For the complete documentation index, see llms.txt. Markdown versions of all pages are available by appending .md to any URL (e.g. /get-started.md).

Mojo function

shape_div

def shape_div[a_t: IntTuple, b_t: IntTuple](a: RuntimeTuple[a_t, element_type=a.element_type], b: RuntimeTuple[b_t, element_type=b.element_type]) -> RuntimeTuple[shape_div(a_t, b_t)]

Performs specialized shape division between RuntimeTuples.

This function implements a special division operation specifically designed for tensor shape calculations. Unlike standard division, it handles special cases:

1. If shapes are directly divisible (a % b == 0), returns a standard division (a // b)
2. If shapes are inversely divisible (b % a == 0), returns the signed reciprocal
3. If shapes are incompatible, aborts with an error

This operation is essential for transformations between tensor layouts and computing broadcasting semantics.

Parameters:

• ​a_t (IntTuple): Type of the first operand.
• ​b_t (IntTuple): Type of the second operand.

Args:

• ​a (RuntimeTuple[a_t, element_type=a.element_type]): The dividend RuntimeTuple.
• ​b (RuntimeTuple[b_t, element_type=b.element_type]): The divisor RuntimeTuple.

Returns:

RuntimeTuple[shape_div(a_t, b_t)]: A new RuntimeTuple containing the result of the shape division.