Skip to content

GH-50906: [C++] Support 1D tensors in SparseCSR/CSC matrix conversion #50907

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

Open
pratyushadk wants to merge 1 commit into
base: main
Choose a base branch
from
+106 −8
Open

GH-50906: [C++] Support 1D tensors in SparseCSR/CSC matrix conversion #50907

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
64 changes: 64 additions & 0 deletions cpp/src/arrow/sparse_tensor_test.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -869,6 +869,70 @@ TEST_F(TestSparseCSRMatrix, TestToTensor) {
ASSERT_TRUE(tensor.Equals(*dense_tensor));
}

TEST_F(TestSparseCSRMatrix, CreationFromNumericTensor1D) {
// Dense 1D vector: [1, 0, 2, 0, 0, 3, 0, 4]
// Expected: indptr = [0, 4], indices = [0, 2, 5, 7], values = [1, 2, 3, 4]
std::vector<int64_t> values = {1, 0, 2, 0, 0, 3, 0, 4};
std::vector<int64_t> shape({static_cast<int64_t>(values.size())});
NumericTensor<Int64Type> dense_vector(Buffer::Wrap(values), shape);

std::shared_ptr<SparseCSRMatrix> st;
ASSERT_OK_AND_ASSIGN(st, SparseCSRMatrix::Make(dense_vector));

ASSERT_EQ(4, st->non_zero_length());
ASSERT_TRUE(st->is_mutable());

const int64_t* raw_data = reinterpret_cast<const int64_t*>(st->raw_data());
AssertNumericDataEqual(raw_data, {1, 2, 3, 4});

auto si = internal::checked_pointer_cast<SparseCSRIndex>(st->sparse_index());
ASSERT_EQ(1, si->indptr()->ndim());
ASSERT_EQ(1, si->indices()->ndim());

const int64_t* indptr_begin =
reinterpret_cast<const int64_t*>(si->indptr()->raw_data());
std::vector<int64_t> indptr_values(indptr_begin,
indptr_begin + si->indptr()->shape()[0]);
ASSERT_EQ(std::vector<int64_t>({0, 4}), indptr_values);

const int64_t* indices_begin =
reinterpret_cast<const int64_t*>(si->indices()->raw_data());
std::vector<int64_t> indices_values(indices_begin,
indices_begin + si->indices()->shape()[0]);
ASSERT_EQ(std::vector<int64_t>({0, 2, 5, 7}), indices_values);
}

TEST_F(TestSparseCSRMatrix, CreationFromAllZeroTensor1D) {
std::vector<int64_t> values = {0, 0, 0, 0};
std::vector<int64_t> shape({static_cast<int64_t>(values.size())});
NumericTensor<Int64Type> dense_vector(Buffer::Wrap(values), shape);

std::shared_ptr<SparseCSRMatrix> st;
ASSERT_OK_AND_ASSIGN(st, SparseCSRMatrix::Make(dense_vector));

ASSERT_EQ(0, st->non_zero_length());

auto si = internal::checked_pointer_cast<SparseCSRIndex>(st->sparse_index());
const int64_t* indptr_begin =
reinterpret_cast<const int64_t*>(si->indptr()->raw_data());
std::vector<int64_t> indptr_values(indptr_begin,
indptr_begin + si->indptr()->shape()[0]);
ASSERT_EQ(std::vector<int64_t>({0, 0}), indptr_values);
}

TEST_F(TestSparseCSRMatrix, TestToTensor1D) {
std::vector<int64_t> values = {1, 0, 2, 0, 0, 3, 0, 4};
std::vector<int64_t> shape({static_cast<int64_t>(values.size())});
Tensor tensor(int64(), Buffer::Wrap(values), shape);

std::shared_ptr<SparseCSRMatrix> st;
ASSERT_OK_AND_ASSIGN(st, SparseCSRMatrix::Make(tensor));

