Module hybridq.gate.property
Author: Salvatore Mandra (salvatore.mandra@nasa.gov)
Copyright © 2021, United States Government, as represented by the Administrator of the National Aeronautics and Space Administration. All rights reserved.
The HybridQ: A Hybrid Simulator for Quantum Circuits platform is licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at http://www.apache.org/licenses/LICENSE-2.0.
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.
Expand source code
"""
Author: Salvatore Mandra (salvatore.mandra@nasa.gov)
Copyright © 2021, United States Government, as represented by the Administrator
of the National Aeronautics and Space Administration. All rights reserved.
The HybridQ: A Hybrid Simulator for Quantum Circuits platform is licensed under
the Apache License, Version 2.0 (the "License"); you may not use this file
except in compliance with the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0.
Unless required by applicable law or agreed to in writing, software distributed
under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR
CONDITIONS OF ANY KIND, either express or implied. See the License for the
specific language governing permissions and limitations under the License.
"""
from __future__ import annotations
from hybridq.base import __Base__, generate
from hybridq.base import staticvars, compare, requires
from hybridq.base.property import Tags as TagGate, Name as NameGate, Params, Tuple, DocString
from scipy.linalg import expm, fractional_matrix_power as powm
from hybridq.utils import isintegral, isnumber
from copy import deepcopy
from hybridq.utils import sort
import numpy as np
def _truncate_print(x):
if x is None:
return ''
else:
if len(x) < 5:
return f'{x}'
else:
_str = '(' if isinstance(x, tuple) else '['
_str += ', '.join(str(x) for x in x[:2])
_str += ', ..., '
_str += ', '.join(str(x) for x in x[-2:])
_str += ')' if isinstance(x, tuple) else ']'
return _str
@compare('n_qubits,qubits')
@staticvars(
'n_qubits',
check=dict(n_qbits=(lambda n: n is any or (isintegral(n) and n >= 0),
"'n_qubits' must be a non-negative integer")))
class QubitGate(__Base__):
"""
Class representing a gate with qubits.
Attributes
----------
qubits: iter[any], optional
"""
def __init__(self, qubits: iter[any] = None, **kwargs) -> None:
# Call super
super().__init__(**kwargs)
# Set qubits
self._on(qubits)
@property
def qubits(self) -> tuple[any, ...]:
return self.__qubits
def __print__(self) -> dict[str, tuple[int, str, int]]:
return {
'n_qubits':
(10, f"n_qubits={self.n_qubits}" if self.qubits is None else '',
0),
'qubits':
(100, f"qubits={_truncate_print(self.qubits)}"
if self.qubits is not None and self.n_qubits > 0 else '', 0),
}
def _on(self, qubits: iter[any]) -> None:
"""
Set `qubits` to `QubitGate`.
"""
self.on(qubits, inplace=True)
def on(self,
qubits: iter[any] = None,
*,
inplace: bool = False) -> QubitGate:
"""
Return `QubitGate` applied to `qubits`. If `inplace` is `True`,
`QubitGate` is modified in place.
Parameters
----------
qubits: iter[any]
Qubits the new Gate will act on.
inplace: bool, optional
If `True`, `QubitGate` is modified in place. Otherwise, a new
`QubitGate` is returned.
Returns
-------
QubitGate
New `QubitGate` acting on `qubits`. If `inplace` is `True`,
`QubitGate` is modified in place.
Example
-------
>>> QubitGate([1, 2]).qubits
[1, 2]
>>> QubitGate().on([42]).qubits
[42]
"""
# Convert to tuple
qubits = None if qubits is None else tuple(qubits)
# Check that all qubits are hashable
if qubits and any(not getattr(q, '__hash__', None) for q in qubits):
raise ValueError("Only hashable 'qubits' are allowed.")
# Check that no qubits are repeated
if qubits is not None and len(qubits) != len(set(qubits)):
raise ValueError("Repeated qubits are not allowed.")
# Check lenght
if qubits is not None and len(qubits) != self.n_qubits:
raise ValueError(f"Wrong number of 'qubits' "
f"(expected {self.n_qubits}, got {len(qubits)})")
# Make a copy if needed
if inplace:
_g = self
else:
_g = deepcopy(self)
# Set
_g.__qubits = qubits
# Return
return _g
@compare('power')
class PowerGate(__Base__):
"""
Class representing a gate that can be raised to a given power.
Attributes
----------
power: any, optional
"""
def __init__(self, power: any = 1, **kwargs) -> None:
# Call super
super().__init__(**kwargs)
# Assign power to object
self._set_power(power)
@property
def power(self) -> any:
return self.__power
def __print__(self) -> dict[str, tuple[int, str, int]]:
_power = ''
if self.power != 1:
if any(
isinstance(self.power, t)
for t in (int, float, np.integer, np.floating)):
_power = f"**{np.round(self.power, 5)}"
else:
_power = f"**{self.power}"
return {'power': (0, f"{_power}", 1)}
def _set_power(self, power: any) -> None:
"""
Set `power` to `PowerGate`.
"""
self.set_power(power, inplace=True)
def set_power(self, power: any, *, inplace: bool = False) -> PowerGate:
"""
Return `PowerGate` to the given `power`. If `inplace` is `True`,
`PowerGate` is modified in place.
Parameters
----------
power: any
Power to elevate `PowerGate`.
inplace: bool, optional
If `True`, `PowerGate` is modified in place. Otherwise, a new
`PowerGate` is returned.
Returns
-------
PowerGate
New `PowerGate` to the given `power`. If `inplace` is `True`,
`PowerGate` is modified in place.
Example
-------
>>> PowerGate(U=[[1, 0], [0, -1]]).matrix()
array([[ 1, 0],
[ 0, -1]])
>>> PowerGate(U=[[1, 0], [0, -1]]).set_power(1.2345).matrix()
array([[ 1. +0.j , 0. +0.j ],
[ 0. +0.j , -0.74068735-0.67184987j]])
"""
# Make a copy if needed
if inplace:
_g = self
else:
_g = deepcopy(self)
# Assign qubits
_g.__power = 1 if power is None else power
return _g
def __pow__(self, p: any) -> PowerGate:
"""
Return `PowerGate`**p.
"""
return self.set_power(self.power * p, inplace=False)
def _inv(self) -> None:
"""
Modify `PowerGate` to its inverse.
"""
self.inv(inplace=True)
def inv(self, *, inplace: bool = False) -> PowerGate:
"""
Return inverse of `PowerGate`. If `inplace` is `True`, `PowerGate` is modified in place.
Parameters
----------
inplace: bool, optional
If `True`, PowerGate is modified in place. Otherwise, a new
`PowerGate` is returned.
Returns
-------
PowerGate
Inverse of `PowerGate`. If `True`, `PowerGate` is modified in place.
Example
-------
>>> g = PowerGate(U=[[1, 0], [0, np.exp(-0.23j)]])
>>> g.matrix()
array([[1. +0.j , 0. +0.j ],
[0. +0.j , 0.9736664-0.22797752j]])
>>> g.inv().matrix()
array([[1. -0.j , 0. -0.j ],
[0. -0.j , 0.9736664+0.22797752j]])
>>> g.inv().matrix() @ g.matrix()
array([[1.+0.j, 0.+0.j],
[0.+0.j, 1.+0.j]])
"""
return self.set_power(self.power * -1, inplace=inplace)
@compare('Matrix')
@staticvars('Matrix', transform=dict(Matrix=lambda Matrix: np.asarray(Matrix)))
class MatrixGate(__Base__):
"""
Class for gates that can be represented as a matrix.
"""
def __print__(self):
return {
'M': (
400,
f'M={type(self.Matrix).__module__}.{type(self.Matrix).__name__}(shape={self.Matrix.shape}, dtype={self.Matrix.dtype})',
0)
}
@requires('Matrix,qubits,n_qubits')
class PowerMatrixGate(PowerGate, __Base__):
"""
Class representing a single matrix gate.
