tensorplay.functional.addcdiv

tensorplay.functional.addcdiv(input, tensor1, tensor2, *, value=1, out=None) → Tensor[source]

Performs the element-wise division of tensor1 by tensor2, multiplies the result by the scalar value and adds it to input.

Warning

Integer division with addcdiv is no longer supported, and in a future release addcdiv will perform a true division of tensor1 and tensor2. The historic addcdiv behavior can be implemented as (input + value * tensorplay.trunc(tensor1 / tensor2)).to(input.dtype) for integer inputs and as (input + value * tensor1 / tensor2) for float inputs. The future addcdiv behavior is just the latter implementation: (input + value * tensor1 / tensor2), for all dtypes.

$$\text{out}_i = \text{input}_i + \text{value} \times \frac{\text{tensor1}_i}{\text{tensor2}_i}$$

The shapes of input, tensor1, and tensor2 must be broadcastable.

For inputs of type FloatTensor or DoubleTensor, value must be a real number, otherwise an integer.

Parameters:

• input (Tensor) – the tensor to be added

• tensor1 (Tensor) – the numerator tensor

• tensor2 (Tensor) – the denominator tensor

Keyword Arguments:

• value (Number, optional) – multiplier for $\text{tensor1} / \text{tensor2}$

• out (Tensor, optional) – the output tensor.

Example:

>>> t = tensorplay.randn(1, 3)
>>> t1 = tensorplay.randn(3, 1)
>>> t2 = tensorplay.randn(1, 3)
>>> tensorplay.addcdiv(t, t1, t2, value=0.1)
tensor([[-0.2312, -3.6496,  0.1312],
        [-1.0428,  3.4292, -0.1030],
        [-0.5369, -0.9829,  0.0430]])