qp.ops.op_math.ChangeOpBasis2

class ChangeOpBasis2(compute_op, target_op, uncompute_op=None)[source]

Bases: CompositeOp2

Composite operator representing a compute-uncompute pattern of Operator2s.

Parameters:
  • compute_op (Operator2) – operator applied before the target

  • target_op (Operator2) – operator applied in the changed basis

  • uncompute_op (Operator2 or None) – Operator applied after the target. None uses the adjoint of compute_op.

Note

The factors are stored and iterated in matrix-product order: uncompute, target, compute.

Example

>>> op = qp.ops.ChangeOpBasis2(qp.H(0), qp.RZ(0.3, 0))
>>> op.compute_op
H(0)
>>> op.target_op
RZ(0.3, wires=[0])

arg_specs

The expected types for the arguments of an operator.

arguments

Dictionary mapping argument names to their values.

arithmetic_depth

Arithmetic depth of the operator.

batch_size

Batch size of the operator if it is used with broadcasted parameters.

compilable_argnames

The names of arguments that are treated as compilable static arguments.

compilable_args

Dictionary mapping compilable argument names to their values.

compute_op

partial(func, *args, **keywords) - new function with partial application of the given arguments and keywords.

data

Create data property

dynamic_argnames

The names of arguments that are treated as dynamic.

dynamic_args

Dictionary mapping dynamic argument names to their values.

eigendecomposition

Return the eigendecomposition of the matrix specified by the operator.

grad_method

grad_recipe

Compute 'grad_recipe' lazily.

has_adjoint

has_decomposition

has_diagonalizing_gates

bool(x) -> bool

has_fixed_sig

Whether the expected signature of an operator is fixed.

has_generator

has_matrix

has_overlapping_wires

Boolean expression that indicates if the factors have overlapping wires.

has_sparse_matrix

hybrid_argnames

The names of arguments that represent dynamic data wrapped in static structures (known as Pytrees).

hybrid_args

Dictionary mapping hybrid argument names to their values.

hyperparameters

Legacy Operator compatibility view of non-dynamic, non-wire constructor arguments.

is_fully_abstract

Whether this operator contains only abstract data.

is_verified_hermitian

Check if the product operator is hermitian.

name

Operator name.

ndim_params

Number of dimensions per trainable parameter of the operator.

num_params

Number of trainable parameters.

num_wires

Number of wires the operator acts on.

operands

The operators in matrix-product order.

overlapping_ops

Groups all operands of the composite operator that act on overlapping wires.

parameters

Legacy Operator compatibility view of dynamic numerical arguments as a list.

pauli_rep

A PauliSentence representation of the Operator, or None if it doesn't have one.

static_argnames

The names of arguments that are treated as static.

static_args

Dictionary mapping static argument names to their values.

target_op

partial(func, *args, **keywords) - new function with partial application of the given arguments and keywords.

uncompute_op

partial(func, *args, **keywords) - new function with partial application of the given arguments and keywords.

wire_argnames

The names of arguments corresponding to wires.

wire_args

Dictionary mapping wire argument names to their values.

wire_sizes

The expected number of wire labels for each wire argument.

wires

Wires that the operator acts on.

arg_specs = None

The expected types for the arguments of an operator. This attribute is optional—not setting it has no loss of functionality. If set, it can be used to perform automatic validation of an operators inputs during construction. Additionally, when defining decomposition rules for an operator, operator types with arg_specs that spans all the arguments with static types can be placed in the rules’ resources without needing to fully construct abstract operators.

Note

A type that is listed in ‘arg_specs’ says what an argument is allowed to be, not what it actually is. For example, if arg_specs contains Complex[-1, -1], the Operator can still be instantiated with a real float64 array, which will then be reported as complex even though it holds real data.

The decomposition graph goes by the reported type, so real and complex inputs will look like the same operator and share one rule. To let them decompose differently, leave the argument out of arg_specs and give each rule a register_condition that checks the type. For a concrete example see BasisRotation.

arguments

Dictionary mapping argument names to their values.

arithmetic_depth
batch_size

Batch size of the operator if it is used with broadcasted parameters.

The batch_size is determined based on ndim_params and the provided parameters for the operator. If (some of) the latter have an additional dimension, and this dimension has the same size for all parameters, its size is the batch size of the operator. If no parameter has an additional dimension, the batch size is None.

