Module numeric 

Numeric utility functions for jit backends

Functions

add

Add two tensors

add_scalar

Add a tensor and a scalar

bitwise_and

Bitwise and two tensors

bitwise_and_scalar

Bitwise and with a scalar

bitwise_or

Bitwise or two tensors

bitwise_or_scalar

Bitwise or with a scalar

bitwise_xor

Bitwise xor two tensors

bitwise_xor_scalar

Bitwise xor with a scalar

cummax

Compute the cumulative maximum along a dimension

cummin

Compute the cumulative minimum along a dimension

cumprod

Compute the cumulative product along a dimension

cumsum

Compute the cumulative sum along a dimension

div

Divide two tensors

div_scalar

Divide a tensor by a scalar

empty_device

Creates a tensor with uninitialized memory

empty_device_dtype

Creates a tensor with uninitialized memory with the specific dtype.

empty_device_optimized

Create a tensor with uninitialized memory

empty_device_optimized_dtype

Create a tensor with uninitialized memory

full

Creates a tensor filled with value

full_client

Creates a tensor filled with value

full_device_dtype

Creates a tensor filled with value

mul

Multiply two tensors

mul_scalar

Multiply a tensor and a scalar

ones

Creates a tensor filled with ones

ones_client

Creates a tensor filled with ones

pow

Calculate the power of two tensors

remainder

Calculate remainder of two tensors

remainder_scalar

Calculate the remainder of a tensor with a scalar

sub

Subtract two tensors

sub_scalar

Subtract a tensor and a scalar

zeros

Creates a tensor filled with zeros

zeros_client

Creates a tensor filled with zeros