Is possible to use TH1F with two variables in the TDataFrame

ROOT
1

Hi,

I’m trying to using TH1F in TDF to do a histogram with two
variable like this example below (This work before):

// Energy Distribution
    TH1F *hist1 = new TH1F("hist1","Energy Distribution LAB",100,-20,20);
    hist1->Fill(eta,E); 

// Tcanvas hist1
    TCanvas *c2 = new TCanvas("c2","Histograms 1 ",600,400);
    c2->SetGrid(1,1);
    c1->SetLogx(0);
    c1->SetLogy(1);
    
    hist1->SetLineColor(kRed);
    hist1->GetYaxis()->SetTitle("dE/d#eta [GeV]");
    hist1->GetXaxis()->SetTitle("#eta ");
    hist1->Draw();

but in the TDF I don’t know how to do this or I don’t find one example that explain this.

What I’m trying in my macro:

auto hist1 = d.Histo1D(TH1F("hist1", "Energy distribution in rapidity", 100, -20, 20), "eta", "E");

ROOT 6.13/03 | Built for linuxx8664gcc heads/master@v6-13-02-120-g3c7fa4a, Mar 27 2018, 09:56:12

Cheers, Andre

2

Hi Andre,

if you mean a weighted histo the syntax would be

auto hist1 = d.Histo1D({"hist1", "Energy distribution in rapidity", 100, -20, 20}, "eta", "E");

as shown here. Does this work for you?

Cheers,
D

3

Hi Danilo,

What I did:

using doubles = TVec<double>;
        
        auto RapCalc = [](doubles Es, doubles pzs) {
            auto all_RapCalc  = 0.5*log((Es+pzs)/(Es-pzs));
            auto good_RapCalc = all_RapCalc[Es > 200.];
            return good_RapCalc;
        };
        
        auto hist1 = d.Define("rap", RapCalc, {"E", "pz"}).Histo1D({"hist1", "Energy distribution in rapidity", 100, -20, 20}, "rap", "E");
        auto hist2 = d1.Histo1D({"hist2", "", 100, -20, 20}, "Eta", "e");

        // Drawing
        auto c1 = new TCanvas("c1", "c1", 10, 10, 700, 500);
        c1->SetGrid(1,1);
        c1->SetLogx(0); // 0 == scale without Log, 1 == scale with Log
        c1->SetLogy(1);
        hist1->GetYaxis()->SetTitle("dE/dy [GeV]");
        hist1->GetXaxis()->SetTitle("y");
        hist1->DrawClone();
        hist2->SetLineColor(kRed);
        hist2->DrawClone("same");

Error:

terminate called after throwing an instance of 'std::runtime_error'
  what():  Cannot fill histogram with values in containers of different sizes.

Cheers, Andre

4

Hi Andre,

what are the sizes of the containers stored in column e and column rap ?

Cheers,
D

5

Hi, Danilo

TTree  Mar 26 21:22 2018 T  "Tree"
  e    "e/D"    10737 (Entries 1334)

rap is calculate using Es and Pzs that have:

TTree  Mar 24 22:09 2018 Particle  "particles produced"
  pz               "pz[nPart]/D"        94620
  E                "E[nPart]/D"         94616  (Entries 11746)

the size is very different, :sweat_smile:

Cheers, Andre

6

Hi Andre,

I am not sure what is the reason of this difference. TDataFrame can put in histograms collections and use collections for both values and weights. Unfortunately, having a values collection and a weights collection with different sizes is not a well defined situation, right? This is why TDataFrame complains.
Can you maybe fix this upfront in your code?

Cheers,
D

7

Hi Danilo,

When the energy cut is applied [Es >200],
the function return nothing (I Think):

using doubles = TVec<double>;
        
        auto RapCalc = [](doubles Es, doubles pzs) {
            auto all_RapCalc  = 0.5*log((Es+pzs)/(Es-pzs));
            auto good_RapCalc = all_RapCalc[Es > 200.];
            return good_RapCalc;
        };

and this can be a problem, i.e., RapCalc ('don’t receive nothing`)

auto dd2 = d.Define("rap", RapCalc, {"E", "pz"});
auto histoy1 = dd2.Histo1D({"histoy1", "Energy distribution", 100, -20, 20}, "rap", "E");

Cheers, Andre

8

Hi Andre,

but if you apply a cut, isn’t it normal that the values in all_RapCalc will be at most the ones before the cut?

Cheers,
D

9

Hi Andre,

just thought again about this. Isn’t the right thing to do also to filter E with E>200 to then make your plot? For example:

auto dd2 = d.Define("rap", RapCalc, {"E", "pz"}).Define("good_E", "E[E>200]");
auto histoy1 = dd2.Histo1D({"histoy1", "Energy distribution", 100, -20, 20}, "rap", "good_E");

This will ensure that the size is the same and you would actually use the right weight for your plot.

Cheers,
D

10

Hi, Danilo,

Thanks, It worked, very well. I will continue to study TDF, this
interface for analyses of data is very nice.

Best, Andre