Returns:

Size of the parameter broadcasting dimension if present, else None.

Return type:

int or None

compilable_argnames = ()

The names of arguments that are treated as compilable static arguments. Compilable static arguments are a subset of static arguments—arguments whose concrete values are known when compiling the program. But, unlike static_argnames, they are lowered to the compiler intermediate representation, making their values visible to the compiler, which may be useful if the compiler needs to interact with such values. Such values may include numbers, strings, and lists, tuples, or dictionaries thereof. This feature is opt-in; if any static arguments are not guaranteed to be compilable, it is safer to place them in static_argnames.

Note

An operator can only specify static_argnames and hybrid_argnames, or compilable_argnames, but not both; if any static or hybrid arguments are not or cannot be lowered to the IR, then all static and hybrid arguments are assumed to not be lowerable.

compilable_args

Dictionary mapping compilable argument names to their values.

compute_op

partial(func, *args, **keywords) - new function with partial application of the given arguments and keywords.

data

Create data property

dynamic_argnames = ()

The names of arguments that are treated as dynamic. Dynamic arguments are numerical arguments whose concrete values may not be known at compile-time, such as rotation angles, matrices, etc., and may include trainable data. Additionally, dynamic arguments are the only ones used to infer batch_size, necessary for broadcasting support.

Inputs for these arguments must be scalars, arrays, or castable to arrays (e.g. multi- dimensional lists with homogenous shapes), and operator constructors will be responsible for “canonicalizing” them (such as casting a list input to an array). At compile-time, the concrete values of these arguments are assumed to be unknown—only their shapes and data types are known.

dynamic_args

Dictionary mapping dynamic argument names to their values.

eigendecomposition

Return the eigendecomposition of the matrix specified by the operator.

This method uses pre-stored eigenvalues for standard observables where possible and stores the corresponding eigenvectors from the eigendecomposition.

It transforms the input operator according to the wires specified.

Returns:

dictionary containing the eigenvalues and the

eigenvectors of the operator.

Return type:

dict[str, array]

grad_method = None
grad_recipe

Compute ‘grad_recipe’ lazily.

has_adjoint = True
has_decomposition = True
has_diagonalizing_gates
has_fixed_sig = False

Whether the expected signature of an operator is fixed. If True, then the operator’s signature will always be fully known. When defining decomposition rules for an operator, operator types with fixed signatures can be placed in the rules’ resources without needing to fully construct abstract operators. This is set automatically when arg_specs covers every dynamic and wire argument, there are no hybrid, static, or compilable arguments, and every declared type is fully fixed (no unknown array shapes or wire counts).

has_generator = False
has_matrix = False
has_overlapping_wires

Boolean expression that indicates if the factors have overlapping wires.

has_sparse_matrix = False
hybrid_argnames = ('compute_op', 'target_op', 'uncompute_op')

The names of arguments that represent dynamic data wrapped in static structures (known as Pytrees). Names in this category must be disjoint from dynamic_argnames, static_argnames, and compilable_argnames, but may overlap with wire_argnames when those arguments contain nested structures of wires. Examples of hybrid arguments include collections of wires or dynamic arrays, operators, etc.

Note

An operator can only specify static_argnames and hybrid_argnames, or compilable_argnames, but not both; if any static or hybrid arguments are not or cannot be lowered to the IR, then all static and hybrid arguments are assumed to not be lowerable.

hybrid_args

Dictionary mapping hybrid argument names to their values.

hyperparameters

Legacy Operator compatibility view of non-dynamic, non-wire constructor arguments.

is_fully_abstract
is_verified_hermitian

Check if the product operator is hermitian.

Note, this check is not exhaustive. There can be hermitian operators for which this check yields false, which ARE hermitian. So a false result only implies that a more explicit check must be performed.

name

Operator name.

ndim_params

Number of dimensions per trainable parameter of the operator.

By default, this property returns the numbers of dimensions of the parameters used for the operator creation. If the parameter sizes for an operator subclass are fixed, this property can be overwritten to return the fixed value.

Returns:

Number of dimensions for each trainable parameter.

Return type:

tuple

num_params
num_wires

Number of wires the operator acts on.

operands

The operators in matrix-product order.

overlapping_ops

Groups all operands of the composite operator that act on overlapping wires.

Returns:

List of lists of operators that act on overlapping wires. All the inner lists commute with each other.

