View source on GitHub |
A helper gate for implementing reversible classical arithmetic.
Inherits From: Gate
Used in the notebooks
Used in the tutorials |
---|
Child classes must override the registers
, with_registers
, and apply
methods.
This class handles the details of ensuring that the scaling of implementing
the gate is O(2^n) instead of O(4^n) where n is the number of qubits
being acted on, by implementing an _apply_unitary_
function in terms of
the registers and the apply function of the child class.
Examples:
class Add(cirq.ArithmeticGate):
def __init__(
self,
target_register: '[int, Sequence[int]]',
input_register: 'Union[int, Sequence[int]]',
):
self.target_register = target_register
self.input_register = input_register
def registers(self) -> 'Sequence[Union[int, Sequence[int]]]':
return self.target_register, self.input_register
def with_registers(
self, *new_registers: 'Union[int, Sequence[int]]'
) -> 'Add':
return Add(*new_registers)
def apply(self, *register_values: int) -> 'Union[int, Iterable[int]]':
return sum(register_values)
cirq.unitary(
Add(target_register=[2, 2],
input_register=1).on(*cirq.LineQubit.range(2))
).astype(np.int32)
array([[0, 0, 0, 1],
[1, 0, 0, 0],
[0, 1, 0, 0],
[0, 0, 1, 0]], dtype=int32)
c = cirq.Circuit(
cirq.X(cirq.LineQubit(3)),
cirq.X(cirq.LineQubit(2)),
cirq.X(cirq.LineQubit(6)),
cirq.measure(*cirq.LineQubit.range(4, 8), key='before_in'),
cirq.measure(*cirq.LineQubit.range(4), key='before_out'),
Add(target_register=[2] * 4,
input_register=[2] * 4).on(*cirq.LineQubit.range(8)),
cirq.measure(*cirq.LineQubit.range(4, 8), key='after_in'),
cirq.measure(*cirq.LineQubit.range(4), key='after_out'),
)
cirq.sample(c).data
before_in before_out after_in after_out
0 2 3 2 5
Methods
apply
@abc.abstractmethod
apply( *register_values ) -> Union[int, Iterable[int]]
Returns the result of the gate operating on classical values.
For example, an addition takes two values (the target and the source), adds the source into the target, then returns the target and source as the new register values.
The apply
method is permitted to be sloppy in three ways:
- The
apply
method is permitted to return values that have more bits than the registers they will be stored into. The extra bits are simply dropped. For example, if the value 5 is returned for a 2 qubit register then 5 % 22 = 1 will be used instead. Negative values are also permitted. For example, for a 3 qubit register the value -2 becomes -2 % 23 = 6. - When the value of the last
k
registers is not changed by the gate, theapply
method is permitted to omit these values from the result. That is to say, when the length of the output is less than the length of the input, it is padded up to the intended length by copying from the same position in the input. - When only the first register's value changes, the
apply
method is permitted to return anint
instead of a sequence of ints.
The apply
method must be reversible. Otherwise the gate will
not be unitary, and incorrect behavior will result.
Examples | |
---|---|
A fully detailed adder:
The same adder, with less boilerplate due to the details being
handled by the
|
controlled
controlled(
num_controls: Optional[int] = None,
control_values: Optional[Union[cv.AbstractControlValues, Sequence[Union[int, Collection[int]]]]
] = None,
control_qid_shape: Optional[Tuple[int, ...]] = None
) -> 'Gate'
Returns a controlled version of this gate. If no arguments are specified, defaults to a single qubit control.
Args | |
---|---|
num_controls
|
Total number of control qubits. |
control_values
|
Which control computational basis state to apply the
sub gate. A sequence of length num_controls where each
entry is an integer (or set of integers) corresponding to the
computational basis state (or set of possible values) where that
control is enabled. When all controls are enabled, the sub gate is
applied. If unspecified, control values default to 1.
|
control_qid_shape
|
The qid shape of the controls. A tuple of the
expected dimension of each control qid. Defaults to
(2,) * num_controls . Specify this argument when using qudits.
|
Returns | |
---|---|
A cirq.Gate representing self controlled by the given control values
and qubits. This is a cirq.ControlledGate in the base
implementation, but subclasses may return a different gate type.
|
num_qubits
num_qubits() -> int
The number of qubits this gate acts on.
on
on(
*qubits
) -> 'Operation'
Returns an application of this gate to the given qubits.
