Loop with RDataFrame

ROOT
1

Hi there.
I have a root file with a branch named ADC, which consists of std::vector<int>.
I have successfully written a code using RDataFrame that draws the histogram of the fist element of ADC(that is, ADC[0]) over all events.
But it failed when I tried to do the same thing to ADC[1], ADC[2] and so on with for-loop.
I would appreciate if you could help me solve the problem.
Cheers

my program

using ints = ROOT::RVec<int>;

void myMacro()
{
    auto fileName = "rootfile.root";
    auto treeName = "tree";
    TString pdfname = "histogram.pdf",
            pdf = "pdf";

    auto c = new TCanvas();
    c->SetLogy();
    c->Print(pdfname + "[", pdf);

    ROOT::RDataFrame df(treeName, fileName, {"ADC"});

    // this works fine :)
    {
        // Draw the histogram of first element of branch `ADC'
        auto first_elm = [](ints v) { return v.at(0); };
        auto histo = df.Define("ADC0", first_elm, {"ADC"})
                         .Histo1D({"histo", "title;ADC(ch);Counts", 2048, 0, 4096}, "ADC0");
        histo->Draw();
        c->Print(pdfname, pdf);
    }

    // this doesn't work fine :(
    /*
    for (int i = 0; i < 30; ++i)
    {
        // Draw the histogram of i-th element of branch `ADC'
        auto ith = [](int i, ints v) { return v.at(i); };
        auto histo = df.Define("ADCi", ith, {i, "ADC"})
                         .Histo1D({"histo", "title;ADC(ch);Counts", 2048, 0, 4096}, "ADCi");
        histo->Draw();
        c->Print(pdfname, pdf);
    }
    */

    c->Print(pdfname + "]", pdf);
}

error message

myMacro.cpp:41:46: error: non-constant-expression cannot be narrowed from type 'int' to 'std::__1::vector<std::__1::basic_string<char>,
      std::__1::allocator<std::__1::basic_string<char> > >::size_type' (aka 'unsigned long') in initializer list [-Wc++11-narrowing]
        auto histo = df.Define("ADCi", ith, {i, "ADC"})
                                             ^
myMacro.cpp:41:46: note: insert an explicit cast to silence this issue
        auto histo = df.Define("ADCi", ith, {i, "ADC"})
                                             ^
                                             static_cast<size_type>( )

environment

❯ uname -a
Darwin ##############.ac.jp 20.4.0 Darwin Kernel Version 20.4.0: Fri Mar  5 01:14:02 PST 2021; root:xnu-7195.101.1~3/RELEASE_ARM64_T8101 arm64

❯ root --version
ROOT Version: 6.22/08
Built for macosxarm64 on Mar 10 2021, 14:20:04
From tags/v6-22-08@v6-22-08
2

Not sure about this: df.Define("ADCi", ith, {i, "ADC"}) (Looking at ROOT::RDataFrame::Define()), but I might be wrong. @eguiraud knows, for sure

3

Hi @haltack,
the third argument of a Define is a list of column names as strings, you cannot pass arbitrary types in there. To pass non-column values into your expressions you can use C++11 lambda captures:

    for (int i = 0; i < 30; ++i)
    {
        // Draw the histogram of i-th element of branch `ADC'
        auto ith = [i](ints v) { return v.at(i); };
        auto histo = df.Define("ADCi", ith, {"ADC"})
                         .Histo1D({"histo", "title;ADC(ch);Counts", 2048, 0, 4096}, "ADCi");
        histo->Draw();
        c->Print(pdfname, pdf);
    }

Unrelated, but to avoid unnecessary copies ints should be RVec<int>& or const RVec<int>& (i.e. pass-by-reference rather than pass-by-copy).

Finally, if you call histo->Draw directly inside the loop, RDataFrame is forced to read all data and fill the histogram 30 times, once per call! This is a better setup (might have typos in it, but it should convey the idea):

    std::vector<ROOT::RDF::RResultPtr<TH1D>> histos;
    for (int i = 0; i < 30; ++i)
    {
        // Book the filling of the histogram of i-th element of branch `ADC'
        auto ith = [i](ints v) { return v.at(i); };
        auto histo = df.Define("ADCi", ith, {"ADC"})
                         .Histo1D({"histo", "title;ADC(ch);Counts", 2048, 0, 4096}, "ADCi");
        histos.push_back(histo);
    }

    for (auto &h : histos) {
        h->Draw();
        c->Print(pdfname, pdf);
   }

Here we book the filling of all histograms first, and then when we first access one of the results RDataFrame can run a single loop over the data that fills all booked histograms. See the RDF user guide for more details.

Cheers,
Enrico

P.S.
and as you are using macros rather than compiled C++ programs: make sure to run the macro as root mymacro.C+ or, in the prompt, load it as .L macro.C+ (with a +) to compile it with optimizations to get the best performance. Otherwise the ROOT interpreter runs everything without optimizations (we are looking into improving the situation).

1 Like
4

@bellenot @eguiraud

Thank you for your help!
@eguiraud 's code works completely fine when executed as root myMacro.cpp.

However, root myMacro.cpp+ doesn’t work fine. It displays quite large amount of stderr output (more than 1,000 lines!). I’m not sure, but the cause is at the step of linking libraries. I attach the stderr output just in case.

Cheers

log.txt (115.0 KB)

5

These are all warnings, no errors, and then the program actually runs.

These warnings are something that came up with the very latest Mac upgrade ~10 days ago and it’s something I’m working on fixing, actually – but as annoying as they are, you can ignore them.

As you see at the bottom you actually get a .pdf out.

Super annoying that these come up in user code that uses 6.22 by the way, argh. I’ll try to fix them soon and then we’ll have to publish a patch release that you can upgrade to :slight_smile:

6

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