"""
def __init__(self, *args, **kwargs):
# Initialize local variables
self.__conj = False
self.__T = False
# Continue
super().__init__(*args, **kwargs)
def __eq__(self, other: __Base__) -> bool:
# Check if gates are conjugated
_c1 = self.is_conjugated()
_c2 = other.is_conjugated() if other.provides(
'is_conjugated') else False
# Check if gates are transposed
_t1 = self.is_transposed()
_t2 = other.is_transposed() if other.provides(
'is_transposed') else False
# If either the conjugation or the transposition is different, the
# gates differ
if _c1 != _c2 or _t1 != _t2:
return False
# Otherwise, keep going
else:
return super().__eq__(other)
def __hash__(self) -> int:
return super().__hash__()
def __print__(self):
return {
'ct': (0, '^+' if self.__conj and self.__T else
'^T' if self.__T else '^*' if self.__conj else '', 2)
}
def _conj(self) -> PowerMatrixGate:
self.conj(inplace=True)
def conj(self, *, inplace: bool = False) -> PowerMatrixGate:
# Create a copy if needed
g = self if inplace else deepcopy(self)
# Change conjugation
g.__conj ^= True
# Return gate
return g
def _T(self) -> PowerMatrixGate:
self.T(inplace=True)
def T(self, *, inplace: bool = False) -> PowerMatrixGate:
# Create a copy if needed
g = self if inplace else deepcopy(self)
# Change transposition
g.__T ^= True
# Return gate
return g
def is_conjugated(self) -> bool:
return self.__conj
def is_transposed(self) -> bool:
return self.__T
def _adj(self) -> PowerMatrixGate:
self.adj(inplace=True)
def adj(self, *, inplace: bool = False) -> PowerMatrixGate:
# Create a copy if needed
g = self if inplace else deepcopy(self)
# Change transposition
g.__T ^= True
# Change conjugation
g.__conj ^= True
# Return gate
return g
def matrix(self, order: iter[any] = None) -> np.ndarray:
"""
Return matrix representing `MatrixPowerGate`. If `order` is provided,
the given order of qubits is used to output its matrix.
Parameters
----------
order: iter[any]
Order of qubits used to output the matrix.
Returns
-------
array_like
Matrix representing `MatrixPowerGate`.
Example
-------
>>> g = PowerMatrixGate(qubits=[0, 1], U=[[1, 0, 0, 0], [0, 1, 0, 0], [0, 0, 0, 1], [0, 0, 1, 0]])
array([[1, 0, 0, 0],
[0, 1, 0, 0],
[0, 0, 0, 1],
[0, 0, 1, 0]])
The order of qubits is `[1, 0]`. On the contrary:
>>> g.on().matrix(order=[1, 0])
array([[1, 0, 0, 0],
[0, 0, 0, 1],
[0, 0, 1, 0],
[0, 1, 0, 0]])
outputs a matrix with the qubits order being if `[0, 1]`.
"""
# Get Unitary
_U = np.asarray(self.Matrix)
if order is not None:
# Covert order to list
order = list(order)
# Order is allowed only if gate.qubits are specified
if self.qubits is None or sort(order) != sort(self.qubits):
raise ValueError(
"'order' is not a permutation of 'gate.qubits'.")
# Reorder matrix in case qubits are out-of-order
if order and order != self.qubits:
# Transpose
_U = np.reshape(
np.transpose(
np.reshape(_U, (2,) * (2 * self.n_qubits)),
[self.qubits.index(q) for q in order] +
[self.n_qubits + self.qubits.index(q) for q in order]),
(2**self.n_qubits, 2**self.n_qubits))
# Apply power
if self.power != 1:
_U = powm(_U, float(self.power))
# Apply conjugation
if self.__conj:
_U = _U.conj()
# Apply transposition
if self.__T:
_U = _U.T
# Return matrix
return _U
def isclose(self, gate: Gate, atol: float = 1e-8) -> bool:
"""
Determine if the matrix of `gate` is close within an absolute
tollerance. If the gates are acting on a different set of qubits,
`isclose` will return `False`.
Parameters
----------
gate: PowerMatrixGate
Gate to compare with.
atol: float, optional
Absolute tollerance.
Returns
-------
bool
`True` if the two gates are close withing the given absolute
tollerance, otherwise `False`.
Example
-------
>>> g1 = PowerMatrixGate(U = [[1, 2], [3, 4]])
>>> g2 = PowerMatrixGate(U = [[4, 5], [3, 4]])
>>> g1.isclose(g1)
True
>>> g1.isclose(g2)
False
>>> g1.on([3]).isclose(g1)
False
>>> g1.on([3]).isclose(g1.on([3])
True
"""
if not (self.qubits is None) ^ (gate.qubits is None):
# The gates differ if they act on a different set of qubits
if self.qubits is not None and sort(self.qubits) != sort(
gate.qubits):
return False
# Get unitaries
_U1 = self.matrix(order=self.qubits if self.qubits else None)
_U2 = gate.matrix(order=self.qubits if self.qubits else None)
# If either matrix does not exist, the two gates differ
return np.allclose(_U1, _U2, atol=atol)
else:
return False
def commutes_with(self, gate: PowerMatrixGate, atol: float = 1e-7) -> bool:
"""
Return `True` if the calling gate commutes with `gate`.
Parameters
----------
gate: PowerMatrixGate
Gate to check commutation with.
atol: float
Absolute tollerance.
Returns
-------
bool
`True` if the calling gate commutes with `gate`, otherwise `False`.
"""
from string import ascii_lowercase as alc, ascii_uppercase as auc
# Check both gates have qubits
if self.qubits is None or gate.qubits is None:
raise ValueError("Cannot check commutation between virtual gates.")
# Get shared qubits
shared_qubits = sort(set(self.qubits).intersection(gate.qubits))
# If no qubits are shared, the gates definitely commute
if not shared_qubits:
return True
# Rename
g1, g2 = self, gate
# Get all qubits
q12 = tuple(sort(set(g1.qubits + g2.qubits)))
# Get number of qubits
n12 = len(q12)
# Get unitaries
U1 = np.reshape(g1.matrix(), (2,) * 2 * g1.n_qubits)
U2 = np.reshape(g2.matrix(), (2,) * 2 * g2.n_qubits)
# Define how to multiply matrices
def _mul(w1, w2):
# Get qubits and unitaries
q1, U1 = w1
q2, U2 = w2
# Get number of qubits
n1 = len(q1)
n2 = len(q2)
# Construct map
_map = ''
_map += ''.join(alc[q12.index(q)] for q in q1)
_map += ''.join(auc[-shared_qubits.index(q) -
1 if q in shared_qubits else q12.index(q)]
for q in q1)
_map += ','
_map += ''.join(auc[-shared_qubits.index(q) -
1] if q in shared_qubits else alc[q12.index(q)]
for q in q2)
_map += ''.join(auc[q12.index(q)] for q in q2)
_map += '->'
_map += ''.join(alc[x] for x in range(n12))
_map += ''.join(auc[x] for x in range(n12))
# Multiply map
return np.einsum(_map, U1, U2)
# Compute products
P1 = _mul((g1.qubits, U1), (g2.qubits, U2))
P2 = _mul((g2.qubits, U2), (g1.qubits, U1))
# Check if the same
return np.allclose(P1, P2, atol=1e-5)
class UnitaryGate(PowerMatrixGate, skip_requirements=True):
def set_power(self, power: any, *, inplace: bool = False) -> UnitaryGate:
# If power is negative, just apply the absolute value and apply conj
# and T as well.
if isnumber(power) and power < 0:
if power != -1:
self = PowerMatrixGate.set_power(self, -power, inplace=inplace)
self = self.adj(inplace=inplace)
# Apply power
else:
self = PowerMatrixGate.set_power(self, power, inplace=inplace)
return self
def unitary(self, *args, **kwargs) -> np.ndarray:
"""
Alias for `self.matrix`.
"""
from hybridq.utils import DeprecationWarning
from warnings import warn
# Warn that `self.matrix` should be used instead of `self.unitary`
warn("Since '0.7.0', 'self.matrix' should be used instead "
"of the less general 'self.unitary'")
# Call self.matrix
return self.matrix(*args, **kwargs)
class SelfAdjointUnitaryGate(UnitaryGate, skip_requirements=True):
def conj(self, *, inplace: bool = False) -> SelfAdjointUnitaryGate:
"""
Apply conjugation to self.matrix().
"""
if self.power == 1 and not self.is_conjugated() and self.is_transposed(
):
return PowerMatrixGate.T(self, inplace=inplace)
else:
return PowerMatrixGate.conj(self, inplace=inplace)
def T(self, *, inplace: bool = False) -> SelfAdjointUnitaryGate:
"""
Apply transposition to self.matrix().
"""
if self.power == 1 and self.is_conjugated(
) and not self.is_transposed():
return PowerMatrixGate.conj(self, inplace=inplace)
else:
return PowerMatrixGate.T(self, inplace=inplace)
def adj(self, *, inplace: bool = False) -> SelfAdjointUnitaryGate:
"""
Apply adjunction to self.matrix().
"""
if self.power == 1:
return self if inplace else deepcopy(self)
else:
return PowerMatrixGate.adj(self, inplace=inplace)
@compare('Matrix_gen',
cmp=dict(Matrix_gen=lambda x, y: x.__code__ == y.__code__))
@staticvars('Matrix_gen',
check=dict(Matrix_gen=(lambda Matrix_gen: callable(Matrix_gen),
"'Matrix_gen' must be callable")))
class ParamGate(Params, n_params=any):
"""
Class representing a gate with qubits.
