How to verify the content of a branch within a Tree?

[Yassine-taw]

I’m using T Multi Layer Perceptron which is asking for branches of a tree as input arguments (input neurons).
My network is not training properly, this may be because of the way I fill my tree (and hence my branches).
Here are all the tree related commands that I use in my code:

//Create tree called ZernikeTree
TTree *ZernikeTree = new TTree(“simulation”, “simulated data”);

//Declare variables that will be used as branches
float ZRX00, ZRX11, ZRX20;

//Create branches
ZernikeTree->Branch(“ZRX00”, &ZRX00, “ZRX00/F”);
ZernikeTree->Branch(“ZRX11”, &ZRX11, “ZRX11/F”);
ZernikeTree->Branch(“ZRX20”, &ZRX20, “ZRX20/F”);

//Fill in Tree from two different variables declared before (ZTR and ZTE)
//(Those variables have already been verified, their content is correct)

for (int i = 0; i < 625; ++i) {

  ZRX00 = ZTR[ i ][ 25 ];
  ZernikeTree->Fill();
  ZRX00 = ZTE[ i ][ 25 ]; 
  ZernikeTree->Fill();

}

//Same syntax for the two other branches.

Is there anything wrong with this syntax? If not, how could I verify the content of the branch to understand where the problem is coming from?

Thank you!

[eguiraud]

You probably want

for (int i = 0; i < 625; ++i) {
    ZRX00 = ZTR[ i ][ 25 ];
    ZRX00 = ZTE[ i ][ 25 ]; 
    ZernikeTree->Fill();
}

First you set the values, then you fill an entry.

See this answer for various methods that can be used to read TTree values.

As a side note, I am not a TMVA expert but I think TMultiLayerPerceptron is deprecated. See the TMVA tutorials here (or in your ROOT installation folder $ROOTSYS/tutorials/tmva) for examples of current supervised learning algorithms.