asdConvolve

Applicable Products

Function Description

API function

asdConvolve: performs one-dimensional filtering on a given signal.

Formula:

• asdConvolve formula: ,

  where w(k) is the element at the kth position of the output, u(j) is the one-dimensional signal at the jth position of the input, and v(k-j+1) is the filter convolution kernel at the (k-j+1)th position. The one-dimensional signal is a complex vector, and the filter convolution kernel is a real vector.

• Example:

  Input u:

  [[1.+1.j 2.+2.j]
   [1.+1.j 2.+2.j]]

  Input v:

   [1. 2. 3. 4.]

  After the asdConvolve operator is called, the output result is as follows:

  [[4.+4.j, 7.+7.j],
  [4.+4.j, 7.+7.j]]

Function Prototype

AspbStatus asdConvolve(const aclTensor *signal, const aclTensor *kernel, aclTensor *output, asdConvolveMode_t mode, void *stream, void *workspace)

Parameter Description

Return Value Description

For details about the return values, see Return Value.

Constraints

asdMul

• The supported CANN version is CANN 8.0 or later.
• The supported data types of the input signal are COMPLEX32 and COMPLEX64. The supported data types of the input filtering convolution kernel are FLOAT16 and FLOAT32.
• The supported output data types are COMPLEX32 and COMPLEX64.

• The shape of the operator input signal is [BatchCount, n]. The output shape is the same as the input shape. The shape of the input filtering convolution kernel is [k].

Calling Example

For details about the operator calling example, see asdConvolve_complex32 or asdConvolve_complex64.