I want to do a 2D fit. How do i do it?

First of all, I’m not good at coding.
I want to do a 2D fit like below.
But I’ve only tried 1D fit, so I don’t know what to do with 2D fit.

As you can see in the bottom left of the picture, I want to do a 2D fit.

The pictures below are the 1D fitting and 2D histograms I did, something is off.

Finally, below is the code I was modifying.
I want to use KDE on the x-axis and argus+crystalball on the y-axis, in 2D fit.
That is, it should be (argus+crystalball)*KDE.
Can you tell me which part of my code I should fix?
Thank you for reading.

#include "RooRealVar.h"
#include "RooDataSet.h"
#include "RooGaussian.h"
#include "TCanvas.h"
#include "RooPlot.h"
#include "TAxis.h"
using namespace RooFit;

void gen_mbc()

 TChain chain1("B_tree");

   auto run_N = std::to_string(2);
   auto file_2 = "/home/belle2/hyuna/MVA_ntuple/MVA_cc.root";
   auto file_3 = "/home/belle2/hyuna/MVA_ntuple/MVA_uu.root";
   auto file_4 = "/home/belle2/hyuna/MVA_ntuple/MVA_dd.root";
   auto file_5 = "/home/belle2/hyuna/MVA_ntuple/MVA_ss.root";
   auto file_6 = "/home/belle2/hyuna/MVA_ntuple/MVA_charged.root";
   auto file_7 = "/home/belle2/hyuna/MVA_ntuple/MVA_mixed.root";


   ROOT::RDataFrame df1(chain1);

   TCut mvaCut = "outMVA < 0.17";

   auto data_2D = new TH2F("y Axion Mass x Mbc",200,5.2,5.29,200,1.84,2.085);
   chain1.Draw("B_axion_M:Mbc >> data_2D",mvaCut);
   TH2F * hh = (TH2F *) gDirectory->Get("data_2D");

//   auto df_m = df1.Filter(mvaCut.GetTitle());

//   auto h = df_m.Histo1D({"Mbc", "Mbc", 200u, 5.2, 5.29}, "Mbc");
//   auto h2 = df_m.Histo1D({"B_axion_M", "B_axion_M", 200u, 1.84, 2.085}, "B_axion_M");

   RooRealVar x("Mass", "Mass(GeV)", 5.2, 5.29);
   RooRealVar y("axion_Mass", "Axion mass(GeV)", 1.84, 2.085);

   auto rd2h = RooDataHist("rd2h","rd2h",RooArgList(x,y),hh);
//   auto rdh = RooDataHist("rdh","rdh",RooArgList(x),h.GetPtr(),1.0);
//   auto rdh_2 = RooDataHist("rdh_2","rdh_2",RooArgList(y),);

//   RooHistPdf m("Mbc", "", x, rdh, 0);
//   RooHistPdf am("axion_Mass", "", y, rdh2, 0);

// Crystal ball PDF 
   RooRealVar m1("m1", "mean1 of crystalball", 5.27942, 5.27942, 5.27942);
   RooRealVar s1("s1", "width1 of crystalball", 0.00315, 0.00315, 0.00315);
   RooRealVar a1("a1", "alpha1 of crystalball", 1.32, 1.32, 1.32);
   RooRealVar n1("n1", "n1 of crystalball", 10, 10, 10);

// Argus PDF
   RooRealVar c("c", "slope parameter of argus", -1., -100., 100.);

   RooCrystalBall crystal("crystal", "crystalball - signal", x, RooConst(5.279), s1, a1, n1);
   RooArgusBG argus("argus", "argus - bkg", x, RooConst(5.29), c);

   RooRealVar Nsig("Nsig", "yield of signal", 0.0, 0.0,10000000);
   RooRealVar Nbkg("Nbkg", "yield of background", 0.0, 0.0, 10000000);

   RooAddPdf model("model", "", RooArgList(crystal, argus), RooArgList(Nsig,Nbkg));


 // Fit pdf to data
   RooFitResult* fitres = model.fitTo(rdh, RooFit::SumW2Error(true), Save());

        RooArgSet fitargs = fitres->floatParsFinal();
        TIterator* iter(fitargs.createIterator());

        for (TObject* Tobj = iter->Next(); Tobj != 0; Tobj = iter->Next()) {
            RooRealVar* rrv = dynamic_cast<RooRealVar*>(Tobj);
            std::string name = rrv->GetName();
            double val = rrv->getVal();
            double err = rrv->getError();

            printf("%s: %lf +- %lf\n", name.c_str(), val, err);

// Print values of mean and sigma (that now reflect fitted values and errors)
   auto C = new TCanvas();
   C -> cd();
   h -> Draw();
   C -> Update();

   RooPlot* xframe = x.frame(Title("BtoK+gg Mbc"));

   model.plotOn(xframe, Components(crystal), LineColor(kGreen), LineStyle(kDotted));
   model.plotOn(xframe, Components(argus), LineColor(kRed), LineStyle(kDashed));

//   model.paramOn(xframe, Label(""), Format("NEU", AutoPrecision(1)), Layout(0.1, 0.45, 0.9), Parameters(RooArgSet(Nsig,Nbkg,m1,s1,a1,n1,c)));


Hi @hyuna!

You are not missing much in your script. You just need to implement your pdf for the other variable (a RooKeysPdf if you want to implement a KDE), and then make a 2D pdf using a RooProdPdf.

In this tutorial notebook, I also have an example on a 2D fit in the section called “multivariate fit”: https://github.com/root-project/training/blob/master/HASCO/2023/exercises/notebooks/roofit-tutorial-01.ipynb

Maybe it helps you! And let me know if you have further questions.


This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.