hpr2 (USM Version)¶
Performs a rank-2 update of a Hermitian packed matrix.
Syntax
-
event
hpr2
(queue &exec_queue, uplo upper_lower, std::int64_t n, T alpha, const T *x, std::int64_t incx, const T *y, std::int64_t incy, T *a, const vector_class<event> &dependencies = {})¶
The USM version of hpr2
supports the following precisions and
devices.
T |
Devices Supported |
---|---|
|
Host, CPU, and GPU |
|
Host, CPU, and GPU |
Description
The hpr2 routines compute two scalar-vector-vector products and add them to a Hermitian packed matrix. The operation is defined as
A <- alpha*x*yH + conjg(alpha)*y*xH + A
where:
alpha
is a scalar,
A
is an n
-by-n
Hermitian matrix, supplied in packed form,
x
and y
are vectors of length n
.
Input Parameters
- exec_queue
The queue where the routine should be executed.
- upper_lower
Specifies whether A is upper or lower triangular. See Data Types for more details.
- n
Number of rows and columns of
A
. Must be at least zero.- alpha
Scaling factor for the matrix-vector product.
- x
Pointer to input vector
x
. The array holding input vectorx
must be of size at least (1 + (n
- 1)*abs(incx
)). See Matrix and Vector Storage for more details.- incx
Stride of vector
x
.- y
Pointer to input/output vector
y
. The array holding input/output vectory
must be of size at least (1 + (n
- 1)*abs(incy
)). See Matrix and Vector Storage for more details.- incy
Stride of vector
y
.- a
Pointer to input matrix
A
. The array holding input matrixA
must have size at least (n
*(n
-1))/2. See Matrix and Vector Storage for more details.The imaginary parts of the diagonal elements need not be set and are assumed to be zero.
- dependencies
List of events to wait for before starting computation, if any. If omitted, defaults to no dependencies.
Output Parameters
- a
Pointer to the updated upper triangularpart of the Hermitian matrix
A
ifupper_lower =upper
, or the updated lower triangular part of theHermitian matrixA
ifupper_lower =lower
.The imaginary parts of the diagonal elements are set tozero.
Return Values
Output event to wait on to ensure computation is complete.