var

tripy.var(input: Tensor, dim: int | Sequence[int] | None = None, keepdim: bool = False, correction: int = 1) → Tensor

Returns a new tensor containing the variance of the elements of the input tensor along the specified dimension.

The variance along a dimension is defined as:

\(\sigma^2 = \Large \frac{1}{max(0, N - \text{correction})} \large \sum_{i=1}^N (x_i - \bar{x})^2\)

where \(N\) is the length of the dimension, \(x_i\) is the \(i^{th}\) element along the dimension, and \(\bar{x}\) is the mean.

Parameters:

• input (Tensor) – [dtype=T1] The input tensor.

• dim (int | Sequence[int] | None) – The dimension or dimensions along which to reduce. If this is not provided, all dimensions are reduced.

• keepdim (bool) – Whether to retain reduced dimensions in the output. If this is False, reduced dimensions will be squeezed.

• correction (int) – Defaults to Bessel’s correction.

Returns:

[dtype=T1] variance of the input tensor

Return type:

Tensor

TYPE CONSTRAINTS:

• T1: bfloat16, float16, float32

Example

Example

input = tp.reshape(tp.arange(6, dtype=tp.float32), (2, 3))
output = tp.var(input, dim=1, keepdim=True)

>>> input
tensor(
    [[0.0000, 1.0000, 2.0000],
     [3.0000, 4.0000, 5.0000]], 
    dtype=float32, loc=gpu:0, shape=(2, 3))
>>> output
tensor(
    [[1.0000],
     [1.0000]], 
    dtype=float32, loc=gpu:0, shape=(2, 1))