mm — ops | Lucid

mm

→Tensor

mm(input: Tensor, mat2: Tensor)

source edit

Implementing kernel

C++mm_opfree fn

Strict 2-D matrix multiplication.

Computes the product of two 2-D tensors of shapes (n, k) and (k, m). No broadcasting is performed; both operands must be exactly 2-D. For broadcasting or batched semantics use matmul / bmm.

Parameters

inputTensor

Left matrix of shape (n, k).

mat2Tensor or scalar

Right matrix of shape (k, m). The inner dimension must match input.

Returns

Tensor

Matrix product of shape (n, m).

Notes

Mathematical definition:

\mathbf{C}_{ij} = \sum_{r=1}^{k} \mathbf{A}_{ir}\,\mathbf{B}_{rj}

Autograd mirrors matmul but without any batch handling. Use this function when the rank assumption can be guaranteed for clearer error messages and slightly leaner dispatch.

Examples

>>> import lucid
>>> a = lucid.tensor([[1.0, 2.0]])
>>> b = lucid.tensor([[3.0], [4.0]])
>>> lucid.mm(a, b)
Tensor([[11.]])

mm