container::kernels::lapack_hegvd< T, Device > Struct Template Reference

#include <lapack.h>

Collaboration diagram for container::kernels::lapack_hegvd< T, Device >:

Public Types
using	Real = typename GetTypeReal< T >::type

Public Member Functions
void	operator() (const int n, const int lda, T *Mat_A, T *Mat_B, Real *eigen_val, T *eigen_vec)
	Computes all the eigenvalues and, optionally, the eigenvectors of a complex generalized Hermitian-definite eigenproblem.

Member Typedef Documentation

Real

template<typename T , typename Device >

using container::kernels::lapack_hegvd< T, Device >::Real = typename GetTypeReal<T>::type

Member Function Documentation

operator()()

template<typename T , typename Device >

void container::kernels::lapack_hegvd< T, Device >::operator()	(	const int	n,
		const int	lda,
		T *	Mat_A,
		T *	Mat_B,
		Real *	eigen_val,
		T *	eigen_vec
	)

Computes all the eigenvalues and, optionally, the eigenvectors of a complex generalized Hermitian-definite eigenproblem.

This function solves the problem A*x = lambda*B*x, where A and B are Hermitian matrices, and B is also positive definite.

Parameters

n	The order of the matrices Mat_A and Mat_B. n >= 0.
lda	The leading dimension of the arrays Mat_A and Mat_B. lda >= max(1, n).
Mat_A	On entry, the Hermitian matrix A. On exit, it may be overwritten.
Mat_B	On entry, the Hermitian positive definite matrix B. On exit, it may be overwritten.
eigen_val	Array to store the computed eigenvalues in ascending order.
eigen_vec	If not nullptr, array to store the computed eigenvectors.

Note

See LAPACK ZHEGVD or CHEGVD documentation for more details. This function assumes that A and B have the same leading dimensions, lda. This function copies B to auxiliary memory to avoid being overwritten.

---

The documentation for this struct was generated from the following file:

• /home/runner/work/abacus-develop/abacus-develop/source/source_base/module_container/ATen/kernels/lapack.h