BatchedBLAS.batched_dot!
— Methodbatched_dot!(o, x, y)
In-place batched vector-vector multiplication, equivalent to o[k] = transpose(x[:,k]) * y[:,k]
for all k
. All inputs can have eltypes of either AbstractFloats or Integers.
BatchedBLAS.batched_gemv!
— Methodbatched_gemv!(tA, alpha, A, x, beta, y)
In-place batched matrix-vector multiplication and addition, equivalent to y[:,k] = alpha[k] * A[:,:,k] * x[:,k] + beta[k] * y[:,k]
for all k
. A
can optionally be transposed with tA
as N
, T
, or C
. All other inputs can have eltypes of either AbstractFloats or Integers. alpha
and beta
can also be scalars.
BatchedBLAS.batched_ger!
— Methodbatched_ger!(alpha, x, y, A)
In-place rank-1 update of matrix A
with vectors x
and y
as alpha[k] * x[:,k] * transpose(y[:,k]) + A[:,:,k]
for all k
. All nputs can have eltypes of either AbstractFloats or Integers. alpha
can also be a scalar.
BatchedBLAS.batched_spmv!
— Methodbatched_spmv!(ul, alpha, A, x, beta, y)
In-place batched matrix-vector multiplication and addition, equivalent to y[:,k] = alpha[k] * A[:,:,k] * x[:,k] + beta[k] * y[:,k]
for all k
. A
must be packed symmetric, and uplo
specifies whether the upper ('U') or lower ('L') triangle was packed. All other inputs can have eltypes of either AbstractFloats or Integers. alpha
and beta
can also be scalars.
BatchedBLAS.batched_spr!
— Methodbatched_spr!(uplo, alpha, x, A)
In-place rank-1 update of packed symmetric matrix A
with vector x
as alpha[k] * x[:,k] * transpose(x[:,k]) + A[:,:,k]
for all k
. A
must be symmetric, and uplo
specifies whether the upper ('U') or lower ('L') triangle was packed. All other inputs can have eltypes of either AbstractFloats or Integers. alpha
can also be a scalar.
BatchedBLAS.batched_symv!
— Methodbatched_symv!(uplo, alpha, A, x, beta, y)
In-place batched matrix-vector multiplication and addition, equivalent to y[:,k] = alpha[k] * A[:,:,k] * x[:,k] + beta[k] * y[:,k]
for all k
. A
is assumed to be symmetric. Only the uplo
(either 'U' or 'L') triangle of A
is used. All other inputs can have eltypes of either AbstractFloats or Integers. alpha
and beta
can also be scalars.
BatchedBLAS.batched_syr!
— Methodbatched_syr!(uplo, alpha, x, A)
In-place rank-1 update of symmetric matrix A
with vector x
as alpha[k] * x[:,k] * transpose(x[:,k]) + A[:,:,k]
for all k
. A
is assumed to be symmetric. Only the uplo
(either 'U' or 'L') triangle of A
is used. All other inputs can have eltypes of either AbstractFloats or Integers. alpha
can also be a scalar.