussian distributions trained to give 0, 0.5 and 1, respectively
 at the output  node.  Plot 5 shows that the  NN results in a clean separation of the distributions.