Skip to content

[df] Handle containers in RHist filling #22899

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Draft
hahnjo wants to merge 2 commits into
base: master
Choose a base branch
from
+326 −3
Draft

[df] Handle containers in RHist filling #22899

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
bcf76f6
[df] Handle containers in RHist filling
hahnjo Jul 23, 2026
613e0bd
WIP: handle containers in RHistEngine filling
hahnjo Jul 23, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
54 changes: 51 additions & 3 deletions tree/dataframe/inc/ROOT/RDF/ActionHelpers.hxx
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -513,6 +513,15 @@ public:
void Initialize() {}
void InitTask(TTreeReader *, unsigned int) {}

template <typename... Iterators>
void ExecLoop(unsigned int slot, std::size_t elements, Iterators... its)
{
for (std::size_t i = 0; i < elements; i++) {
Exec(slot, *its...);
(std::advance(its, 1), ...);
}
}

template <typename... ColumnTypes, const std::size_t... I>
void
ExecWithWeight(unsigned int slot, const std::tuple<const ColumnTypes &...> &columnValues, std::index_sequence<I...>)
Expand All @@ -526,7 +535,22 @@ public:
template <typename... ColumnTypes>
void Exec(unsigned int slot, const ColumnTypes &...columnValues)
{
if constexpr (WithWeight) {
if constexpr (std::disjunction_v<IsDataContainer<ColumnTypes>...>) {
constexpr std::array<bool, sizeof...(ColumnTypes)> isContainer{IsDataContainer<ColumnTypes>::value...};
constexpr std::size_t firstContainerIdx = FindIdxTrue(isContainer);
std::array<std::size_t, sizeof...(columnValues)> sizes = {{GetSize(columnValues)...}};
std::size_t elements = 0;
for (std::size_t i = 0; i < isContainer.size(); i++) {
if (isContainer[i]) {
if (i == firstContainerIdx) {
elements = sizes[i];
} else if (elements != sizes[i]) {
throw std::runtime_error("Cannot fill values in containers of different sizes.");
}
}
}
ExecLoop(slot, elements, MakeBegin(columnValues)...);
} else if constexpr (WithWeight) {
auto t = std::forward_as_tuple(columnValues...);
ExecWithWeight(slot, t, std::make_index_sequence<sizeof...(ColumnTypes) - 1>());
} else {
Expand Down Expand Up @@ -566,6 +590,15 @@ public:
void Initialize() {}
void InitTask(TTreeReader *, unsigned int) {}

template <typename... Iterators>
void ExecLoop(unsigned int slot, std::size_t elements, Iterators... its)
{
for (std::size_t i = 0; i < elements; i++) {
Exec(slot, *its...);
(std::advance(its, 1), ...);
}
}

template <typename... ColumnTypes, const std::size_t... I>
void ExecWithWeight(const std::tuple<const ColumnTypes &...> &columnValues, std::index_sequence<I...>)
{
Expand All @@ -576,9 +609,24 @@ public:
}

template <typename... ColumnTypes>
void Exec(unsigned int, const ColumnTypes &...columnValues)
void Exec(unsigned int slot, const ColumnTypes &...columnValues)
{
if constexpr (WithWeight) {
if constexpr (std::disjunction_v<IsDataContainer<ColumnTypes>...>) {
constexpr std::array<bool, sizeof...(ColumnTypes)> isContainer{IsDataContainer<ColumnTypes>::value...};
constexpr std::size_t firstContainerIdx = FindIdxTrue(isContainer);
std::array<std::size_t, sizeof...(columnValues)> sizes = {{GetSize(columnValues)...}};
std::size_t elements = 0;
for (std::size_t i = 0; i < isContainer.size(); i++) {
if (isContainer[i]) {
if (i == firstContainerIdx) {
elements = sizes[i];
} else if (elements != sizes[i]) {
throw std::runtime_error("Cannot fill values in containers of different sizes.");
}
}
}
ExecLoop(slot, elements, MakeBegin(columnValues)...);
} else if constexpr (WithWeight) {
auto t = std::forward_as_tuple(columnValues...);
ExecWithWeight(t, std::make_index_sequence<sizeof...(ColumnTypes) - 1>());
} else {
Expand Down
275 changes: 275 additions & 0 deletions tree/dataframe/test/dataframe_hist.cxx
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -8,6 +8,7 @@
#include <ROOT/RHistEngine.hxx>
#include <ROOT/RRegularAxis.hxx>
#include <ROOT/RVariableBinAxis.hxx>
#include <ROOT/RVec.hxx>

