mindspore.ops.vecdot

mindspore.ops.vecdot(x, y, *, axis=- 1)

Calculates the dot product of two batches of vectors across the specified dimension.

The formula of calculation is as follows. $\overline{x_i}$ represents the conjugate for complex vectors, and $\overline{x_i}$ is the raw value for real vectors.

$$\sum _{i=1}^n\overline{x_i}{y}_i$$

Warning

This is an experimental API that is subject to change or deletion.

Parameters

• x (Tensor) – First batch of vectors. The shape of Tensor is $(\ast ,N)$ where $\ast$ means, any number of additional dimensions. Supporting broadcasting. The dtype of Tensor should be one of the following types: float, double, int, complex64 and complex128.

• y (Tensor) – Second batch of vectors. The shape of Tensor is $(\ast ,N)$ where $\ast$ means, any number of additional dimensions. Supporting broadcasting. The dtype of Tensor should be one of the following types: float, double, int, complex64 and complex128.

Keyword Arguments

axis (int) – Dimension across which to calculate the dot product. Default: -1 .

Returns

Tensor, the shape is almost same as the shape of Tensor after broadcasting, while the specified dimension axis in shape has been removed.

Raises

• TypeError – If x or y is not a Tensor.

• TypeError – If type of axis is not int.

• ValueError – If axis is out of range.

Supported Platforms:

Ascend GPU CPU

Note

Currently, complex numbers are not supported on GPU.

Examples

>>> import mindspore as ms
>>> from mindspore import ops
>>> x = ms.Tensor([[1, 3], [5, 7], [9, 8]], dtype=ms.float32)
>>> y = ms.Tensor([[4, 5], [6, 7], [3, 2]], dtype=ms.float32)
>>> output = ops.vecdot(x, y, axis=-1)
>>> print(output)
[19. 79. 43.]