Re: [TMVA-users] TTree::Fill error using TMVA MLP plus getting a .cxx file.

[Jan T. <the...@ph...>]

Hi Diego,

the problem is probably, that ROOT is trying to assign the 
training/testing trees to the input file, because this is the last 
file you opened. Just placing a

outputFile->cd();

just in front of the loop where you're filling the events should do 
the job.

The stand-alone class should be generate automatically during training 
and can be found in the weights/ directory. It will be named something 
like TMVAClassification_MLP.class.C

Cheers
Jan



On Thu, 21 Jul 2011 17:04:58 -0500
  Diego Menezes <dom...@gm...> wrote:
> Dear all,
> 
> Two questions about the TMVA mlp method. First, I've been using a 
>modified
> version of ROOT's   /root/tmva/test/MVAnalysis.C to study signal and
> background separation using TMVA  mlp with this set up:
> 
>  TFile *inputsig = new TFile("ttbar_signal.root");
>   TFile *inputback = new TFile("ttbar_back.root");
> 
> 
>      TTree *signal     = (TTree*)inputsig->Get("TopTauTree");
>      TTree *background = (TTree*)inputback->Get("TopTauTree");
> 
>    factory->AddVariable( "Et", 'F' );
>    factory->AddVariable( "cent", 'F' );
>    factory->AddVariable( "ht",   'F' );
>    factory->AddVariable( "sqrts", 'F' );
> 
> and then
> 
>  std::vector<Double_t> vars( 4 ); // vector has size of number of 
>input
> variables
>   Float_t  treevars[4];
>    signal->SetBranchAddress("Et", &(treevars[0]) );
>    signal->SetBranchAddress("cent", &(treevars[1]) );
>    signal->SetBranchAddress("ht", &(treevars[2]) );
>    signal->SetBranchAddress("sqrts", &(treevars[3]) );
>   for (Int_t i=0; i<signal->GetEntries(); i++) {
>      signal->GetEntry(i);
>      for (Int_t ivar=0; ivar<4; ivar++) vars[ivar] = treevars[ivar];
>      // add training and test events; here: first half is training, 
>second
> is testing
>      // note that the weight can also be event-wise
>      if (i < signal->GetEntries()/2) 
>factory->AddSignalTrainingEvent( vars,
> signalWeight );
>      else                            factory->AddSignalTestEvent 
>   ( vars,
> signalWeight );
>   }
> 
> Plus something similar for the background part. The error I see when 
>running
> the macro is
> 
> 
> Error in <TTree::Fill>: Failed filling 
>branch:LocalSigTrainTree.weight,
> nbytes=-1
> This error is symptomatic of a Tree created as a memory-resident 
>Tree
> Instead of doing:
>    TTree *T = new TTree(...)
>    TFile *f = new TFile(...)
> you should do:
>    TFile *f = new TFile(...)
>    TTree *T = new TTree(...)
> Error in <TTree::Fill>: Failed filling 
>branch:LocalSigTrainTree.metl,
> nbytes=-1
> Error in <TTree::Fill>: Failed filling 
>branch:LocalSigTrainTree.cent,
> nbytes=-1
> Error in <TTree::Fill>: Failed filling 
>branch:LocalSigTrainTree.sqrts,
> nbytes=-1
> Error in <TTree::Fill>: Failed filling branch:LocalSigTrainTree.ht,
> nbytes=-1
> 
> Any ideas on how to fix it? Second, my old ROOT's 
>TMultiLayerPerceptron
> macro generates as one the output a .cxx file that I use in my 
>analysis
> later on. Can TMVA mlp do the same when training the NN? If so, how?
> 
> Thanks,
> 
> Diego.

--------------

Jan Therhaag
Physikalisches Institut, Universität Bonn
Office: CERN bat. 304-1-042 Tel: +41 22 76 78052
eMail: the...@ph...