"""
def __init_subclass__(cls, **kwargs):
# Bind Matrix_gen to cls
cls.Matrix_gen = cls.__get_staticvar__('Matrix_gen')
# Continue
super().__init_subclass__(**kwargs)
def __setattr__(self, name, value):
if name == 'Matrix_gen':
raise AttributeError("Cannot set 'Matrix_gen'")
else:
super().__setattr__(name, value)
@property
def Matrix(self) -> np.ndarray:
if self.params is None:
raise ValueError("'params' must be provided.")
return np.asarray(self.Matrix_gen(*self.params))
@staticvars('RMatrix',
transform=dict(RMatrix=lambda RMatrix: np.asarray(RMatrix)))
class RotationGate(
ParamGate,
PowerGate,
__Base__,
n_params=1,
Matrix_gen=lambda self, r: expm(-1j * float(r) * self.RMatrix / 2)):
"""
Gate with form U = exp(-1j * r / 2 * O), with O an
arbitrary matrix.
"""
def __init_subclass__(cls, n_params=None, Matrix_gen=None, **kwargs):
# Initialize everything
super().__init_subclass__(n_params=cls._RotationGate__n_params,
Matrix_gen=cls._RotationGate__Matrix_gen,
**kwargs)
def __print__(self) -> dict[str, tuple[int, str, int]]:
_params = ''
if self.params:
from hybridq.utils import isnumber
if isnumber(self.params[0]):
_params = f'φ={np.round(self.params[0]/np.pi, 5)}π'
else:
_params = f'φ={self.params[0]}'
return {'params': (101, _params, 0)}
def set_params(self,
params: iter[any],
*,
inplace: bool = False) -> RotationGate:
try:
# Update parameters multiplying them by self.power
_g = ParamGate.set_params(
self, [(p * self.power) % (4 * np.pi) for p in params],
inplace=inplace)
# Set power to 1
_g = PowerGate.set_power(_g, 1, inplace=inplace)
# Return gate
return _g
except:
return ParamGate.set_params(self, params, inplace=inplace)
def set_power(self, power: any, *, inplace: bool = False) -> RotationGate:
try:
return self.set_params([p * power for p in self.params],
inplace=inplace)
except:
return PowerGate.set_power(self, power, inplace=inplace)
class CliffordGate(__Base__):
pass
@compare('apply', cmp=dict(apply=lambda x, y: x.__code__ == y.__code__))
@staticvars('apply',
check=dict(apply=(lambda f: callable(f), "'f' must be callable")))
class FunctionalGate(QubitGate, __Base__, n_qubits=any):
"""
`FunctionalGate` to manipulate state.
"""
def __init_subclass__(cls, **kwargs):
# Bind apply to cls
cls.apply = cls.__get_staticvar__('apply')
# Continue
super().__init_subclass__(**kwargs)
def __setattr__(self, name, value):
if name == 'apply':
raise AttributeError("Cannot set 'apply'")
else:
super().__setattr__(name, value)
def __call__(self, psi: np.ndarray, order: iter[any], **kwargs):
# Check qubits are specified
if self.qubits is None:
raise ValueError("'qubits' must be specified.")
# Convert order to tuple
order = tuple(order)
# Check qubits
if any(q not in order for q in self.qubits):
raise ValueError(
"'FunctionalGate' is expecting qubits not in 'order'.")
# Apply transformation (it should return new_psi, new_order)
return self.apply(psi, order, **kwargs)
class BaseTupleGate(Tuple):
"""
Gate defined as a tuple of gates.
"""
@property
def qubits(self) -> tuple[any, ...]:
from hybridq.utils import sort
# If empty, return empty tuple
if not len(self):
return tuple()
# Get all qubits
_qubits = tuple(
g.qubits if g.provides('qubits') else None for g in self)
# If any None is present, return None
if any(q is None for q in _qubits):
return None
# Flatten list and remove duplicates
else:
return tuple(sort(set(y for x in _qubits for y in x)))
@property
def n_qubits(self) -> int:
# Get qubits
qubits = self.qubits
return None if qubits is None else len(self.qubits)
def _gate_transform(gates):
# Get gates
try:
l_gates, r_gates = gates
l_gates = BaseTupleGate(l_gates)
r_gates = BaseTupleGate(r_gates)
except:
raise ValueError("'gates' must be a pair of lists of 'Gate's")
# Return gates
return (l_gates, r_gates)
def _s_transform(s):
# Get np.array. If s == None, set to 1
s = np.array(1) if s is None else np.asarray(s)
# If scalar or vector, just return
if s.ndim <= 1:
return s
# If diagonal, just return the diagonal
elif s.ndim == 2:
# Return
return _s_transform(
np.diag(s)) if s.shape[0] == s.shape[1] and np.allclose(
s, np.diag(np.diag(s))) else s
# Raise an implementation error
else:
raise NotImplementedError
@compare('gates,s,_conj_rgates')
@staticvars('gates,s,_conj_rgates,_use_cache',
_conj_rgates=False,
_use_cache=True,
transform=dict(gates=_gate_transform,
s=_s_transform,
_use_cache=lambda x: bool(x)),
check=dict(s=(lambda s: s is None or 0 <= s.ndim <= 2,
"'s' cannot have more than two dimensions.")))
class SchmidtGate(__Base__):
def __init_subclass__(cls, **kwargs):
# Get gates and s
l_gates, r_gates = cls.__get_staticvar__('gates')
s = cls.__get_staticvar__('s')
# Get number of gates
nl = len(l_gates)
nr = len(r_gates)
# Initialize error
_err = False
# If s is scalar, skip control
if s.ndim == 0:
if nr != nl:
_err = True
# s is a vector
elif s.ndim == 1:
if s.shape[0] != nl or (nr and nr != nl):
_err = True
# s is a matrix
elif s.ndim == 2:
if s.shape[0] != nl or s.shape[1] != (nr if nr else nl):
_err = True
# No other ndim are supported
else:
_err = True
# Raise if there is any error
if _err:
raise ValueError("'s' must be consistent with number of 'gates'")
# Continue
super().__init_subclass__(**kwargs)
def __print__(self):
return {
'gates':
(200,
f'gates={self.gates if len(self.gates[1]) else self.gates[0]}',
0),
's': (
400, f's={self.s}' if self.s.ndim == 0 else
f's={type(self.s).__module__}.{type(self.s).__name__}(shape={self.s.shape}, dtype={self.s.dtype})',
0)
}
def __reduce__(self,
*,
ignore_sdict: tuple[str, ...] = tuple(),
ignore_methods: tuple[str, ...] = tuple(),
ignore_keys: tuple[str, ...] = tuple()):
return super().__reduce__(ignore_sdict=ignore_sdict,
ignore_methods=ignore_methods,
ignore_keys=ignore_keys +
('_cached_hash', '_cached_Matrix'))
@property
def Matrix(self):
"""
Construct Matrix representing the Map. Order of qubits for `Matrix`
will be `SchmidtGate.gates[0].qubits + SchmidtGate.gates[1].qubits`,
even if left and right gates have overlapping qubits.
"""
from hybridq.gate import TupleGate, MatrixGate, NamedGate
from scipy.sparse import dok_matrix, diags
from hybridq.utils import sort
# Check if a cached value is already present. If yes, return it
if self._use_cache:
# Get cached hash
cached_hash = getattr(self, '_cached_hash', None)
# Get cached Matrix
cached_Matrix = getattr(self, '_cached_Matrix', None)
# Compute new hash
new_hash = hash(self)
# Return cached matrix
if new_hash == cached_hash and cached_Matrix is not None:
return cached_Matrix
# Get left and right gates
l_gates, r_gates = self.gates
# Cannot build map if qubits are not specified
if not (l_gates.qubits and r_gates.qubits):
raise ValueError(
"Cannot build 'Matrix' if 'qubits' are not specified.")
