Logo ROOT  
Reference Guide
 
Loading...
Searching...
No Matches
RTensorUtils.hxx
Go to the documentation of this file.
1#ifndef TMVA_RTENSOR_UTILS
2#define TMVA_RTENSOR_UTILS
3
4#include <vector>
5#include <string>
6
7#include "TMVA/RTensor.hxx"
8#include "ROOT/RDataFrame.hxx"
10
11namespace TMVA {
12namespace Experimental {
13
14/// \brief Convert the content of an RDataFrame to an RTensor
15/// \param[in] dataframe RDataFrame node
16/// \param[in] columns Vector of column names
17/// \param[in] layout Memory layout
18/// \return RTensor with content from selected columns
19template <typename T, typename U>
20RTensor<T>
21AsTensor(U &dataframe, std::vector<std::string> columns = {}, MemoryLayout layout = MemoryLayout::RowMajor)
22{
23 // If no columns are specified, get all columns from dataframe
24 if (columns.size() == 0) {
25 columns = dataframe.GetColumnNames();
26 }
27
28 // Book actions to read-out columns of dataframe in vectors
29 using ResultPtr = ROOT::RDF::RResultPtr<std::vector<T>>;
30 std::vector<ResultPtr> resultPtrs;
31 for (auto &col : columns) {
32 resultPtrs.emplace_back(dataframe.template Take<T>(col));
33 }
34
35 // Copy data to tensor based on requested memory layout
36 const auto numCols = resultPtrs.size();
37 const auto numEntries = resultPtrs[0]->size();
38 RTensor<T> x({numEntries, numCols}, layout);
39 const auto data = x.GetData();
40 if (layout == MemoryLayout::RowMajor) {
41 for (std::size_t i = 0; i < numEntries; i++) {
42 const auto entry = data + numCols * i;
43 for (std::size_t j = 0; j < numCols; j++) {
44 entry[j] = resultPtrs[j]->at(i);
45 }
46 }
47 } else if (layout == MemoryLayout::ColumnMajor) {
48 for (std::size_t i = 0; i < numCols; i++) {
49 // TODO: Replace by RVec<T>::insert as soon as available.
50 std::memcpy(data + numEntries * i, &resultPtrs[i]->at(0), numEntries * sizeof(T));
51 }
52 } else {
53 throw std::runtime_error("Memory layout is not known.");
54 }
55
56 // Remove dimensions of 1
57 x.Squeeze();
58
59 return x;
60}
61
62} // namespace TMVA::Experimental
63} // namespace TMVA
64
65#endif
Option_t Option_t TPoint TPoint const char GetTextMagnitude GetFillStyle GetLineColor GetLineWidth GetMarkerStyle GetTextAlign GetTextColor GetTextSize void data
Smart pointer for the return type of actions.
Double_t x[n]
Definition legend1.C:17
RTensor< T > AsTensor(U &dataframe, std::vector< std::string > columns={}, MemoryLayout layout=MemoryLayout::RowMajor)
Convert the content of an RDataFrame to an RTensor.
MemoryLayout
Memory layout type (copy from RTensor.hxx)
Definition CudaTensor.h:47
create variable transformations