quairkit.operator.gate.custom¶

The source file of the oracle class and the control oracle class.

class quairkit.operator.gate.custom.Oracle(oracle, system_idx, acted_system_dim=2, gate_info=None)¶

An oracle as a gate.

Parameters:¶

oracle : Tensor¶: Unitary oracle to be implemented.
system_idx : List[int] | int¶: Indices of the systems on which the gates are applied.
acted_system_dim : List[int] | int¶: dimension of systems that this gate acts on. Can be a list of system dimensions or an int representing the dimension of all systems. Defaults to be qubit case.

Note

The forward function of this class will not create a new instance of state.

property info : OperatorInfoType¶: Information of this gate.

class quairkit.operator.gate.custom.ControlOracle(oracle, system_idx, index, acted_system_dim=2, gate_info=None)¶

A controlled oracle as a gate.

Parameters:¶

oracle : Tensor¶: Unitary oracle to be implemented.
system_idx : List[int | List[int]]¶: Indices of the systems on which the gates are applied. The first element in the list is the control system, represented by a int or a list of ints, and the remaining elements represent the oracle system.
index : int¶: control index that activates the matrix.
acted_system_dim : List[int] | int¶: dimension of systems that this gate acts on. Can be a list of system dimensions or an int representing the dimension of all systems. Defaults to be qubit case.
gate_info : Dict | None¶: information of this gate that will be placed into the gate history or plotted by a Circuit. Defaults to None.

Note

The forward function of this class will not create a new instance of state.

property info : OperatorInfoType¶: Information of this gate.

class quairkit.operator.gate.custom.ParamOracle(generator, system_idx, param=None, num_acted_param=1, acted_system_dim=2, gate_info=None, support_batch=True)¶

An parameterized oracle as a gate

Parameters:¶

generator : Callable[[Tensor], Tensor]¶: function that generates the oracle.
system_idx : List[int] | int¶: indices of the system on which this gate acts on.
param : Tensor | float | List[float]¶: input parameters of quantum parameterized gates. Defaults to None i.e. randomized.
num_acted_param : int¶: the number of parameters required for a single operation.
acted_system_dim : List[int] | int¶: dimension of systems that this gate acts on. Can be a list of system dimensions or an int representing the dimension of all systems. Defaults to be qubit case.
gate_info : Dict¶: information of this gate that will be placed into the gate history or plotted by a Circuit. Defaults to None.
support_batch : bool¶: whether the generator supports batched input. Defaults to True.

Note

The forward function of this class will not create a new instance of state.

class quairkit.operator.gate.custom.ControlParamOracle(generator, system_idx, index, param=None, num_acted_param=1, acted_system_dim=2, gate_info=None, support_batch=True)¶

An parameterized oracle as a gate

Parameters:¶

generator : Callable[[Tensor], Tensor]¶: function that generates the oracle.
system_idx : List[int] | int¶: Indices of the systems on which the gates are applied. The first element in the list is the control system, represented by a int or a list of ints, and the remaining elements represent the oracle system.
index : int¶: control index that activates the matrix.
param : Tensor | float | List[float]¶: input parameters of quantum parameterized gates. Defaults to None i.e. randomized.
num_acted_param : int¶: the number of parameters required for a single operation.
acted_system_dim : List[int] | int¶: dimension of systems that this gate acts on. Can be a list of system dimensions or an int representing the dimension of all systems. Defaults to be qubit case.
gate_info : Dict¶: information of this gate that will be placed into the gate history or plotted by a Circuit. Defaults to None.
support_batch : bool¶: whether the generator supports batched input. Defaults to True.

Note

The forward function of this class will not create a new instance of state.

class quairkit.operator.gate.custom.UniversalQudits(system_idx, acted_system_dim, param=None, param_sharing=False, identity_init=False)¶

A collection of universal qudit gates. One of such a gate requires \(d^2 - 1\) parameters.

Parameters:¶

system_idx : Iterable[int] | str | None¶: Indices of the qubits on which the gates are applied. Defaults to the first qubit.
param : Tensor | float | None¶: Parameters of the gates. Defaults to None.
param_sharing : bool¶: Whether gates in the same layer share a parameter. Defaults to False.
acted_system_dim : Iterable[int]¶: dimension of systems that this gate acts on. Can be a list of system dimensions or an int representing the dimension of all systems. Defaults to be qubit case.
identity_init : bool¶: If True, initialize to identity matrix (all-zero Gell-Mann coefficients). Cannot be used together with param.

Note

The forward function of this class will not create a new instance of state.

class quairkit.operator.gate.custom.ManifoldUniversalQudits(system_idx, acted_system_dim, param=None, param_sharing=False, identity_init=False)¶

A universal qudit gate optimized on the unitary manifold via Riemannian gradient descent.

Instead of parametrizing unitaries via Gell-Mann generators (exponential map), this gate stores the unitary matrix directly and uses SVD projection (retraction) plus tangent-space gradient projection to stay on the unitary manifold during optimization.

Parameters:¶

system_idx : Iterable[int] | str | None¶: Indices of the systems on which the gates are applied.
acted_system_dim : Iterable[int]¶: Dimension of systems that this gate acts on. Must be a list of system dimensions (cannot be a single int).
param : Tensor | None¶: Initial unitary matrix(ces). If None, trainable with random (or identity if identity_init) init. If nn.Parameter, trainable with fixed init. If plain Tensor, not trainable with fixed init. When provided, must be reshapable to [N, B, d, d] where d = prod(acted_system_dim).
param_sharing : bool¶: Whether gates in the same layer share a parameter. Defaults to False.
identity_init : bool¶: If True, initialize to identity matrix instead of random. Cannot be used together with param.

Note

The forward function of this class will not create a new instance of state.

register_parameter(name, param)¶

Override to re-attach Riemannian grad hook when theta is replaced.

property matrix : Tensor¶: Return the unitary matrix after SVD projection to U(d).

class quairkit.operator.gate.custom.QFT(system_idx, acted_system_dim=2, is_dagger=False)¶

A collection of quantum Fourier transform (QFT) gates.

Parameters:¶

system_idx : List[int | List[int]] | int¶: Indices of the systems on which the gates are applied.
acted_system_dim : List[int] | int¶: Dimension of systems that this gate acts on. Can be a list of system dimensions or an int representing the dimension of all systems. Defaults to qubit case.
is_dagger : bool¶: Whether initialize this gate as inverse QFT. Defaults to False.

Note

The forward function of this class will not create a new instance of state.

property info : OperatorInfoType¶: Information of this gate.

class quairkit.operator.gate.custom.Permutation(perm, system_idx, acted_system_dim=2)¶

permutation matrix as a gate.

Parameters:¶

perm : List[int]¶: A list representing the permutation of subsystems. For example, [1, 0, 2] swaps the first two subsystems.
system_idx : List[int | List[int]]¶: Indices of the systems on which the gates are applied.
acted_system_dim : List[int] | int¶: dimension of systems that this gate acts on. Can be a list of system dimensions or an int representing the dimension of all systems. Defaults to be qubit case.

property info : OperatorInfoType¶: Information of this gate.