#include <cstddef>
#include <memory>
Expand Down Expand Up @@ -455,6 +456,280 @@ TEST_P(RDFHist, WeightInvalidNumberOfArgumentsJit)
EXPECT_THROW(dfXW.Hist(engine, {"x", "x"}, "w"), std::invalid_argument);
}

TEST_P(RDFHist, Container)
{
RDataFrame df(10);
auto dfX = df.Define("x", [](ULong64_t e) { return ROOT::RVecD{e + 5.5}; }, {"rdfentry_"});

const RRegularAxis axis(10, {5.0, 15.0});
auto hist = dfX.Hist</*BinContentType=*/double, ROOT::RVecD>({axis}, {"x"});
EXPECT_EQ(hist->GetNEntries(), 10);
for (auto index : axis.GetNormalRange()) {
EXPECT_EQ(hist->GetBinContent(index), 1.0);
}
}

TEST_P(RDFHist, ContainerEmpty)
{
RDataFrame df(10);
auto dfX = df.Define("x", [] { return ROOT::RVecD{}; });

const RRegularAxis axis(10, {5.0, 15.0});
auto hist = dfX.Hist</*BinContentType=*/double, ROOT::RVecD>({axis}, {"x"});
EXPECT_EQ(hist->GetNEntries(), 0);
}

TEST_P(RDFHist, ContainerElements)
{
RDataFrame df(10);
auto dfX = df.Define("x", [](ULong64_t e) { return ROOT::RVecD{e + 5.5, e + 15.5}; }, {"rdfentry_"});

const RRegularAxis axis(20, {5.0, 25.0});
auto hist = dfX.Hist</*BinContentType=*/double, ROOT::RVecD>({axis}, {"x"});
EXPECT_EQ(hist->GetNEntries(), 20);
for (auto index : axis.GetNormalRange()) {
EXPECT_EQ(hist->GetBinContent(index), 1.0);
}
}

TEST_P(RDFHist, ContainerNested)
{
RDataFrame df(10);
auto dfX = df.Define("x", [](ULong64_t e) { return ROOT::RVec<ROOT::RVecD>{{e + 5.5}, {e + 15.5}}; }, {"rdfentry_"});

const RRegularAxis axis(20, {5.0, 25.0});
auto hist = dfX.Hist</*BinContentType=*/double, ROOT::RVec<ROOT::RVecD>>({axis}, {"x"});
EXPECT_EQ(hist->GetNEntries(), 20);
for (auto index : axis.GetNormalRange()) {
EXPECT_EQ(hist->GetBinContent(index), 1.0);
}
}

TEST_P(RDFHist, ContainerJit)
{
RDataFrame df(10);
auto dfX = df.Define("x", "ROOT::RVecD{rdfentry_ + 5.5}");

const RRegularAxis axis(10, {5.0, 15.0});
auto hist = dfX.Hist({axis}, {"x"});
EXPECT_EQ(hist->GetNEntries(), 10);
for (auto index : axis.GetNormalRange()) {
EXPECT_EQ(hist->GetBinContent(index), 1.0);
}
}

TEST_P(RDFHist, ContainerMultiDim)
{
RDataFrame df(10);
auto dfXY = df.Define("x", [](ULong64_t e) { return ROOT::RVecD{e + 5.5}; }, {"rdfentry_"})
.Define("y", [](ULong64_t e) { return ROOT::RVecD{2 * e + 0.5}; }, {"rdfentry_"});

const RRegularAxis regularAxis(10, {5.0, 15.0});
static constexpr std::size_t BinsY = 20;
std::vector<double> bins;
for (std::size_t i = 0; i < BinsY; i++) {
bins.push_back(i);
}
bins.push_back(BinsY);
const RVariableBinAxis variableBinAxis(bins);

auto hist =
dfXY.Hist</*BinContentType=*/double, ROOT::RVecD, ROOT::RVecD>({regularAxis, variableBinAxis}, {"x", "y"});
EXPECT_EQ(hist->GetNEntries(), 10);
for (auto x : regularAxis.GetNormalRange()) {
for (auto y : variableBinAxis.GetNormalRange()) {
if (2 * x.GetIndex() == y.GetIndex()) {
EXPECT_EQ(hist->GetBinContent(x, y), 1.0);
} else {
EXPECT_EQ(hist->GetBinContent(x, y), 0.0);
}
}
}
}