Args | |
---|---|
*qubits
|
The collection of qubits to potentially apply the gate to. |
Returns: a cirq.Operation
which is this gate applied to the given
qubits.
on_each
on_each(
*targets
) -> List['cirq.Operation']
Returns a list of operations applying the gate to all targets.
Args | |
---|---|
*targets
|
The qubits to apply this gate to. For single-qubit gates
this can be provided as varargs or a combination of nested
iterables. For multi-qubit gates this must be provided as an
Iterable[Sequence[Qid]] , where each sequence has num_qubits
qubits.
|
Returns | |
---|---|
Operations applying this gate to the target qubits. |
Raises | |
---|---|
ValueError
|
If targets are not instances of Qid or Iterable[Qid]. If the gate qubit number is incompatible. |
TypeError
|
If a single target is supplied and it is not iterable. |
registers
@abc.abstractmethod
registers() -> Sequence[Union[int, Sequence[int]]]
The data acted upon by the arithmetic gate.
Each register in the list can either be a classical constant (an int
),
or else a list of qubit/qudit dimensions. Registers that are set to a
classical constant must not be mutated by the arithmetic gate
(their value must remain fixed when passed to apply
).
Registers are big endian. The first qubit is the most significant, the last qubit is the 1s qubit, the before last qubit is the 2s qubit, etc.
Returns | |
---|---|
A list of constants and qubit groups that the gate will act upon. |
validate_args
validate_args(
qubits: Sequence['cirq.Qid']
) -> None
Checks if this gate can be applied to the given qubits.
By default checks that:
- inputs are of type
Qid
- len(qubits) == num_qubits()
- qubit_i.dimension == qid_shape[i] for all qubits
Child classes can override. The child implementation should call
super().validate_args(qubits)
then do custom checks.
Args | |
---|---|
qubits
|
The sequence of qubits to potentially apply the gate to. |
Raises | |
---|---|
ValueError
|
The gate can't be applied to the qubits. |
with_probability
with_probability(
probability: 'cirq.TParamVal'
) -> 'cirq.Gate'
Creates a probabilistic channel with this gate.
Args | |
---|---|
probability
|
floating point value between 0 and 1, giving the probability this gate is applied. |
Returns | |
---|---|
cirq.RandomGateChannel that applies self with probability
probability and the identity with probability 1-p .
|
with_registers
@abc.abstractmethod
with_registers( *new_registers ) -> Self
Returns the same fate targeting different registers.
Args | |
---|---|
*new_registers
|
The new values that should be returned by the
registers method.
|
Returns | |
---|---|
An instance of the same kind of gate, but acting on different registers. |
wrap_in_linear_combination
wrap_in_linear_combination(
coefficient: Union[complex, float, int] = 1
) -> 'cirq.LinearCombinationOfGates'
Returns a LinearCombinationOfGates with this gate.
Args | |
---|---|
coefficient
|
number coefficient to use in the resulting
cirq.LinearCombinationOfGates object.
|
Returns | |
---|---|
cirq.LinearCombinationOfGates containing self with a
coefficient of coefficient .
|
__add__
__add__(
other: Union['Gate', 'cirq.LinearCombinationOfGates']
) -> 'cirq.LinearCombinationOfGates'
__call__
__call__(
*qubits, **kwargs
)
Call self as a function.
__mul__
__mul__(
other: Union[complex, float, int]
) -> 'cirq.LinearCombinationOfGates'
__neg__
__neg__() -> 'cirq.LinearCombinationOfGates'
__pow__
__pow__(
power
)
__rmul__
__rmul__(
other: Union[complex, float, int]
) -> 'cirq.LinearCombinationOfGates'
__sub__
__sub__(
other: Union['Gate', 'cirq.LinearCombinationOfGates']
) -> 'cirq.LinearCombinationOfGates'
__truediv__
__truediv__(
other: Union[complex, float, int]
) -> 'cirq.LinearCombinationOfGates'