abs_op

Structs

Struct: Abs

Fields

Methods


fwd(arg0: Array) -> Array

Computes the absolute value of the input array element-wise.

Args

arg0: Array The input array.

Returns

Array - An array containing the absolute value of each element in the input array.

Examples:


a = Array([[1, 2], [3, 4]])
result = to_abs(a)
print(result)

Note: This function supports:

Automatic differentiation (forward and reverse modes).
Complex valued arguments.


jvp(primals: List[Array], tangents: List[Array]) -> Array

Computes the Jacobian-vector product for the absolute value function.

Args

primals: List[Array] A list containing the primal input array.
tangents: List[Array] A list containing the tangent vector.

Returns

Array - The Jacobian-vector product for the absolute value function.

Implements forward-mode automatic differentiation for the absolute value function.

Note: The Jacobian-vector product for the absolute value is computed as sign(x) * dx, where x is the primal input and dx is the tangent vector.


vjp(primals: List[Array], grad: Array, out: Array) -> List[Array]

Computes the vector-Jacobian product for the absolute value function.

Args

primals: List[Array] A list containing the primal input array.
grad: Array The gradient of the output with respect to some scalar function.
out: Array The output of the forward pass (unused in this function).

Returns

List[Array] - A list containing the gradient with respect to the input.

Implements reverse-mode automatic differentiation for the absolute value function.

Note: The vector-Jacobian product for the absolute value is computed as sign(x) * grad, where x is the primal input and grad is the incoming gradient.


unary_simd_op(arg0_real: SIMD[float32, nelts[::DType]().__mul__(2).__floordiv__(2)], arg0_imag: SIMD[float32, nelts[::DType]().__mul__(2).__floordiv__(2)]) -> Tuple[SIMD[float32, nelts[::DType]().__mul__(2).__floordiv__(2)], SIMD[float32, nelts[::DType]().__mul__(2).__floordiv__(2)]]

Low-level function to compute the absolute value of a complex number represented as SIMD vectors.

Args

arg0_real: SIMD[float32, nelts[::DType]().__mul__(2).__floordiv__(2)] The real part of the complex number.
arg0_imag: SIMD[float32, nelts[::DType]().__mul__(2).__floordiv__(2)] The imaginary part of the complex number.

Returns

Tuple[SIMD[float32, nelts[::DType]().__mul__(2).__floordiv__(2)], SIMD[float32, nelts[::DType]().__mul__(2).__floordiv__(2)]] - The real and imaginary parts of the absolute value of the complex number as a tuple.


__call__(mut curr: Array, args: List[Array])

Performs the forward pass for element-wise absolute value computation of an array.

Args

curr: Array The current array to store the result (modified in-place).
args: List[Array] A list containing the input array.

Computes the absolute value of each element in the input array and stores the result in the current array. Initializes the current array if not already set up.

Note: This function assumes that the shape and data of the args are already set up. If the current array (curr) is not initialized, it computes the shape based on the input array and sets up the data accordingly.

Functions

to_abs


to_abs(arg0: Array) -> Array

Computes the absolute value of the input array element-wise.

Args

arg0: Array The input array.

Returns

Array - An array containing the absolute value of each element in the input array.

Examples:


a = Array([[1, 2], [3, 4]])
result = to_abs(a)
print(result)

Note: This function supports:

Automatic differentiation (forward and reverse modes).
Complex valued arguments.