TEST_P(RDFHist, ContainerMultiDimBroadcast)
{
RDataFrame df(10);
auto dfXY = df.Define("x", [](ULong64_t e) { return e + 5.5; }, {"rdfentry_"})
.Define("y", [](ULong64_t e) { return ROOT::RVecD{2 * e + 0.5}; }, {"rdfentry_"});

const RRegularAxis regularAxis(10, {5.0, 15.0});
static constexpr std::size_t BinsY = 20;
std::vector<double> bins;
for (std::size_t i = 0; i < BinsY; i++) {
bins.push_back(i);
}
bins.push_back(BinsY);
const RVariableBinAxis variableBinAxis(bins);

auto hist = dfXY.Hist</*BinContentType=*/double, double, ROOT::RVecD>({regularAxis, variableBinAxis}, {"x", "y"});
EXPECT_EQ(hist->GetNEntries(), 10);
for (auto x : regularAxis.GetNormalRange()) {
for (auto y : variableBinAxis.GetNormalRange()) {
if (2 * x.GetIndex() == y.GetIndex()) {
EXPECT_EQ(hist->GetBinContent(x, y), 1.0);
} else {
EXPECT_EQ(hist->GetBinContent(x, y), 0.0);
}
}
}
}

TEST_P(RDFHist, ContainerMultiDimNestedBroadcast)
{
RDataFrame df(10);
auto dfXY = df.Define("x", [](ULong64_t e) { return e + 5.5; }, {"rdfentry_"})
.Define("xV", [](ULong64_t e) { return ROOT::RVecD{e + 5.5}; }, {"rdfentry_"})
.Define("y", [](ULong64_t e) { return ROOT::RVec<ROOT::RVecD>{{2 * e + 0.5}}; }, {"rdfentry_"});

const RRegularAxis regularAxis(10, {5.0, 15.0});
static constexpr std::size_t BinsY = 20;
std::vector<double> bins;
for (std::size_t i = 0; i < BinsY; i++) {
bins.push_back(i);
}
bins.push_back(BinsY);
const RVariableBinAxis variableBinAxis(bins);

auto hist =
dfXY.Hist</*BinContentType=*/double, double, ROOT::RVec<ROOT::RVecD>>({regularAxis, variableBinAxis}, {"x", "y"});
EXPECT_EQ(hist->GetNEntries(), 10);
for (auto x : regularAxis.GetNormalRange()) {
for (auto y : variableBinAxis.GetNormalRange()) {
if (2 * x.GetIndex() == y.GetIndex()) {
EXPECT_EQ(hist->GetBinContent(x, y), 1.0);
} else {
EXPECT_EQ(hist->GetBinContent(x, y), 0.0);
}
}
}

hist = dfXY.Hist</*BinContentType=*/double, ROOT::RVecD, ROOT::RVec<ROOT::RVecD>>({regularAxis, variableBinAxis},
{"xV", "y"});
EXPECT_EQ(hist->GetNEntries(), 10);
}

TEST_P(RDFHist, ContainerInvalidElements)
{
RDataFrame df(10);
auto dfXY = df.Define("x0", [] { return ROOT::RVecD{}; })
.Define("x2", [](ULong64_t e) { return ROOT::RVecD{e + 5.5, e + 15.5}; }, {"rdfentry_"})
.Define("y1", [](ULong64_t e) { return ROOT::RVecD{2 * e + 0.5}; }, {"rdfentry_"});

const RRegularAxis regularAxis(20, {5.0, 25.0});
static constexpr std::size_t BinsY = 20;
std::vector<double> bins;
for (std::size_t i = 0; i < BinsY; i++) {
bins.push_back(i);
}
bins.push_back(BinsY);
const RVariableBinAxis variableBinAxis(bins);

try {
// Cannot use EXPECT_THROW because of template arguments...
*dfXY.Hist</*BinContentType=*/double, ROOT::RVecD, ROOT::RVecD>({regularAxis, variableBinAxis}, {"x0", "y1"});
FAIL() << "expected std::runtime_error";
} catch (const std::runtime_error &) {
// expected
}

try {
// Cannot use EXPECT_THROW because of template arguments...
*dfXY.Hist</*BinContentType=*/double, ROOT::RVecD, ROOT::RVecD>({regularAxis, variableBinAxis}, {"x2", "y1"});
FAIL() << "expected std::runtime_error";
} catch (const std::runtime_error &e) {
// expected
}
}

