warp.max

warp.max(a: Scalar, b: Scalar) → Scalar

Kernel

Differentiable

Compute the maximum value.

On float types, NaN is treated as missing (C fmax semantics): the operation returns the non-NaN operand when exactly one is NaN, and NaN only when both are NaN.

warp.max(

a: Vector[Scalar, Any],

b: Vector[Scalar, Any],

) → Vector[Scalar, Any]

Kernel

Differentiable

Compute the maximum value.

Returns:

The element-wise maximum of a and b.

warp.max(a: Vector[Scalar, Any]) → Scalar

Kernel

Differentiable

Compute the maximum value.

On float types, NaN elements are treated as missing (C fmax semantics); the reduction returns the largest non-NaN element, or NaN only if every element is NaN.

Returns:

The maximum element of a.