Hello RooFit experts,
When I try splitting the dataset, the errors seem not to be passed to subsets correctly.
My code is here:
void repro()
{
using namespace RooFit;
std::cout << "ROOT version: " << gROOT->GetVersion() << std::endl;
// Add data1
RooRealVar x1{"x1", "x1", 0, 10};
RooCategory cat1{"cat1", "cat1", {{"sample_1", 0}}};
RooRealVar weight1{"weight1", "weight1", 1.0};
RooDataSet data1{"data1", "data1", {x1, cat1, weight1}, WeightVar(weight1), StoreError(weight1)};
data1.add({x1, cat1}, 2.0, 0.3);
// Add data2
RooRealVar x2{"x2", "x2", 0, 10};
RooCategory cat2{"cat2", "cat2", {{"sample_2", 1}}};
RooRealVar weight2{"weight2", "weight2", 1.0};
RooDataSet data2{"data2", "data2", {x2, cat2, weight2}, WeightVar(weight2), StoreError(weight2)};
data2.add({x2, cat2}, 3.0, 0.4);
// combine data1 and data2 into combData
RooRealVar weightComb{"weightComb", "weightComb", 1.0};
RooCategory catComb{"catComb", "catComb"};
catComb.defineType("sample_1");
catComb.defineType("sample_2");
RooArgSet args;
args.add(x1);
args.add(x2);
args.add(catComb);
args.add(weightComb);
args.add(weight1);
args.add(weight2);
RooDataSet combData{"combData", "Combined data", args, Index(catComb), Import({{"sample_1", &data1}, {"sample_2", &data2}}), WeightVar(weightComb), StoreError({weight1,weight2,weightComb})};
// split combData into data1Comb and data2Comb
std::unique_ptr<TList> dataList{combData.split(catComb, true)};
auto& data1Comb = static_cast<RooDataSet&>(*dataList->At(0));
auto& data2Comb = static_cast<RooDataSet&>(*dataList->At(1));
// Check original data1 and data2 weight errors
data1.get(0);
data1Comb.get(0);
std::cout << "data1 weightError: " << data1.weightError() << ", data1Comb weightError: " << data1Comb.weightError() << std::endl;
data2.get(0);
data2Comb.get(0);
std::cout << "data2 weightError: " << data2.weightError() << ", data2Comb weightError: " << data2Comb.weightError() << std::endl;
// debug:print all datasets
std::cout << "\nCombined data:" << std::endl;
combData.Print("v");
combData.weightVar()->Print();
std::cout << "\ndata1:" << std::endl;
data1.Print("v");
data1.weightVar()->Print();
std::cout << "\ndata2:" << std::endl;
data2.Print("v");
data1.weightVar()->Print();
std::cout << "\ndata1Comb:" << std::endl;
data1Comb.Print("v");
data1Comb.weightVar()->Print();
std::cout << "\ndata2Comb:" << std::endl;
data2Comb.Print("v");
data2Comb.weightVar()->Print();
}
And the output is:
------------------------------------------------------------------
| Welcome to ROOT 6.36.02 https://root.cern |
| (c) 1995-2025, The ROOT Team; conception: R. Brun, F. Rademakers |
| Built for linuxx8664gcc on Jul 10 2025, 20:02:19 |
| From tags/6-36-02@6-36-02 |
| With g++ (GCC) 13.1.0 |
| Try '.help'/'.?', '.demo', '.license', '.credits', '.quit'/'.q' |
------------------------------------------------------------------
Processing repro.cpp...
ROOT version: 6.36.02
data1 weightError: 0.3, data1Comb weightError: 0
data2 weightError: 0.4, data2Comb weightError: 0
Combined data:
DataStore combData (data2)
Contains 2 entries
Observables:
1) x1 = 5 L(0 - 10) "x1"
2) x2 = 5 L(0 - 10) "x2"
3) catComb = sample_2(idx = 1)
"catComb"
4) weight1 = 2 +/- 0.3 C L(-INF - +INF) "weight1"
5) weight2 = 3 +/- 0.4 C L(-INF - +INF) "weight2"
Dataset variable "weightComb" is interpreted as the event weight
RooRealVar::weightComb = 3 C L(-INF - +INF)
data1:
DataStore data1 (data1)
Contains 1 entries
Observables:
1) x1 = 5 L(0 - 10) "x1"
2) cat1 = sample_1(idx = 0)
"cat1"
Dataset variable "weight1" is interpreted as the event weight
RooRealVar::weight1 = 2 +/- 0.3 C L(-INF - +INF)
data2:
DataStore data2 (data2)
Contains 1 entries
Observables:
1) x2 = 5 L(0 - 10) "x2"
2) cat2 = sample_2(idx = 1)
"cat2"
Dataset variable "weight2" is interpreted as the event weight
RooRealVar::weight1 = 2 +/- 0.3 C L(-INF - +INF)
data1Comb:
DataStore sample_1 (sample_1)
Contains 1 entries
Observables:
1) x1 = 5 L(0 - 10) "x1"
2) x2 = 5 L(0 - 10) "x2"
3) weight1 = 2 +/- 0.3 C L(-INF - +INF) "weight1"
4) weight2 = 1 C L(-INF - +INF) "weight2"
Dataset variable "weight" is interpreted as the event weight
RooRealVar::weight = 2 C L(-INF - +INF)
data2Comb:
DataStore sample_2 (sample_2)
Contains 1 entries
Observables:
1) x1 = 5 L(0 - 10) "x1"
2) x2 = 5 L(0 - 10) "x2"
3) weight1 = 2 +/- 0.3 C L(-INF - +INF) "weight1"
4) weight2 = 3 +/- 0.4 C L(-INF - +INF) "weight2"
Dataset variable "weight" is interpreted as the event weight
RooRealVar::weight = 3 C L(-INF - +INF)
From the output
data1 weightError: 0.3, data1Comb weightError: 0
data2 weightError: 0.4, data2Comb weightError: 0
The information of errors seem missing.
It said that for data1Comb and data2Comb, Dataset variable "weight" is interpreted as the event weight. I even don’t know where the variable weight come from.
Thanks for your attention! I would appreciate if you could give me some suggestions.