Error in <TTreeReaderValueBase::CreateProxy()>

Newbie
,
1

Hello,

I am trying to create histogram which will have only positively charged hadronic taus in it using RDataFrame and I get following error :

Error in <TTreeReaderValueBase::CreateProxy()>: The branch GenVisTau_charge contains data of type {UNDETERMINED TYPE}, which does not have a dictionary.
terminate called after throwing an instance of 'std::runtime_error'
  what():  An error was encountered while processing the data. TTreeReader status code is: 6

The main script which is supposed to return tau pt for only + charge is following:

auto genVisTau_pt_plus = df.Define("GenVisTau_pt_plus", [] (
			const uint& nGenVisTau,
			const int& charge,
			const rvec<float>& genVisTau_pt){
				rvec<float> gentau_pt;
				for(uint i = 0 ; i < nGenVisTau ; i++){
						if( charge == 1){
							gentau_pt.push_back(genVisTau_pt[i]);
						}	
				}
				return gentau_pt;
			} , {"nGenVisTau", "GenVisTau_charge", "GenVisTau_pt"});

What I normally would have tried is instead of
if ( charge == 1 )
if ( charge[i] == 1 ) , but it does not seem to work. With the later one I get invalid types 'int[int]' for array subscript error, after which I googled and tried this if ( charge[i][0] == 1 ) , but it also does not work.
As it seems I have a caveat in understanding how this works in RDataFrame. I would appreciate any help, thanks!

2

Worked like this, maybe can be useful for someone else :slight_smile:

			const uint& nGenVisTau,
			const rvec<int>& charge,
			const rvec<float>& genVisTau_pt){
				rvec<float> gentau_pt;
				for(uint i = 0 ; i < nGenVisTau ; i++){
					
						if( charge[i] == 1 && charge[i] != -1){
							gentau_pt.push_back(genVisTau_pt[i]);
						}
				}
				return gentau_pt;
			} , {"nGenVisTau", "GenVisTau_charge", "GenVisTau_pt"});
3

Hi @beg1nner ,

sorry for the trouble! If GenVisTau_charge is an array of integers, then the right type to use to read it is ROOT::RVec<int> (or, with the latest ROOT, ROOT::RVecI). You can check what type the column is with df.GetColumnType("GenVisTau_charge").

Now, you should not get that error message if you use the wrong type for a column. RDF should print a message that’s a bit friendlier and tells you about the type mismatch issue.
What ROOT version are you using?

Cheers,
Enrico

4

Hi @eguiraud ,

Thank you for your response. I have posted the solution to my question shortly before your answer. Indeed, using RVec solved the issue.
I’m using ROOT 6.24/00 with g++ (GCC) 10.1.0 . As you said, it would be nicer to see a friendlier error, because with this one I did not immediately get that it has problems with type.

Cheers

5

it would be nicer to see a friendlier error, because with this one I did not immediately get that it has problems with type

I totally agree – the weird thing is, you should have gotten a friendlier error :smiley: If you can share a minimal reproducer with me (including data) I could check what the behavior is with the latest ROOT version.

For example the following code:

#include <ROOT/RDataFrame.hxx>

int main()
{
   // write an array of ints
   ROOT::RDataFrame(10).Define("arr", "ROOT::RVec<int>{1,2,3}").Snapshot("t", "test.root");
   // read it back as int instead of RVec<int>
   ROOT::RDataFrame("t", "test.root").Define("x", [](int a) { return a; }, {"arr"}).Count().GetValue();
}

produces:

Error in <TTreeReaderValueBase::CreateProxy()>: The branch arr contains data of type ROOT::VecOps::RVec<int>. It cannot be accessed by a TTreeReaderValue<int>
terminate called after throwing an instance of 'std::runtime_error'
  what():  An error was encountered while processing the data. TTreeReader status code is: 6
fish: Job 1, './foo' terminated by signal SIGABRT (Abort)

Cheers,
Enrico