Return type:

List[List[Operator]]

parameters

Legacy Operator compatibility view of dynamic numerical arguments as a list.

pauli_rep

A PauliSentence representation of the Operator, or None if it doesn’t have one.

static_argnames = ()

The names of arguments that are treated as static. Static arguments are those whose concrete values are known when compiling the program. Arguments specified here are not lowered to a compiler intermediate representation (IR). Alternatively, if an argument is of a type that can be lowered, it can be moved to compilable_argnames.

Note

An operator can only specify static_argnames and hybrid_argnames, or compilable_argnames, but not both; if any static or hybrid arguments are not or cannot be lowered to the IR, then all static and hybrid arguments are assumed to not be lowerable.

static_args

Dictionary mapping static argument names to their values.

target_op

partial(func, *args, **keywords) - new function with partial application of the given arguments and keywords.

uncompute_op

partial(func, *args, **keywords) - new function with partial application of the given arguments and keywords.

wire_argnames = ()

The names of arguments corresponding to wires. Values for these arguments are automatically wrapped in Wires objects by the Operator2 constructor. If an argument is expected to be a structure that wraps wires (known as pytrees), such as a list of wire registers, make sure to include its name in hybrid_argnames as well. Additionally, Operator2 does not descend into the pytree structure of such arguments, so subclasses must ensure every wire leaf inside a hybrid wire argument is already a Wires object before forwarding it to super().__init__.

The order in which names appear in wire_argnames determines the order in which their wires appear in op.wires (see Operator2.wires). For hybrid wire arguments, the contained Wires leaves are ordered by pytree traversal order. Wires contributed by arguments which are themselves Operator2 objects are appended after all wire_argnames wires. The special names "work_wires" and "work_wire" may be included in wire_argnames but their values are excluded from op.wires.

wire_args

Dictionary mapping wire argument names to their values.

wire_sizes = ()

The expected number of wire labels for each wire argument. If any wire arguments support an arbitrary size, None must be used. By default, all wire arguments are assumed to support arbitrary sizes. For hybrid wire arguments, the wire size must always be None. Specifying wire sizes allows for additional automatic validation to be implemented, but, specifying it is optional if such validation is not needed.

wires

Wires that the operator acts on.

By default, the returned Wires are collected from the operator’s arguments in the following order:

  1. For each name in wire_argnames (in declaration order):

    • If the name is not in hybrid_argnames, the canonical value of that argument is added.

    • If the name is in hybrid_argnames, every Wires leaf of the argument’s pytree is added in pytree traversal order.

  2. After all wire_argnames have been processed, for each name in hybrid_argnames that is not in wire_argnames, the wires of any Operator2 leaves inside that argument are appended to the end in pytree traversal order.

Duplicate wires are removed while preserving the first occurrence, so the final result is the ordered union of all wires above.

Note

By default, work wires are not included in op.wires. In particular, wire arguments named work_wires or work_wire are excluded.

Note

This property may be overridden by developers if the default behaviour is not desired, for reasons such as including work wires or changing the order of the wires.

Returns:

wires

Return type:

Wires

adjoint()

Create an operation that is the adjoint of this one.

compute_decomposition(*args, **kwargs)

Representation of the operator as a product of other operators (static method).

compute_diagonalizing_gates([operands, ...])

Sequence of gates that diagonalize the operator in the computational basis.

compute_eigvals([operands, _init_pauli_rep])

Return the eigenvalues of the specified operator.

compute_matrix(*args, **kwargs)

Representation of the operator as a canonical matrix in the computational basis (static method).

compute_sparse_matrix(*args[, format])

Representation of the operator as a sparse matrix in the computational basis (static method).

decomposition()

Representation of the operator as a product of other operators.

diagonalizing_gates()

Sequence of gates that diagonalize the operator in the computational basis.

eigvals()

Eigenvalues of the operator in the computational basis.

generator()

Generator of an operator that is in single-parameter-form.

label([decimals, base_label, cache])

How the composite operator is represented in diagrams and drawings.

map_wires(wire_map)

Returns a new operator with its wires changed according to the given wire map.

matrix([wire_order])

Representation of the operator as a matrix in the computational basis.

pow(z)

A list of new operators equal to this one raised to the given power.

queue([context])

Append the operator to the Operator queue.

simplify()

Reduce the depth of nested operators to the minimum.

sparse_matrix([wire_order, format])

Representation of the operator as a sparse matrix in the computational basis.

terms()

Representation of the operator as a linear combination of other operators.

adjoint()[source]

Create an operation that is the adjoint of this one. Used to simplify Adjoint operators constructed by adjoint(). TODO: [sc-120844] Fix docstring after Adjoint is added

Adjointed operations are the conjugated and transposed version of the original operation. Adjointed ops are equivalent to the inverted operation for unitary gates.