# Get order
order = tuple((0, q) for q in l_gates.qubits) + tuple(
(1, q) for q in r_gates.qubits)
# Convert to MatrixGate (to speedup calculation)
l_gates = TupleGate(
MatrixGate(U=g.matrix(), qubits=((0, q)
for q in g.qubits))
for g in l_gates)
r_gates = TupleGate(
MatrixGate(U=(g.conj() if self._conj_rgates else g).matrix(),
qubits=((1, q) for q in g.qubits)) for g in r_gates)
# Define how to merge gates
def _merge(l_g, r_g):
from hybridq.gate.utils import merge, pad
# Merge the two gates and pad to the right number of qubits
return pad(merge(l_g, r_g),
qubits=order,
order=order,
return_matrix_only=True)
# Get s
s = diags(
[self.s] * len(l_gates)).todok() if self.s.ndim == 0 else diags(
self.s).todok() if self.s.ndim == 1 else dok_matrix(self.s)
# Merge all gates
Matrix = np.sum(
[s * _merge(l_gates[x], r_gates[y]) for (x, y), s in s.items()],
axis=0)
# Save cache
if self._use_cache:
# Cache hash
self._cached_hash = new_hash
# Cache Matrix
self._cached_Matrix = Matrix
# Return Matrix
return Matrix
@requires('sample')
class StochasticGate(__Base__):
pass
@compare('gate,c_qubits')
@staticvars(
'gate,c_qubits',
transform=dict(c_qubits=lambda c_qubits: tuple(c_qubits)),
check=dict(
gate=(lambda gate: isinstance(gate, __Base__) and gate.n_qubits > 0,
f"Not a valid 'gate'.")))
class ControlledGate(__Base__):
def __init_subclass__(cls, **kwargs):
# Get gate and controlling qubits
gate = cls.__get_staticvar__('gate')
c_qubits = cls.__get_staticvar__('c_qubits')
# Check gate provides qubits and n_qubits
if any(not hasattr(gate, w) for w in ('qubits', 'n_qubits')):
raise ValueError(f"'{type(gate).__name__}' must "
"provide 'qubits' and 'n_qubits'.")
# Few checks
if gate.n_qubits == 0 or gate.qubits is None:
raise ValueError("'gate' must provide qubits.")
if len(c_qubits) == 0:
raise ValueError("'c_qubits' must non empty.")
if len(set(c_qubits)) != len(c_qubits):
raise ValueError("'c_qubits' cannot have repeated qubits.")
if set(c_qubits).intersection(gate.qubits):
raise ValueError(
"Controlled and controlling qubits must be different.")
# Continue
super().__init_subclass__(**kwargs)
@property
def n_qubits(self):
return len(self.qubits)
@property
def qubits(self):
return tuple(self.c_qubits) + tuple(self.gate.qubits)
def __print__(self):
return {
'n_qubits': (10, f"n_qubits={self.n_qubits}", 0),
'c_qubits': (99, f"c_qubits={_truncate_print(self.c_qubits)}", 0),
'qubits': (100, f"qubits={_truncate_print(self.qubits)}", 0),
'gate': (101, f'gate={self.gate}', 0),
}Classes
class BaseTupleGate (elements=(), tags=None, **kwargs)-
Gate defined as a tuple of gates.
Expand source code
class BaseTupleGate(Tuple): """ Gate defined as a tuple of gates. """ @property def qubits(self) -> tuple[any, ...]: from hybridq.utils import sort # If empty, return empty tuple if not len(self): return tuple() # Get all qubits _qubits = tuple( g.qubits if g.provides('qubits') else None for g in self) # If any None is present, return None if any(q is None for q in _qubits): return None # Flatten list and remove duplicates else: return tuple(sort(set(y for x in _qubits for y in x))) @property def n_qubits(self) -> int: # Get qubits qubits = self.qubits return None if qubits is None else len(self.qubits)Ancestors
Instance variables
var n_qubits : int-
Expand source code
@property def n_qubits(self) -> int: # Get qubits qubits = self.qubits return None if qubits is None else len(self.qubits) var qubits : tuple[any, ...]-
Expand source code
@property def qubits(self) -> tuple[any, ...]: from hybridq.utils import sort # If empty, return empty tuple if not len(self): return tuple() # Get all qubits _qubits = tuple( g.qubits if g.provides('qubits') else None for g in self) # If any None is present, return None if any(q is None for q in _qubits): return None # Flatten list and remove duplicates else: return tuple(sort(set(y for x in _qubits for y in x)))
Inherited members
class CliffordGate-
Basic features.
Expand source code
class CliffordGate(__Base__): passAncestors
- hybridq.base.base.__Base__
class ControlledGate-
Basic features.
Expand source code
class ControlledGate(__Base__): def __init_subclass__(cls, **kwargs): # Get gate and controlling qubits gate = cls.__get_staticvar__('gate') c_qubits = cls.__get_staticvar__('c_qubits') # Check gate provides qubits and n_qubits if any(not hasattr(gate, w) for w in ('qubits', 'n_qubits')): raise ValueError(f"'{type(gate).__name__}' must " "provide 'qubits' and 'n_qubits'.") # Few checks if gate.n_qubits == 0 or gate.qubits is None: raise ValueError("'gate' must provide qubits.") if len(c_qubits) == 0: raise ValueError("'c_qubits' must non empty.") if len(set(c_qubits)) != len(c_qubits): raise ValueError("'c_qubits' cannot have repeated qubits.") if set(c_qubits).intersection(gate.qubits): raise ValueError( "Controlled and controlling qubits must be different.") # Continue super().__init_subclass__(**kwargs) @property def n_qubits(self): return len(self.qubits) @property def qubits(self): return tuple(self.c_qubits) + tuple(self.gate.qubits) def __print__(self): return { 'n_qubits': (10, f"n_qubits={self.n_qubits}", 0), 'c_qubits': (99, f"c_qubits={_truncate_print(self.c_qubits)}", 0), 'qubits': (100, f"qubits={_truncate_print(self.qubits)}", 0), 'gate': (101, f'gate={self.gate}', 0), }Ancestors
- hybridq.base.base.__Base__
Instance variables
var n_qubits-
Expand source code
@property def n_qubits(self): return len(self.qubits) var qubits-
Expand source code
@property def qubits(self): return tuple(self.c_qubits) + tuple(self.gate.qubits)
class FunctionalGate (qubits: iter[any] = None, **kwargs)-
FunctionalGateto manipulate state.Expand source code
class FunctionalGate(QubitGate, __Base__, n_qubits=any): """ `FunctionalGate` to manipulate state. """ def __init_subclass__(cls, **kwargs): # Bind apply to cls cls.apply = cls.__get_staticvar__('apply') # Continue super().__init_subclass__(**kwargs) def __setattr__(self, name, value): if name == 'apply': raise AttributeError("Cannot set 'apply'") else: super().__setattr__(name, value) def __call__(self, psi: np.ndarray, order: iter[any], **kwargs): # Check qubits are specified if self.qubits is None: raise ValueError("'qubits' must be specified.") # Convert order to tuple order = tuple(order) # Check qubits if any(q not in order for q in self.qubits): raise ValueError( "'FunctionalGate' is expecting qubits not in 'order'.") # Apply transformation (it should return new_psi, new_order) return self.apply(psi, order, **kwargs)Ancestors
- QubitGate
- hybridq.base.base.__Base__
Subclasses
Inherited members
class MatrixGate-
Class for gates that can be represented as a matrix.
Expand source code
class MatrixGate(__Base__): """ Class for gates that can be represented as a matrix. """ def __print__(self): return { 'M': ( 400, f'M={type(self.Matrix).__module__}.{type(self.Matrix).__name__}(shape={self.Matrix.shape}, dtype={self.Matrix.dtype})', 0) }Ancestors
- hybridq.base.base.__Base__
class ParamGate (params: iter[any] = None, **kwargs)-
Class representing a gate with qubits.
Expand source code
class ParamGate(Params, n_params=any): """ Class representing a gate with qubits. """ def __init_subclass__(cls, **kwargs): # Bind Matrix_gen to cls cls.Matrix_gen = cls.__get_staticvar__('Matrix_gen') # Continue super().__init_subclass__(**kwargs) def __setattr__(self, name, value): if name == 'Matrix_gen': raise AttributeError("Cannot set 'Matrix_gen'") else: super().__setattr__(name, value) @property def Matrix(self) -> np.ndarray: if self.params is None: raise ValueError("'params' must be provided.") return np.asarray(self.Matrix_gen(*self.params))Ancestors
- Params
- hybridq.base.base.__Base__
Subclasses
Instance variables
var Matrix : numpy.ndarray-
Expand source code
@property def Matrix(self) -> np.ndarray: if self.params is None: raise ValueError("'params' must be provided.") return np.asarray(self.Matrix_gen(*self.params))
Inherited members
class PowerGate (power: any = 1, **kwargs)-
Class representing a gate that can be raised to a given power.
