CXX Interface

Overview

The TBLIS [C interface](C Interface) is complicated by requiring explicit user specification and management of tensor layouts, generic typing, etc. The C++ interface is designed to streamline this process by providing convenient tensor, matrix, and vector container and view objects. TBLIS containers behave similarly to STL containers in that they own the tensor/matrix/vector data and handle allocation and deallocation based on the container lifetime. Views offer a similar interface for accessing the underlying data, but do not imply ownership and are relatively lightweight. View come in two flavors: const_views that do not allow modification of the underlying data (but can be shifted, resized, etc.), and non-const views, that do allow modification. Container and view objects are templated over the underlying data type.

Tensor, Matrix, and Vector Objects

Object Type	Tensor	Matrix	Vector
container	tensor<T, Allocator=...>	matrix<T, Allocator=...>	row<T, Allocator=...>
const view	const_tensor_view<T>	const_matrix_view<T>	const_row_view<T>
non-const view	tensor_view<T>	matrix_view<T>	row_view<T>

C++ Interface for Tensor Operations

The six basic [tensor operations](Tensor Operations) (and their matrix and vector counterparts) are accessible via:

Operation	Interface
add	template void add(T alpha, const_row_view A, T beta, row_view B); template void add(single_t s, T alpha, const_row_view A, T beta, row_view B); template void add(const communicator& comm, T alpha, const_row_view A, T beta, row_view B); template void add(T alpha, const_matrix_view A, T beta, matrix_view B); template void add(single_t s, T alpha, const_matrix_view A, T beta, matrix_view B); template void add(const communicator& comm, T alpha, const_matrix_view A, T beta, matrix_view B); template void add(T alpha, const_tensor_view A, const label_type* idx_A, T beta, tensor_view B, const label_type* idx_B); template void add(single_t s, T alpha, const_tensor_view A, const label_type* idx_A, T beta, tensor_view B, const label_type* idx_B); template void add(const communicator& comm, T alpha, const_tensor_view A, const label_type* idx_A, T beta, tensor_view B, const label_type* idx_B);
dot	template void dot(const_row_view A, const_row_view B, T& result); template void dot(single_t s, const_row_view A, const_row_view B, T& result); template void dot(const communicator& comm, const_row_view A, const_row_view B, T& result); template T dot(const_row_view A, const_row_view B); template T dot(single_t s, const_row_view A, const_row_view B); template T dot(const communicator& comm, const_row_view A, const_row_view B); template void dot(const_matrix_view A, const_matrix_view B, T& result); template void dot(single_t s, const_matrix_view A, const_matrix_view B, T& result); template void dot(const communicator& comm, const_matrix_view A, const_matrix_view B, T& result); template T dot(const_matrix_view A, const_matrix_view B); template T dot(single_t s, const_matrix_view A, const_matrix_view B); template T dot(const communicator& comm, const_matrix_view A, const_matrix_view B); template void dot(const_tensor_view A, const label_type* idx_A, const_tensor_view B, const label_type* idx_B, T& result); template void dot(single_t s, const_tensor_view A, const label_type* idx_A, const_tensor_view B, const label_type* idx_B, T& result); template void dot(const communicator& comm, const_tensor_view A, const label_type* idx_A, const_tensor_view B, const label_type* idx_B, T& result); template T dot(const_tensor_view A, const label_type* idx_A, const_tensor_view B, const label_type* idx_B); template T dot(single_t s, const_tensor_view A, const label_type* idx_A, const_tensor_view B, const label_type* idx_B); template T dot(const communicator& comm, const_tensor_view A, const label_type* idx_A, const_tensor_view B, const label_type* idx_B);
reduce	template void reduce(reduce_t op, const_row_view A, T& result, len_type& idx); template void reduce(single_t s, reduce_t op, const_row_view A, T& result, len_type& idx); template void reduce(const communicator& comm, reduce_t op, const_row_view A, T& result, len_type& idx); template std::pair<T,len_type> reduce(reduce_t op, const_row_view A); template std::pair<T,len_type> reduce(single_t s, reduce_t op, const_row_view A); template std::pair<T,len_type> reduce(const communicator& comm, reduce_t op, const_row_view A); template void reduce(reduce_t op, const_matrix_view A, T& result, len_type& idx); template void reduce(single_t s, reduce_t op, const_matrix_view A, T& result, len_type& idx) template void reduce(const communicator& comm, reduce_t op, const_matrix_view A, T& result, len_type& idx); template std::pair<T,len_type> reduce(reduce_t op, const_matrix_view A); template std::pair<T,len_type> reduce(single_t s, reduce_t op, const_matrix_view A); template std::pair<T,len_type> reduce(const communicator& comm, reduce_t op, const_matrix_view A); template void reduce(reduce_t op, const_tensor_view A, const label_type* idx_A, T& result, len_type& idx); template void reduce(single_t s, reduce_t op, const_tensor_view A, const label_type* idx_A, T& result, len_type& idx); template void reduce(const communicator& comm, reduce_t op, const_tensor_view A, const label_type* idx_A, T& result, len_type& idx); template std::pair<T,len_type> reduce(reduce_t op, const_tensor_view A, const label_type* idx_A); template std::pair<T,len_type> reduce(single_t s, reduce_t op, const_tensor_view A, const label_type* idx_A); template std::pair<T,len_type> reduce(const communicator& comm, reduce_t op, const_tensor_view A, const label_type* idx_A);
scale	template void scale(T alpha, row_view A); template void scale(single_t s, T alpha, row_view A); template void scale(const communicator& comm, T alpha, row_view A); template void scale(T alpha, matrix_view A); template void scale(single_t s, T alpha, matrix_view A); template void scale(const communicator& comm, T alpha, matrix_view A); template void scale(T alpha, tensor_view A, const label_type* idx_A); template void scale(single_t s, T alpha, tensor_view A, const label_type* idx_A); template void scale(const communicator& comm, T alpha, tensor_view A, const label_type* idx_A);
set	template void set(T alpha, row_view A); template void set(single_t s, T alpha, row_view A); template void set(const communicator& comm, T alpha, row_view A); template void set(T alpha, matrix_view A); template void set(single_t s, T alpha, matrix_view A); template void set(const communicator& comm, T alpha, matrix_view A); template void set(T alpha, tensor_view A, const label_type* idx_A); template void set(single_t s, T alpha, tensor_view A, const label_type* idx_A); template void set(const communicator& comm, T alpha, tensor_view A, const label_type* idx_A);
mult	template void mult(T alpha, const_matrix_view A, const_matrix_view B, T beta, matrix_view C); template void mult(single_t s, T alpha, const_matrix_view A, const_matrix_view B, T beta, matrix_view C); template void mult(const communicator& comm, T alpha, const_matrix_view A, const_matrix_view B, T beta, matrix_view C); template void mult(T alpha, const_tensor_view A, const label_type* idx_A, const_tensor_view B, const label_type* idx_B, T beta, tensor_view C, const label_type* idx_C); template void mult(single_t s, T alpha, const_tensor_view A, const label_type* idx_A, const_tensor_view B, const label_type* idx_B, T beta, tensor_view C, const label_type* idx_C); template void mult(const communicator& comm, T alpha, const_tensor_view A, const label_type* idx_A, const_tensor_view B, const label_type* idx_B, T beta, tensor_view C, const label_type* idx_C);

Containers and Views

Examples