Operator2.adjoint can be optionally defined by Operator developers, while adjoint() is the entry point for constructing generic adjoint representations.

Returns:

The adjointed operation.

>>> from pennylane.core.operator import Operator2
>>> class MyClass(Operator2):
...     dynamic_argnames = ("phi",)
...
...     def __init__(self, phi, wires):
...         super().__init__(phi, wires)
...
...     def adjoint(self):
...         return self
...
>>> op = MyClass(0.5, wires=0).adjoint()
>>> op
MyClass(0.5, wires=[0])
classmethod compute_decomposition(*args, **kwargs)

Representation of the operator as a product of other operators (static method).

\[O = O_1 O_2 \dots O_n.\]

Note

Operations making up the decomposition should be queued within the compute_decomposition method.

See also

decomposition().

Parameters:
  • *args (tuple) – Positional arguments of the operator

  • **kwargs (dict) – Keyword arguments of the operator

Returns:

decomposition of the operator

Return type:

list[Operator2]

static compute_diagonalizing_gates(operands=(), _init_pauli_rep=None)

Sequence of gates that diagonalize the operator in the computational basis.

Given the eigendecomposition \(O = U \Sigma U^{\dagger}\) where \(\Sigma\) is a diagonal matrix containing the eigenvalues, the sequence of diagonalizing gates implements the unitary \(U^{\dagger}\).

The diagonalizing gates rotate the state into the eigenbasis of the operator.

A DiagGatesUndefinedError is raised if no representation by decomposition is defined.

Returns:

a list of operators

Return type:

list[.Operator] or None

static compute_eigvals(operands=(), _init_pauli_rep=None)

Return the eigenvalues of the specified operator.

This method uses pre-stored eigenvalues for standard observables where possible and stores the corresponding eigenvectors from the eigendecomposition.

Returns:

array containing the eigenvalues of the operator

Return type:

array

static compute_matrix(*args, **kwargs)

Representation of the operator as a canonical matrix in the computational basis (static method).

The canonical matrix is the textbook matrix representation that does not consider wires. Implicitly, this assumes that the wires of the operator correspond to the global wire order.

Parameters:
  • *args (tuple) – Positional arguments of the operator

  • **kwargs (dict) – Keyword arguments of the operator

Returns:

matrix representation

Return type:

tensor_like

static compute_sparse_matrix(*args, format='csr', **kwargs)

Representation of the operator as a sparse matrix in the computational basis (static method).

The canonical matrix is the textbook matrix representation that does not consider wires. Implicitly, this assumes that the wires of the operator correspond to the global wire order.

See also

sparse_matrix()

Parameters:
  • *args (tuple) – Positional arguments of the operator

  • format (str) – format of the returned scipy sparse matrix, for example ‘csr’

  • **kwargs (dict) – Keyword arguments of the operator

Returns:

sparse matrix representation

Return type:

scipy.sparse._csr.csr_matrix

decomposition()

Representation of the operator as a product of other operators.

\[O = O_1 O_2 \dots O_n\]

A DecompositionUndefinedError is raised if no representation by decomposition is defined.

Returns:

decomposition of the operator

Return type:

list[Operator2]

diagonalizing_gates()[source]

Sequence of gates that diagonalize the operator in the computational basis.

Given the eigendecomposition \(O = U \Sigma U^{\dagger}\) where \(\Sigma\) is a diagonal matrix containing the eigenvalues, the sequence of diagonalizing gates implements the unitary \(U^{\dagger}\).

The diagonalizing gates rotate the state into the eigenbasis of the operator.

A DiagGatesUndefinedError is raised if no representation by decomposition is defined.

Returns:

a list of operators

Return type:

list[.Operator2] or None

eigvals()

Eigenvalues of the operator in the computational basis.