Attributes
power:any, optional
Expand source code
class PowerGate(__Base__): """ Class representing a gate that can be raised to a given power. Attributes ---------- power: any, optional """ def __init__(self, power: any = 1, **kwargs) -> None: # Call super super().__init__(**kwargs) # Assign power to object self._set_power(power) @property def power(self) -> any: return self.__power def __print__(self) -> dict[str, tuple[int, str, int]]: _power = '' if self.power != 1: if any( isinstance(self.power, t) for t in (int, float, np.integer, np.floating)): _power = f"**{np.round(self.power, 5)}" else: _power = f"**{self.power}" return {'power': (0, f"{_power}", 1)} def _set_power(self, power: any) -> None: """ Set `power` to `PowerGate`. """ self.set_power(power, inplace=True) def set_power(self, power: any, *, inplace: bool = False) -> PowerGate: """ Return `PowerGate` to the given `power`. If `inplace` is `True`, `PowerGate` is modified in place. Parameters ---------- power: any Power to elevate `PowerGate`. inplace: bool, optional If `True`, `PowerGate` is modified in place. Otherwise, a new `PowerGate` is returned. Returns ------- PowerGate New `PowerGate` to the given `power`. If `inplace` is `True`, `PowerGate` is modified in place. Example ------- >>> PowerGate(U=[[1, 0], [0, -1]]).matrix() array([[ 1, 0], [ 0, -1]]) >>> PowerGate(U=[[1, 0], [0, -1]]).set_power(1.2345).matrix() array([[ 1. +0.j , 0. +0.j ], [ 0. +0.j , -0.74068735-0.67184987j]]) """ # Make a copy if needed if inplace: _g = self else: _g = deepcopy(self) # Assign qubits _g.__power = 1 if power is None else power return _g def __pow__(self, p: any) -> PowerGate: """ Return `PowerGate`**p. """ return self.set_power(self.power * p, inplace=False) def _inv(self) -> None: """ Modify `PowerGate` to its inverse. """ self.inv(inplace=True) def inv(self, *, inplace: bool = False) -> PowerGate: """ Return inverse of `PowerGate`. If `inplace` is `True`, `PowerGate` is modified in place. Parameters ---------- inplace: bool, optional If `True`, PowerGate is modified in place. Otherwise, a new `PowerGate` is returned. Returns ------- PowerGate Inverse of `PowerGate`. If `True`, `PowerGate` is modified in place. Example ------- >>> g = PowerGate(U=[[1, 0], [0, np.exp(-0.23j)]]) >>> g.matrix() array([[1. +0.j , 0. +0.j ], [0. +0.j , 0.9736664-0.22797752j]]) >>> g.inv().matrix() array([[1. -0.j , 0. -0.j ], [0. -0.j , 0.9736664+0.22797752j]]) >>> g.inv().matrix() @ g.matrix() array([[1.+0.j, 0.+0.j], [0.+0.j, 1.+0.j]]) """ return self.set_power(self.power * -1, inplace=inplace)Ancestors
- hybridq.base.base.__Base__
Subclasses
Instance variables
var power :-
Expand source code
@property def power(self) -> any: return self.__power
Methods
def inv(self, *, inplace: bool = False) ‑> PowerGate-
Return inverse of
PowerGate. IfinplaceisTrue,PowerGateis modified in place.Parameters
inplace:bool, optional- If
True, PowerGate is modified in place. Otherwise, a newPowerGateis returned.
Returns
Example
>>> g = PowerGate(U=[[1, 0], [0, np.exp(-0.23j)]]) >>> g.matrix() array([[1. +0.j , 0. +0.j ], [0. +0.j , 0.9736664-0.22797752j]]) >>> g.inv().matrix() array([[1. -0.j , 0. -0.j ], [0. -0.j , 0.9736664+0.22797752j]]) >>> g.inv().matrix() @ g.matrix() array([[1.+0.j, 0.+0.j], [0.+0.j, 1.+0.j]])Expand source code
def inv(self, *, inplace: bool = False) -> PowerGate: """ Return inverse of `PowerGate`. If `inplace` is `True`, `PowerGate` is modified in place. Parameters ---------- inplace: bool, optional If `True`, PowerGate is modified in place. Otherwise, a new `PowerGate` is returned. Returns ------- PowerGate Inverse of `PowerGate`. If `True`, `PowerGate` is modified in place. Example ------- >>> g = PowerGate(U=[[1, 0], [0, np.exp(-0.23j)]]) >>> g.matrix() array([[1. +0.j , 0. +0.j ], [0. +0.j , 0.9736664-0.22797752j]]) >>> g.inv().matrix() array([[1. -0.j , 0. -0.j ], [0. -0.j , 0.9736664+0.22797752j]]) >>> g.inv().matrix() @ g.matrix() array([[1.+0.j, 0.+0.j], [0.+0.j, 1.+0.j]]) """ return self.set_power(self.power * -1, inplace=inplace) def set_power(self, power: any, *, inplace: bool = False) ‑> PowerGate-
Return
PowerGateto the givenpower. IfinplaceisTrue,PowerGateis modified in place.Parameters
power:any- Power to elevate
PowerGate. inplace:bool, optional- If
True,PowerGateis modified in place. Otherwise, a newPowerGateis returned.
Returns
Example
>>> PowerGate(U=[[1, 0], [0, -1]]).matrix() array([[ 1, 0], [ 0, -1]]) >>> PowerGate(U=[[1, 0], [0, -1]]).set_power(1.2345).matrix() array([[ 1. +0.j , 0. +0.j ], [ 0. +0.j , -0.74068735-0.67184987j]])Expand source code
def set_power(self, power: any, *, inplace: bool = False) -> PowerGate: """ Return `PowerGate` to the given `power`. If `inplace` is `True`, `PowerGate` is modified in place. Parameters ---------- power: any Power to elevate `PowerGate`. inplace: bool, optional If `True`, `PowerGate` is modified in place. Otherwise, a new `PowerGate` is returned. Returns ------- PowerGate New `PowerGate` to the given `power`. If `inplace` is `True`, `PowerGate` is modified in place. Example ------- >>> PowerGate(U=[[1, 0], [0, -1]]).matrix() array([[ 1, 0], [ 0, -1]]) >>> PowerGate(U=[[1, 0], [0, -1]]).set_power(1.2345).matrix() array([[ 1. +0.j , 0. +0.j ], [ 0. +0.j , -0.74068735-0.67184987j]]) """ # Make a copy if needed if inplace: _g = self else: _g = deepcopy(self) # Assign qubits _g.__power = 1 if power is None else power return _g
class PowerMatrixGate (*args, **kwargs)-
Class representing a single matrix gate.
Expand source code
class PowerMatrixGate(PowerGate, __Base__): """ Class representing a single matrix gate. """ def __init__(self, *args, **kwargs): # Initialize local variables self.__conj = False self.__T = False # Continue super().__init__(*args, **kwargs) def __eq__(self, other: __Base__) -> bool: # Check if gates are conjugated _c1 = self.is_conjugated() _c2 = other.is_conjugated() if other.provides( 'is_conjugated') else False # Check if gates are transposed _t1 = self.is_transposed() _t2 = other.is_transposed() if other.provides( 'is_transposed') else False # If either the conjugation or the transposition is different, the # gates differ if _c1 != _c2 or _t1 != _t2: return False # Otherwise, keep going else: return super().__eq__(other) def __hash__(self) -> int: return super().__hash__() def __print__(self): return { 'ct': (0, '^+' if self.__conj and self.__T else '^T' if self.__T else '^*' if self.__conj else '', 2) } def _conj(self) -> PowerMatrixGate: self.conj(inplace=True) def conj(self, *, inplace: bool = False) -> PowerMatrixGate: # Create a copy if needed g = self if inplace else deepcopy(self) # Change conjugation g.__conj ^= True # Return gate return g def _T(self) -> PowerMatrixGate: self.T(inplace=True) def T(self, *, inplace: bool = False) -> PowerMatrixGate: # Create a copy if needed g = self if inplace else deepcopy(self) # Change transposition g.__T ^= True # Return gate return g def is_conjugated(self) -> bool: return self.__conj def is_transposed(self) -> bool: return self.__T def _adj(self) -> PowerMatrixGate: self.adj(inplace=True) def adj(self, *, inplace: bool = False) -> PowerMatrixGate: # Create a copy if needed g = self if inplace else deepcopy(self) # Change transposition g.__T ^= True # Change conjugation g.