13.6.4. Macro “modelerFlowrateNeuralNetworks.C”

13.6.4.1. Objective

The objective of this macro is to build a surrogate model (an Artificial Neural Network) from a database. From a first database created from an ASCII file flowrate_sampler_launcher_500.dat (which defines values for eight variables described in Presentation of the problem on 500 patterns), two ASCII files are created: one with the 300st patterns (_flowrate_sampler_launcher_app_.dat), the other with the 200 last patterns (_flowrate_sampler_launcher_val_.dat). The surrogate model is built with the database extracted from the first of the two files, the second allowing to perform calculations with a function.

13.6.4.2. Macro Uranie

    // Create a TDataServer
    TDataServer * tdsu = new TDataServer("tdsu","tds u");
    TDataServer * tds = new TDataServer("tdsFlowrate", "tds for flowrate");
    TDataServer * tdsv = new TDataServer();

    // Load a database in an ASCII file
    tdsu->fileDataRead("flowrate_sampler_launcher_500.dat");

    //Create database
    tdsu->exportData("_flowrate_sampler_launcher_app_.dat", "","tdsFlowrate__n__iter__<=300");
    tdsu->exportData("_flowrate_sampler_launcher_val_.dat", "","tdsFlowrate__n__iter__>300");

    tds->fileDataRead("_flowrate_sampler_launcher_app_.dat");

    TCanvas *c = new TCanvas("c1", "Graph for the Macro modeler",5,64,1270,667);
    // Buils a surrogate model (Artificial Neural Networks) from the DataBase
    TANNModeler* tann = new TANNModeler(tds, "rw:r:tu:tl:hu:hl:l:kw,3,yhat");
    tann->setFcnTol(1e-5);
    //tann->setLog();
    tann->train(3, 2, "test");

    tann->exportFunction("pmmlc++","uranie_ann_flowrate","ANNflowrate");

    gROOT->LoadMacro("uranie_ann_flowrate.C");

    tdsv->fileDataRead("_flowrate_sampler_launcher_val_.dat");

    cout << tdsv->getNPatterns() << endl;

    // evaluate the surrogate model on the database
    TLauncherFunction * tlf = new TLauncherFunction(tdsv, "ANNflowrate", "rw:r:tu:tl:hu:hl:l:kw", "yann");
    tlf->run();

    tdsv->startViewer();

    tdsv->Draw("yann:yhat");
    //  tdsv->draw("yhat");

The main TDataServer loads the main ASCII data file flowrate_sampler_launcher_500.dat

    TDataServer * tdsu = new TDataServer("tdsu","tds u");
    tdsu->fileDataRead("flowrate_sampler_launcher_500.dat");

The database is split in two parts by exporting the 300st patterns in a file and the remaining 200 in another one:

    tdsu->exportData("_flowrate_sampler_launcher_app_.dat", "","tdsFlowrate__n__iter__<=300");
    tdsu->exportData("_flowrate_sampler_launcher_val_.dat", "","tdsFlowrate__n__iter__>300");

A second TDataServer loads _flowrate_sampler_launcher_app_.dat and builds the surrogate model over all the variables:

    TDataServer * tds = new TDataServer("tdsFlowrate", "tds for flowrate");
    tds->fileDataRead("_flowrate_sampler_launcher_app_.dat");
    TANNModeler* tann = new TANNModeler(tds, "rw:r:tu:tl:hu:hl:l:kw,3,yhat");
    tann->setFcnTol(1e-5);
    tann->train(3, 2, "test");

The model is exported in an external file in C++ language "uranie_ann_flowrate.C" where the function name is ANNflowrate:

tann->exportFunction("c++", "uranie_ann_flowrate.C","ANNflowrate");

The model is loaded from the macro "uranie_ann_flowrate.C" and applied on the second database with the function ANNflowrate:

    TDataServer * tdsv = new TDataServer();
    gROOT->LoadMacro("uranie_ann_flowrate.C");
    tdsv->fileDataRead("_flowrate_sampler_launcher_val_.dat");
    TLauncherFunction * tlf = new TLauncherFunction(tdsv, "ANNflowrate", "rw:r:tu:tl:hu:hl:l:kw", "yann");
    tlf->run();

13.6.4.3. Graph

../../_images/modelerFlowrateNeuralNetworks.png — Figure 13.40 Graph of the macro **“modelerFlowrateNeuralNetworks.C”**

13.6.4.4. Console

--- Uranie v4.11/0 --- Developed with ROOT (6.36.06)
                      Copyright (C) 2013-2026 CEA/DES 
                      Contact: support-uranie@cea.fr 
                      Date: Thu Feb 12, 2026

 <URANIE::WARNING> 
 <URANIE::WARNING> *** URANIE WARNING ***
 <URANIE::WARNING> *** File[${SOURCEDIR}/dataSERVER/souRCE/TDataServer.cxx] Line[760]
 <URANIE::WARNING> TDataServer::fileDataRead: Expected iterator tdsu__n__iter__ not found but tdsFlowrate__n__iter__ looks like an URANIE iterator => Will be used as so.
 <URANIE::WARNING> *** END of URANIE WARNING ***
 <URANIE::WARNING> 
 ** TANNModeler::train niter[3] ninit[2]
 ** init the ANN
 ** Input  (1) Name[rw] Min[0.101018] Max[0.0295279]
 ** Input  (2) Name[r] Min[25668.6] Max[14838.3]
 ** Input  (3) Name[tu] Min[89914.4] Max[14909]
 ** Input  (4) Name[tl] Min[89.0477] Max[15.0121]
 ** Input  (5) Name[hu] Min[1048.92] Max[34.8615]
 ** Input  (6) Name[hl] Min[763.058] Max[33.615]
 ** Input  (7) Name[l] Min[1401.09] Max[163.611]
 ** Input  (8) Name[kw] Min[10950.2] Max[635.503]
 ** Output  (1) Name[yhat] Min[78.0931] Max[44.881]
 ** Tolerance  (1e-05)
 ** sHidden    (3) (3)
 ** Nb Weights (31)
 ** ActivationFunction[LOGISTIC]
 **

 ** iter[1/3] : *= : mse_min[0.00150425]
 ** iter[2/3] : ** : mse_min[0.0024702]
 ** iter[3/3] : *= : mse_min[0.00122167]

 ** solutions : 3
 ** isol[1] iter[0] learn[0.00126206] test[0.00150425] *
 ** isol[2] iter[1] learn[0.00152949] test[0.0024702]
 ** isol[3] iter[2] learn[0.00107926] test[0.00122167] *
 ** CPU training finished. Total elapsed time: 1.84 sec

*******************************
*** TModeler::exportFunction lang[pmmlc++] file[uranie_ann_flowrate] name[ANNflowrate] soption[]

*******************************

*******************************
*** exportFunction lang[pmmlc++] file[uranie_ann_flowrate] name[ANNflowrate]

*******************************
PMML Constructor: uranie_ann_flowrate.pmml
*** End Of exportModelPMML 
*******************************
*** End Of exportFunction
*******************************