If diagonalizing_gates are specified and implement a unitary \(U^{\dagger}\), the operator can be reconstructed as

\[O = U \Sigma U^{\dagger},\]

where \(\Sigma\) is the diagonal matrix containing the eigenvalues.

Otherwise, no particular order for the eigenvalues is guaranteed.

Note

When eigenvalues are not explicitly defined, they are computed automatically from the matrix representation. Currently, this computation is not differentiable.

A EigvalsUndefinedError is raised if the eigenvalues have not been defined and cannot be inferred from the matrix representation.

Returns:

eigenvalues

Return type:

tensor_like

generator()

Generator of an operator that is in single-parameter-form.

For example, for operator

\[U(\phi) = e^{i\phi (0.5 Y + Z\otimes X)}\]

we get the generator

>>> U.generator()
0.5 * Y(0) + Z(0) @ X(1)

The generator may also be provided in the form of a dense or sparse Hamiltonian (using LinearCombination and SparseHamiltonian respectively).

label(decimals=None, base_label=None, cache=None)

How the composite operator is represented in diagrams and drawings.

Parameters:
  • decimals (int) – If None, no parameters are included. Else, how to round the parameters. Defaults to None.

  • base_label (Iterable[str]) – Overwrite the non-parameter component of the label. Must be same length as operands attribute. Defaults to None.

  • cache (dict) – Dictionary that carries information between label calls in the same drawing. Defaults to None.

Returns:

label to use in drawings

Return type:

str

Example (using the Sum composite operator)

>>> op = qp.S(0) + qp.X(0) + qp.Rot(1,2,3, wires=[1])
>>> op.label()
'𝓗'
map_wires(wire_map)[source]

Returns a new operator with its wires changed according to the given wire map.

Parameters:

wire_map (dict) – dictionary containing the old wires as keys and the new wires as values

Returns:

new operator

Return type:

.Operator2

matrix(wire_order=None)[source]

Representation of the operator as a matrix in the computational basis.

If wire_order is provided, the numerical representation considers the position of the operator’s wires in the global wire order. Otherwise, the wire order defaults to the operator’s wires.

If the matrix depends on trainable parameters, the result will be cast in the same autodifferentiation framework as the parameters.

A MatrixUndefinedError is raised if the matrix representation has not been defined.

See also

compute_matrix()

Parameters:

wire_order (Iterable) – global wire order, must contain all wire labels from the operator’s wires

Returns:

matrix representation

Return type:

tensor_like

pow(z)

A list of new operators equal to this one raised to the given power. This method is used to simplify Pow instances created by pow() or op ** power. TODO: [sc-120843] Fix docstring after Pow is added

Operator2.pow can be optionally defined by Operator developers, while pow() or op ** power are the entry point for constructing generic powers to exponents.

Parameters:

z (float) – exponent for the operator

Returns:

list[Operator2]

>>> from pennylane.core.operator import Operator2
>>> class MyClass(Operator2):
...     dynamic_argnames = ("phi",)
...
...     def __init__(self, phi, wires):
...         super().__init__(phi, wires)
...
...     def pow(self, z):
...         return [MyClass(self.phi*z, self.wires)]
...
>>> MyClass(0.5, 0).pow(2)
[MyClass(1.0, wires=[0])]
queue(context=<class 'pennylane.core.queuing.QueuingManager'>)

Append the operator to the Operator queue.

simplify()

Reduce the depth of nested operators to the minimum.

Returns:

simplified operator

Return type:

.Operator2

sparse_matrix(wire_order=None, format='csr')[source]

Representation of the operator as a sparse matrix in the computational basis.

If wire_order is provided, the numerical representation considers the position of the operator’s wires in the global wire order. Otherwise, the wire order defaults to the operator’s wires.

A SparseMatrixUndefinedError is raised if the sparse matrix representation has not been defined.

Parameters:
  • wire_order (Iterable) – global wire order, must contain all wire labels from the operator’s wires

  • format (str) – format of the returned scipy sparse matrix, for example ‘csr’

Returns:

sparse matrix representation

Return type:

scipy.sparse._csr.csr_matrix

terms()

Representation of the operator as a linear combination of other operators.

\[O = \sum_i c_i O_i\]

A TermsUndefinedError is raised if no representation by terms is defined.

Returns:

list of coefficients \(c_i\) and list of operations \(O_i\)

Return type:

tuple[list[tensor_like or float], list[.Operator2]]