__conj ^= True # Return gate return g def matrix(self, order: iter[any] = None) -> np.ndarray: """ Return matrix representing `MatrixPowerGate`. If `order` is provided, the given order of qubits is used to output its matrix. Parameters ---------- order: iter[any] Order of qubits used to output the matrix. Returns ------- array_like Matrix representing `MatrixPowerGate`. Example ------- >>> g = PowerMatrixGate(qubits=[0, 1], U=[[1, 0, 0, 0], [0, 1, 0, 0], [0, 0, 0, 1], [0, 0, 1, 0]]) array([[1, 0, 0, 0], [0, 1, 0, 0], [0, 0, 0, 1], [0, 0, 1, 0]]) The order of qubits is `[1, 0]`. On the contrary: >>> g.on().matrix(order=[1, 0]) array([[1, 0, 0, 0], [0, 0, 0, 1], [0, 0, 1, 0], [0, 1, 0, 0]]) outputs a matrix with the qubits order being if `[0, 1]`. """ # Get Unitary _U = np.asarray(self.Matrix) if order is not None: # Covert order to list order = list(order) # Order is allowed only if gate.qubits are specified if self.qubits is None or sort(order) != sort(self.qubits): raise ValueError( "'order' is not a permutation of 'gate.qubits'.") # Reorder matrix in case qubits are out-of-order if order and order != self.qubits: # Transpose _U = np.reshape( np.transpose( np.reshape(_U, (2,) * (2 * self.n_qubits)), [self.qubits.index(q) for q in order] + [self.n_qubits + self.qubits.index(q) for q in order]), (2**self.n_qubits, 2**self.n_qubits)) # Apply power if self.power != 1: _U = powm(_U, float(self.power)) # Apply conjugation if self.__conj: _U = _U.conj() # Apply transposition if self.__T: _U = _U.T # Return matrix return _U def isclose(self, gate: Gate, atol: float = 1e-8) -> bool: """ Determine if the matrix of `gate` is close within an absolute tollerance. If the gates are acting on a different set of qubits, `isclose` will return `False`. Parameters ---------- gate: PowerMatrixGate Gate to compare with. atol: float, optional Absolute tollerance. Returns ------- bool `True` if the two gates are close withing the given absolute tollerance, otherwise `False`. Example ------- >>> g1 = PowerMatrixGate(U = [[1, 2], [3, 4]]) >>> g2 = PowerMatrixGate(U = [[4, 5], [3, 4]]) >>> g1.isclose(g1) True >>> g1.isclose(g2) False >>> g1.on([3]).isclose(g1) False >>> g1.on([3]).isclose(g1.on([3]) True """ if not (self.qubits is None) ^ (gate.qubits is None): # The gates differ if they act on a different set of qubits if self.qubits is not None and sort(self.qubits) != sort( gate.qubits): return False # Get unitaries _U1 = self.matrix(order=self.qubits if self.qubits else None) _U2 = gate.matrix(order=self.qubits if self.qubits else None) # If either matrix does not exist, the two gates differ return np.allclose(_U1, _U2, atol=atol) else: return False def commutes_with(self, gate: PowerMatrixGate, atol: float = 1e-7) -> bool: """ Return `True` if the calling gate commutes with `gate`. Parameters ---------- gate: PowerMatrixGate Gate to check commutation with. atol: float Absolute tollerance. Returns ------- bool `True` if the calling gate commutes with `gate`, otherwise `False`. """ from string import ascii_lowercase as alc, ascii_uppercase as auc # Check both gates have qubits if self.qubits is None or gate.qubits is None: raise ValueError("Cannot check commutation between virtual gates.") # Get shared qubits shared_qubits = sort(set(self.qubits).intersection(gate.qubits)) # If no qubits are shared, the gates definitely commute if not shared_qubits: return True # Rename g1, g2 = self, gate # Get all qubits q12 = tuple(sort(set(g1.qubits + g2.qubits))) # Get number of qubits n12 = len(q12) # Get unitaries U1 = np.reshape(g1.matrix(), (2,) * 2 * g1.n_qubits) U2 = np.reshape(g2.matrix(), (2,) * 2 * g2.n_qubits) # Define how to multiply matrices def _mul(w1, w2): # Get qubits and unitaries q1, U1 = w1 q2, U2 = w2 # Get number of qubits n1 = len(q1) n2 = len(q2) # Construct map _map = '' _map += ''.join(alc[q12.index(q)] for q in q1) _map += ''.join(auc[-shared_qubits.index(q) - 1 if q in shared_qubits else q12.index(q)] for q in q1) _map += ',' _map += ''.join(auc[-shared_qubits.index(q) - 1] if q in shared_qubits else alc[q12.index(q)] for q in q2) _map += ''.join(auc[q12.index(q)] for q in q2) _map += '->' _map += ''.join(alc[x] for x in range(n12)) _map += ''.join(auc[x] for x in range(n12)) # Multiply map return np.einsum(_map, U1, U2) # Compute products P1 = _mul((g1.qubits, U1), (g2.qubits, U2)) P2 = _mul((g2.qubits, U2), (g1.qubits, U1)) # Check if the same return np.allclose(P1, P2, atol=1e-5)Ancestors
- PowerGate
- hybridq.base.base.__Base__
Subclasses
Methods
def T(self, *, inplace: bool = False) ‑> PowerMatrixGate-
Expand source code
def T(self, *, inplace: bool = False) -> PowerMatrixGate: # Create a copy if needed g = self if inplace else deepcopy(self) # Change transposition g.__T ^= True # Return gate return g def adj(self, *, inplace: bool = False) ‑> PowerMatrixGate-
Expand source code
def adj(self, *, inplace: bool = False) -> PowerMatrixGate: # Create a copy if needed g = self if inplace else deepcopy(self) # Change transposition g.__T ^= True # Change conjugation g.__conj ^= True # Return gate return g def commutes_with(self, gate: PowerMatrixGate, atol: float = 1e-07) ‑> bool-
Return
Trueif the calling gate commutes withgate.Parameters
gate:PowerMatrixGate- Gate to check commutation with.
atol:float- Absolute tollerance.
Returns
boolTrueif the calling gate commutes withgate, otherwiseFalse.
Expand source code
def commutes_with(self, gate: PowerMatrixGate, atol: float = 1e-7) -> bool: """ Return `True` if the calling gate commutes with `gate`. Parameters ---------- gate: PowerMatrixGate Gate to check commutation with. atol: float Absolute tollerance. Returns ------- bool `True` if the calling gate commutes with `gate`, otherwise `False`. """ from string import ascii_lowercase as alc, ascii_uppercase as auc # Check both gates have qubits if self.qubits is None or gate.qubits is None: raise ValueError("Cannot check commutation between virtual gates.") # Get shared qubits shared_qubits = sort(set(self.qubits).intersection(gate.qubits)) # If no qubits are shared, the gates definitely commute if not shared_qubits: return True # Rename g1, g2 = self, gate # Get all qubits q12 = tuple(sort(set(g1.qubits + g2.qubits))) # Get number of qubits n12 = len(q12) # Get unitaries U1 = np.reshape(g1.matrix(), (2,) * 2 * g1.n_qubits) U2 = np.reshape(g2.matrix(), (2,) * 2 * g2.n_qubits) # Define how to multiply matrices def _mul(w1, w2): # Get qubits and unitaries q1, U1 = w1 q2, U2 = w2 # Get number of qubits n1 = len(q1) n2 = len(q2) # Construct map _map = '' _map += ''.join(alc[q12.index(q)] for q in q1) _map += ''.join(auc[-shared_qubits.index(q) - 1 if q in shared_qubits else q12.index(q)] for q in q1) _map += ',' _map += ''.join(auc[-shared_qubits.index(q) - 1] if q in shared_qubits else alc[q12.index(q)] for q in q2) _map += ''.join(auc[q12.index(q)] for q in q2) _map += '->' _map += ''.join(alc[x] for x in range(n12)) _map += ''.join(auc[x] for x in range(n12)) # Multiply map return np.einsum(_map, U1, U2) # Compute products P1 = _mul((g1.qubits, U1), (g2.qubits, U2)) P2 = _mul((g2.qubits, U2), (g1.qubits, U1)) # Check if the same return np.allclose(P1, P2, atol=1e-5) def conj(self, *, inplace: bool = False) ‑> PowerMatrixGate-
Expand source code
def conj(self, *, inplace: bool = False) -> PowerMatrixGate: # Create a copy if needed g = self if inplace else deepcopy(self) # Change conjugation g.__conj ^= True # Return gate return g def is_conjugated(self) ‑> bool-
Expand source code
def is_conjugated(self) -> bool: return self.__conj def is_transposed(self) ‑> bool-
Expand source code
def is_transposed(self) -> bool: return self.__T def isclose(self, gate: Gate, atol: float = 1e-08) ‑> bool-
Determine if the matrix of
gateis close within an absolute tollerance. If the gates are acting on a different set of qubits,isclosewill returnFalse.Parameters
gate:PowerMatrixGate- Gate to compare with.
atol:float, optional- Absolute tollerance.
Returns
boolTrueif the two gates are close withing the given absolute tollerance, otherwiseFalse.
