qp.capture.expand_plxpr_transforms¶

expand_plxpr_transforms(f)[source]¶

Function for applying transforms to plxpr.

Currently, when program capture is enabled, transforms are used as higher-order primitives. These primitives are present in the program, but their respective transform is not applied when a transformed function is called. expand_plxpr_transforms further “transforms” the input function to apply any transform primitives that are present in the program being run.

Example

In the below example, we can see that the qp.transforms.cancel_inverses transform has been applied to a function. However, the resulting program representation leaves the cancel_inverses transform as a primitive without actually transforming the program.

qp.capture.enable()

@qp.transforms.cancel_inverses
def circuit():
    qp.X(0)
    qp.S(1)
    qp.X(0)
    qp.adjoint(qp.S(1))
    return qp.expval(qp.Z(1))

>>> qp.capture.make_plxpr(circuit)()
{ lambda ; . let
    a:AbstractMeasurement(n_wires=None) = cancel_inverses_transform[
    args_slice=slice(0, 0, None)
    consts_slice=slice(0, 0, None)
    inner_jaxpr={ lambda ; . let
        _:AbstractOperator() = PauliX[n_wires=1] 0
        _:AbstractOperator() = S[n_wires=1] 1
        _:AbstractOperator() = PauliX[n_wires=1] 0
        b:AbstractOperator() = S[n_wires=1] 1
        _:AbstractOperator() = Adjoint b
        c:AbstractOperator() = PauliZ[n_wires=1] 1
        d:AbstractMeasurement(n_wires=None) = expval_obs c
      in (d,) }
    targs_slice=slice(0, None, None)
    tkwargs={}
    ]
  in (a,) }

To apply the transform, we can use expand_plxpr_transforms as follows:

>>> transformed_circuit = qp.capture.expand_plxpr_transforms(circuit)
>>> qp.capture.make_plxpr(transformed_circuit)()
{ lambda ; . let
    a:AbstractOperator() = PauliZ[n_wires=1] 1
    b:AbstractMeasurement(n_wires=None) = expval_obs a
  in (b,) }

As seen, the transform primitive is no longer present, but it has been applied to the original program, indicated by the inverse operators being cancelled.

Parameters:: f (Callable) – The callable to which any present transforms should be applied.
Returns:: Callable with transforms applied.
Return type:: Callable

code/api/pennylane.capture.expand_plxpr_transforms

Download Python script

Download Notebook

View on GitHub

qp.capture.expand_plxpr_transforms¶

Contents