qml.math.unwrap

unwrap(values, max_depth=None)

Unwrap a sequence of objects to NumPy arrays.

Note that tensors on GPUs will automatically be copied to the CPU.

Parameters

• values (Sequence[tensor_like]) – sequence of tensor-like objects to unwrap

• max_depth (int) – Positive integer indicating the depth of unwrapping to perform for nested tensor-objects. This argument only applies when unwrapping Autograd ArrayBox objects.

Example

>>> values = [np.array([0.1, 0.2]), torch.tensor(0.1, dtype=torch.float64), torch.tensor([0.5, 0.2])]
>>> math.unwrap(values)
[array([0.1, 0.2]), 0.1, array([0.5, 0.2], dtype=float32)]

This function will continue to work during backpropagation:

>>> def cost_fn(params):
...     unwrapped_params = math.unwrap(params)
...     print("Unwrapped:", [(i, type(i)) for i in unwrapped_params])
...     return np.sum(np.sin(params))
>>> params = np.array([0.1, 0.2, 0.3])
>>> grad = autograd.grad(cost_fn)(params)
Unwrapped: [(0.1, <class 'numpy.float64'>), (0.2, <class 'numpy.float64'>), (0.3, <class 'numpy.float64'>)]
>>> print(grad)
[0.99500417 0.98006658 0.95533649]

code/api/pennylane.math.unwrap

 

Download Python script

 

Download Notebook

 

View on GitHub