Example
>>> g1 = PowerMatrixGate(U = [[1, 2], [3, 4]]) >>> g2 = PowerMatrixGate(U = [[4, 5], [3, 4]]) >>> g1.isclose(g1) True >>> g1.isclose(g2) False >>> g1.on([3]).isclose(g1) False >>> g1.on([3]).isclose(g1.on([3]) TrueExpand source code
def isclose(self, gate: Gate, atol: float = 1e-8) -> bool: """ Determine if the matrix of `gate` is close within an absolute tollerance. If the gates are acting on a different set of qubits, `isclose` will return `False`. Parameters ---------- gate: PowerMatrixGate Gate to compare with. atol: float, optional Absolute tollerance. Returns ------- bool `True` if the two gates are close withing the given absolute tollerance, otherwise `False`. Example ------- >>> g1 = PowerMatrixGate(U = [[1, 2], [3, 4]]) >>> g2 = PowerMatrixGate(U = [[4, 5], [3, 4]]) >>> g1.isclose(g1) True >>> g1.isclose(g2) False >>> g1.on([3]).isclose(g1) False >>> g1.on([3]).isclose(g1.on([3]) True """ if not (self.qubits is None) ^ (gate.qubits is None): # The gates differ if they act on a different set of qubits if self.qubits is not None and sort(self.qubits) != sort( gate.qubits): return False # Get unitaries _U1 = self.matrix(order=self.qubits if self.qubits else None) _U2 = gate.matrix(order=self.qubits if self.qubits else None) # If either matrix does not exist, the two gates differ return np.allclose(_U1, _U2, atol=atol) else: return False def matrix(self, order: iter[any] = None) ‑> np.ndarray-
Return matrix representing
MatrixPowerGate. Iforderis provided, the given order of qubits is used to output its matrix.Parameters
order:iter[any]- Order of qubits used to output the matrix.
Returns
array_like- Matrix representing
MatrixPowerGate.
Example
>>> g = PowerMatrixGate(qubits=[0, 1], U=[[1, 0, 0, 0], [0, 1, 0, 0], [0, 0, 0, 1], [0, 0, 1, 0]]) array([[1, 0, 0, 0], [0, 1, 0, 0], [0, 0, 0, 1], [0, 0, 1, 0]])The order of qubits is
[1, 0]. On the contrary:>>> g.on().matrix(order=[1, 0]) array([[1, 0, 0, 0], [0, 0, 0, 1], [0, 0, 1, 0], [0, 1, 0, 0]])outputs a matrix with the qubits order being if
[0, 1].Expand source code
def matrix(self, order: iter[any] = None) -> np.ndarray: """ Return matrix representing `MatrixPowerGate`. If `order` is provided, the given order of qubits is used to output its matrix. Parameters ---------- order: iter[any] Order of qubits used to output the matrix. Returns ------- array_like Matrix representing `MatrixPowerGate`. Example ------- >>> g = PowerMatrixGate(qubits=[0, 1], U=[[1, 0, 0, 0], [0, 1, 0, 0], [0, 0, 0, 1], [0, 0, 1, 0]]) array([[1, 0, 0, 0], [0, 1, 0, 0], [0, 0, 0, 1], [0, 0, 1, 0]]) The order of qubits is `[1, 0]`. On the contrary: >>> g.on().matrix(order=[1, 0]) array([[1, 0, 0, 0], [0, 0, 0, 1], [0, 0, 1, 0], [0, 1, 0, 0]]) outputs a matrix with the qubits order being if `[0, 1]`. """ # Get Unitary _U = np.asarray(self.Matrix) if order is not None: # Covert order to list order = list(order) # Order is allowed only if gate.qubits are specified if self.qubits is None or sort(order) != sort(self.qubits): raise ValueError( "'order' is not a permutation of 'gate.qubits'.") # Reorder matrix in case qubits are out-of-order if order and order != self.qubits: # Transpose _U = np.reshape( np.transpose( np.reshape(_U, (2,) * (2 * self.n_qubits)), [self.qubits.index(q) for q in order] + [self.n_qubits + self.qubits.index(q) for q in order]), (2**self.n_qubits, 2**self.n_qubits)) # Apply power if self.power != 1: _U = powm(_U, float(self.power)) # Apply conjugation if self.__conj: _U = _U.conj() # Apply transposition if self.__T: _U = _U.T # Return matrix return _U
Inherited members
class QubitGate (qubits: iter[any] = None, **kwargs)-
Class representing a gate with qubits.
Attributes
qubits:iter[any], optional
Expand source code
class QubitGate(__Base__): """ Class representing a gate with qubits. Attributes ---------- qubits: iter[any], optional """ def __init__(self, qubits: iter[any] = None, **kwargs) -> None: # Call super super().__init__(**kwargs) # Set qubits self._on(qubits) @property def qubits(self) -> tuple[any, ...]: return self.__qubits def __print__(self) -> dict[str, tuple[int, str, int]]: return { 'n_qubits': (10, f"n_qubits={self.n_qubits}" if self.qubits is None else '', 0), 'qubits': (100, f"qubits={_truncate_print(self.qubits)}" if self.qubits is not None and self.n_qubits > 0 else '', 0), } def _on(self, qubits: iter[any]) -> None: """ Set `qubits` to `QubitGate`. """ self.on(qubits, inplace=True) def on(self, qubits: iter[any] = None, *, inplace: bool = False) -> QubitGate: """ Return `QubitGate` applied to `qubits`. If `inplace` is `True`, `QubitGate` is modified in place. Parameters ---------- qubits: iter[any] Qubits the new Gate will act on. inplace: bool, optional If `True`, `QubitGate` is modified in place. Otherwise, a new `QubitGate` is returned. Returns ------- QubitGate New `QubitGate` acting on `qubits`. If `inplace` is `True`, `QubitGate` is modified in place. Example ------- >>> QubitGate([1, 2]).qubits [1, 2] >>> QubitGate().on([42]).qubits [42] """ # Convert to tuple qubits = None if qubits is None else tuple(qubits) # Check that all qubits are hashable if qubits and any(not getattr(q, '__hash__', None) for q in qubits): raise ValueError("Only hashable 'qubits' are allowed.") # Check that no qubits are repeated if qubits is not None and len(qubits) != len(set(qubits)): raise ValueError("Repeated qubits are not allowed.") # Check lenght if qubits is not None and len(qubits) != self.n_qubits: raise ValueError(f"Wrong number of 'qubits' " f"(expected {self.n_qubits}, got {len(qubits)})") # Make a copy if needed if inplace: _g = self else: _g = deepcopy(self) # Set _g.__qubits = qubits # Return return _gAncestors
- hybridq.base.base.__Base__
Subclasses
- FunctionalGate
- hybridq.noise.channel.channel._MatrixChannel
Instance variables
var qubits : tuple[any, ...]-
Expand source code
@property def qubits(self) -> tuple[any, ...]: return self.__qubits
Methods
def on(self, qubits: iter[any] = None, *, inplace: bool = False) ‑> QubitGate-
Return
QubitGateapplied toqubits. IfinplaceisTrue,QubitGateis modified in place.Parameters
qubits:iter[any]- Qubits the new Gate will act on.
inplace:bool, optional- If
True,QubitGateis modified in place. Otherwise, a newQubitGateis returned.
Returns
Example
>>> QubitGate([1, 2]).qubits [1, 2] >>> QubitGate().on([42]).qubits [42]Expand source code
def on(self, qubits: iter[any] = None, *, inplace: bool = False) -> QubitGate: """ Return `QubitGate` applied to `qubits`. If `inplace` is `True`, `QubitGate` is modified in place. Parameters ---------- qubits: iter[any] Qubits the new Gate will act on. inplace: bool, optional If `True`, `QubitGate` is modified in place. Otherwise, a new `QubitGate` is returned. Returns ------- QubitGate New `QubitGate` acting on `qubits`. If `inplace` is `True`, `QubitGate` is modified in place. Example ------- >>> QubitGate([1, 2]).qubits [1, 2] >>> QubitGate().on([42]).qubits [42] """ # Convert to tuple qubits = None if qubits is None else tuple(qubits) # Check that all qubits are hashable if qubits and any(not getattr(q, '__hash__', None) for q in qubits): raise ValueError("Only hashable 'qubits' are allowed.") # Check that no qubits are repeated if qubits is not None and len(qubits) != len(set(qubits)): raise ValueError("Repeated qubits are not allowed.") # Check lenght if qubits is not None and len(qubits) != self.n_qubits: raise ValueError(f"Wrong number of 'qubits' " f"(expected {self.n_qubits}, got {len(qubits)})") # Make a copy if needed if inplace: _g = self else: _g = deepcopy(self) # Set _g.__qubits = qubits # Return return _g
class RotationGate (params: iter[any] = None, **kwargs)-
Gate with form U = exp(-1j * r / 2 * O), with O an arbitrary matrix.
