mul_inplace_op

→TensorImplPtr

int mul_inplace_op(const int & a, const int & b)

In-place elementwise multiplication: writes $a \cdot b$ back into a.

Math

$a_i \leftarrow a_i \cdot b_i$

See Also

mul_op — out-of-place counterpart.

aTensorImplPtr

Left operand and in-place target. Must be non-null and must not share storage with a view.

bTensorImplPtr

Right operand. Must be non-null and broadcastable into a->shape().

TensorImplPtr

The same pointer as a, now holding the product storage.

ShapeMismatch

If the broadcast output would change a's shape.

RuntimeError

If a shares storage with a view.

Common usage: scaling gradients (grad_.mul_(scale)) or applying element-wise masks (out_.mul_(mask)) without allocating a new buffer per step.