ASSERT_EQ(4, st->non_zero_length());
ASSERT_OK_AND_ASSIGN(auto dense_tensor, st->ToTensor());
ASSERT_TRUE(tensor.Equals(*dense_tensor));
}

template <typename ValueType>
class TestSparseCSRMatrixEquality : public TestSparseTensorBase<ValueType> {
public:
Expand Down
50 changes: 42 additions & 8 deletions cpp/src/arrow/tensor/csx_converter.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -56,13 +56,10 @@ class SparseCSXMatrixConverter : private SparseTensorConverterMixin {
const int value_elsize = tensor_.type()->byte_width();

const int64_t ndim = tensor_.ndim();
if (ndim > 2) {
if (ndim == 0 || ndim > 2) {
return Status::Invalid("Invalid tensor dimension");
}

const int major_axis = static_cast<int>(axis_);
const int64_t n_major = tensor_.shape()[major_axis];
const int64_t n_minor = tensor_.shape()[1 - major_axis];
ARROW_ASSIGN_OR_RAISE(int64_t nonzero_count, tensor_.CountNonZero());

std::shared_ptr<Buffer> indptr_buffer;
Expand All @@ -74,9 +71,45 @@ class SparseCSXMatrixConverter : private SparseTensorConverterMixin {

const auto* tensor_data = tensor_.raw_data();

if (ndim <= 1) {
return Status::NotImplemented("TODO for ndim <= 1");
int64_t indptr_length; // n_major + 1 for 2D; 2 for 1D

if (ndim == 1) {
// A 1D vector is treated as a single-row (CSR) or single-column (CSC)
// matrix. indptr has exactly 2 entries: [0, nnz].
const int64_t n = tensor_.shape()[0];

ARROW_ASSIGN_OR_RAISE(indptr_buffer, AllocateBuffer(index_elsize * 2, pool_));
auto* indptr = indptr_buffer->mutable_data();

ARROW_ASSIGN_OR_RAISE(indices_buffer,
AllocateBuffer(index_elsize * nonzero_count, pool_));
auto* indices = indices_buffer->mutable_data();

// indptr[0] = 0
AssignIndex(indptr, 0, index_elsize);
indptr += index_elsize;

int64_t k = 0;
for (int64_t i = 0; i < n; ++i) {
const int64_t offset = i * value_elsize;
if (std::any_of(tensor_data + offset, tensor_data + offset + value_elsize,
IsNonZero)) {
std::copy_n(tensor_data + offset, value_elsize, values);
values += value_elsize;
AssignIndex(indices, i, index_elsize);
indices += index_elsize;
k++;
}
}

// indptr[1] = nnz
AssignIndex(indptr, k, index_elsize);
indptr_length = 2;
} else {
const int major_axis = static_cast<int>(axis_);
const int64_t n_major = tensor_.shape()[major_axis];
const int64_t n_minor = tensor_.shape()[1 - major_axis];

ARROW_ASSIGN_OR_RAISE(indptr_buffer,
AllocateBuffer(index_elsize * (n_major + 1), pool_));
auto* indptr = indptr_buffer->mutable_data();
Expand Down Expand Up @@ -111,9 +144,10 @@ class SparseCSXMatrixConverter : private SparseTensorConverterMixin {
AssignIndex(indptr, k, index_elsize);
indptr += index_elsize;
}
indptr_length = n_major + 1;
}

std::vector<int64_t> indptr_shape({n_major + 1});
std::vector<int64_t> indptr_shape({indptr_length});
std::shared_ptr<Tensor> indptr_tensor =
std::make_shared<Tensor>(index_value_type_, indptr_buffer, indptr_shape);

Expand Down Expand Up @@ -181,7 +215,7 @@ Result<std::shared_ptr<Tensor>> MakeTensorFromSparseCSXMatrix(
std::vector<int64_t> strides;
RETURN_NOT_OK(ComputeRowMajorStrides(fw_value_type, shape, &strides));

const auto nc = shape[1];
const auto nc = shape.size() > 1 ? shape[1] : shape[0];

int64_t offset = 0;
for (int64_t i = 0; i < indptr->size() - 1; ++i) {
Expand Down
Loading