TEST_P(RDFHist, ContainerNestedInvalidElements)
{
RDataFrame df(10);
auto dfXY =
df.Define("x11", [](ULong64_t e) { return ROOT::RVec<ROOT::RVecD>{{e + 5.5}}; }, {"rdfentry_"})
.Define("y12", [](ULong64_t e) { return ROOT::RVec<ROOT::RVecD>{{2 * e + 0.5, 2 * e + 0.5}}; }, {"rdfentry_"});

const RRegularAxis regularAxis(20, {5.0, 25.0});
static constexpr std::size_t BinsY = 20;
std::vector<double> bins;
for (std::size_t i = 0; i < BinsY; i++) {
bins.push_back(i);
}
bins.push_back(BinsY);
const RVariableBinAxis variableBinAxis(bins);

try {
// Cannot use EXPECT_THROW because of template arguments...
*dfXY.Hist</*BinContentType=*/double, ROOT::RVec<ROOT::RVecD>, ROOT::RVec<ROOT::RVecD>>(
{regularAxis, variableBinAxis}, {"x11", "y12"});
FAIL() << "expected std::runtime_error";
} catch (const std::runtime_error &e) {
// expected
}
}

TEST_P(RDFHist, WeightContainer)
{
RDataFrame df(10);
auto dfXW =
df.Define("xV", [](ULong64_t e) { return ROOT::RVecD{e + 5.5, e + 15.5}; }, {"rdfentry_"})
.Define("wV", [](ULong64_t e) { return ROOT::RVecD{0.1 + e * 0.03, 0.2 + e * e * 0.06}; }, {"rdfentry_"});

const RRegularAxis axis(20, {5.0, 25.0});
auto hist = dfXW.Hist</*BinContentType=*/RBinWithError, ROOT::RVecD, ROOT::RVecD>({axis}, {"xV"}, "wV");
EXPECT_EQ(hist->GetNEntries(), 20);
for (auto index : axis.GetNormalRange()) {
auto &bin = hist->GetBinContent(index);
auto i = index.GetIndex();
double weight = i >= 10 ? (0.2 + (i - 10) * (i - 10) * 0.06) : (0.1 + i * 0.03);
EXPECT_FLOAT_EQ(bin.fSum, weight);
EXPECT_FLOAT_EQ(bin.fSum2, weight * weight);
}
}

TEST_P(RDFHist, WeightContainerJit)
{
RDataFrame df(10);
auto dfXW = df.Define("xV", "ROOT::RVecD{rdfentry_ + 5.5, rdfentry_ + 15.5}")
.Define("wV", "ROOT::RVecD{0.1 + rdfentry_ * 0.03, 0.2 + rdfentry_ * rdfentry_ * 0.06}");

const RRegularAxis axis(20, {5.0, 25.0});
auto hist = dfXW.Hist({axis}, {"xV"}, "wV");
EXPECT_EQ(hist->GetNEntries(), 20);
for (auto index : axis.GetNormalRange()) {
auto &bin = hist->GetBinContent(index);
auto i = index.GetIndex();
double weight = i >= 10 ? (0.2 + (i - 10) * (i - 10) * 0.06) : (0.1 + i * 0.03);
EXPECT_FLOAT_EQ(bin.fSum, weight);
EXPECT_FLOAT_EQ(bin.fSum2, weight * weight);
}
}

TEST_P(RDFHist, WeightContainerBroadcast)
{
RDataFrame df(10);
auto dfXW =
df.Define("x", [](ULong64_t e) { return e + 5.5; }, {"rdfentry_"})
.Define("xV", [](ULong64_t e) { return ROOT::RVecD{e + 5.5, e + 15.5}; }, {"rdfentry_"})
.Define("w", [](ULong64_t e) { return 0.1 + e * 0.03; }, {"rdfentry_"})
.Define("wV", [](ULong64_t e) { return ROOT::RVecD{0.1 + e * 0.03, 0.2 + e * e * 0.06}; }, {"rdfentry_"});

const RRegularAxis axis(20, {5.0, 25.0});
auto hist = dfXW.Hist</*BinContentType=*/RBinWithError, ROOT::RVecD, double>({axis}, {"xV"}, "w");
EXPECT_EQ(hist->GetNEntries(), 20);
for (auto index : axis.GetNormalRange()) {
auto &bin = hist->GetBinContent(index);
auto i = index.GetIndex();
double weight = 0.1 + (i >= 10 ? i - 10 : i) * 0.03;
EXPECT_FLOAT_EQ(bin.fSum, weight);
EXPECT_FLOAT_EQ(bin.fSum2, weight * weight);
}

hist = dfXW.Hist</*BinContentType=*/RBinWithError, double, ROOT::RVecD>({axis}, {"x"}, "wV");
EXPECT_EQ(hist->GetNEntries(), 20);
EXPECT_EQ(hist->GetBinContent(2).fSum, 0.1 + 2 * 0.03 + 0.2 + 2 * 2 * 0.06);
}

INSTANTIATE_TEST_SUITE_P(Seq, RDFHist, ::testing::Values(false));

#ifdef R__USE_IMT
Expand Down
Loading