Expand source code
class RotationGate( ParamGate, PowerGate, __Base__, n_params=1, Matrix_gen=lambda self, r: expm(-1j * float(r) * self.RMatrix / 2)): """ Gate with form U = exp(-1j * r / 2 * O), with O an arbitrary matrix. """ def __init_subclass__(cls, n_params=None, Matrix_gen=None, **kwargs): # Initialize everything super().__init_subclass__(n_params=cls._RotationGate__n_params, Matrix_gen=cls._RotationGate__Matrix_gen, **kwargs) def __print__(self) -> dict[str, tuple[int, str, int]]: _params = '' if self.params: from hybridq.utils import isnumber if isnumber(self.params[0]): _params = f'φ={np.round(self.params[0]/np.pi, 5)}π' else: _params = f'φ={self.params[0]}' return {'params': (101, _params, 0)} def set_params(self, params: iter[any], *, inplace: bool = False) -> RotationGate: try: # Update parameters multiplying them by self.power _g = ParamGate.set_params( self, [(p * self.power) % (4 * np.pi) for p in params], inplace=inplace) # Set power to 1 _g = PowerGate.set_power(_g, 1, inplace=inplace) # Return gate return _g except: return ParamGate.set_params(self, params, inplace=inplace) def set_power(self, power: any, *, inplace: bool = False) -> RotationGate: try: return self.set_params([p * power for p in self.params], inplace=inplace) except: return PowerGate.set_power(self, power, inplace=inplace)Ancestors
Methods
def Matrix_gen(self, r)-
Expand source code
Matrix_gen=lambda self, r: expm(-1j * float(r) * self.RMatrix / 2)):
Inherited members
class SchmidtGate-
Basic features.
Expand source code
class SchmidtGate(__Base__): def __init_subclass__(cls, **kwargs): # Get gates and s l_gates, r_gates = cls.__get_staticvar__('gates') s = cls.__get_staticvar__('s') # Get number of gates nl = len(l_gates) nr = len(r_gates) # Initialize error _err = False # If s is scalar, skip control if s.ndim == 0: if nr != nl: _err = True # s is a vector elif s.ndim == 1: if s.shape[0] != nl or (nr and nr != nl): _err = True # s is a matrix elif s.ndim == 2: if s.shape[0] != nl or s.shape[1] != (nr if nr else nl): _err = True # No other ndim are supported else: _err = True # Raise if there is any error if _err: raise ValueError("'s' must be consistent with number of 'gates'") # Continue super().__init_subclass__(**kwargs) def __print__(self): return { 'gates': (200, f'gates={self.gates if len(self.gates[1]) else self.gates[0]}', 0), 's': ( 400, f's={self.s}' if self.s.ndim == 0 else f's={type(self.s).__module__}.{type(self.s).__name__}(shape={self.s.shape}, dtype={self.s.dtype})', 0) } def __reduce__(self, *, ignore_sdict: tuple[str, ...] = tuple(), ignore_methods: tuple[str, ...] = tuple(), ignore_keys: tuple[str, ...] = tuple()): return super().__reduce__(ignore_sdict=ignore_sdict, ignore_methods=ignore_methods, ignore_keys=ignore_keys + ('_cached_hash', '_cached_Matrix')) @property def Matrix(self): """ Construct Matrix representing the Map. Order of qubits for `Matrix` will be `SchmidtGate.gates[0].qubits + SchmidtGate.gates[1].qubits`, even if left and right gates have overlapping qubits. """ from hybridq.gate import TupleGate, MatrixGate, NamedGate from scipy.sparse import dok_matrix, diags from hybridq.utils import sort # Check if a cached value is already present. If yes, return it if self._use_cache: # Get cached hash cached_hash = getattr(self, '_cached_hash', None) # Get cached Matrix cached_Matrix = getattr(self, '_cached_Matrix', None) # Compute new hash new_hash = hash(self) # Return cached matrix if new_hash == cached_hash and cached_Matrix is not None: return cached_Matrix # Get left and right gates l_gates, r_gates = self.gates # Cannot build map if qubits are not specified if not (l_gates.qubits and r_gates.qubits): raise ValueError( "Cannot build 'Matrix' if 'qubits' are not specified.") # Get order order = tuple((0, q) for q in l_gates.qubits) + tuple( (1, q) for q in r_gates.qubits) # Convert to MatrixGate (to speedup calculation) l_gates = TupleGate( MatrixGate(U=g.matrix(), qubits=((0, q) for q in g.qubits)) for g in l_gates) r_gates = TupleGate( MatrixGate(U=(g.conj() if self._conj_rgates else g).matrix(), qubits=((1, q) for q in g.qubits)) for g in r_gates) # Define how to merge gates def _merge(l_g, r_g): from hybridq.gate.utils import merge, pad # Merge the two gates and pad to the right number of qubits return pad(merge(l_g, r_g), qubits=order, order=order, return_matrix_only=True) # Get s s = diags( [self.s] * len(l_gates)).todok() if self.s.ndim == 0 else diags( self.s).todok() if self.s.ndim == 1 else dok_matrix(self.s) # Merge all gates Matrix = np.sum( [s * _merge(l_gates[x], r_gates[y]) for (x, y), s in s.items()], axis=0) # Save cache if self._use_cache: # Cache hash self._cached_hash = new_hash # Cache Matrix self._cached_Matrix = Matrix # Return Matrix return MatrixAncestors
- hybridq.base.base.__Base__
Instance variables
var Matrix-
Construct Matrix representing the Map. Order of qubits for
Matrixwill beSchmidtGate.gates[0].qubits + SchmidtGate.gates[1].qubits, even if left and right gates have overlapping qubits.Expand source code
@property def Matrix(self): """ Construct Matrix representing the Map. Order of qubits for `Matrix` will be `SchmidtGate.gates[0].qubits + SchmidtGate.gates[1].qubits`, even if left and right gates have overlapping qubits. """ from hybridq.gate import TupleGate, MatrixGate, NamedGate from scipy.sparse import dok_matrix, diags from hybridq.utils import sort # Check if a cached value is already present. If yes, return it if self._use_cache: # Get cached hash cached_hash = getattr(self, '_cached_hash', None) # Get cached Matrix cached_Matrix = getattr(self, '_cached_Matrix', None) # Compute new hash new_hash = hash(self) # Return cached matrix if new_hash == cached_hash and cached_Matrix is not None: return cached_Matrix # Get left and right gates l_gates, r_gates = self.gates # Cannot build map if qubits are not specified if not (l_gates.qubits and r_gates.qubits): raise ValueError( "Cannot build 'Matrix' if 'qubits' are not specified.") # Get order order = tuple((0, q) for q in l_gates.qubits) + tuple( (1, q) for q in r_gates.qubits) # Convert to MatrixGate (to speedup calculation) l_gates = TupleGate( MatrixGate(U=g.matrix(), qubits=((0, q) for q in g.qubits)) for g in l_gates) r_gates = TupleGate( MatrixGate(U=(g.conj() if self._conj_rgates else g).matrix(), qubits=((1, q) for q in g.qubits)) for g in r_gates) # Define how to merge gates def _merge(l_g, r_g): from hybridq.gate.utils import merge, pad # Merge the two gates and pad to the right number of qubits return pad(merge(l_g, r_g), qubits=order, order=order, return_matrix_only=True) # Get s s = diags( [self.s] * len(l_gates)).todok() if self.s.ndim == 0 else diags( self.s).todok() if self.s.ndim == 1 else dok_matrix(self.s) # Merge all gates Matrix = np.sum( [s * _merge(l_gates[x], r_gates[y]) for (x, y), s in s.items()], axis=0) # Save cache if self._use_cache: # Cache hash self._cached_hash = new_hash # Cache Matrix self._cached_Matrix = Matrix # Return Matrix return Matrix
class SelfAdjointUnitaryGate (*args, **kwargs)-
Class representing a single matrix gate.
Expand source code
class SelfAdjointUnitaryGate(UnitaryGate, skip_requirements=True): def conj(self, *, inplace: bool = False) -> SelfAdjointUnitaryGate: """ Apply conjugation to self.matrix(). """ if self.power == 1 and not self.is_conjugated() and self.is_transposed( ): return PowerMatrixGate.T(self, inplace=inplace) else: return PowerMatrixGate.conj(self, inplace=inplace) def T(self, *, inplace: bool = False) -> SelfAdjointUnitaryGate: """ Apply transposition to self.matrix(). """ if self.power == 1 and self.is_conjugated( ) and not self.is_transposed(): return PowerMatrixGate.conj(self, inplace=inplace) else: return PowerMatrixGate.T(self, inplace=inplace) def adj(self, *, inplace: bool = False) -> SelfAdjointUnitaryGate: """ Apply adjunction to self.matrix(). """ if self.power == 1: return self if inplace else deepcopy(self) else: return PowerMatrixGate.adj(self, inplace=inplace)Ancestors
- UnitaryGate
- PowerMatrixGate
- PowerGate
- hybridq.base.base.__Base__
Methods
def T(self, *, inplace: bool = False) ‑> SelfAdjointUnitaryGate-
Apply transposition to self.matrix().
Expand source code
def T(self, *, inplace: bool = False) -> SelfAdjointUnitaryGate: """ Apply transposition to self.matrix(). """ if self.power == 1 and self.is_conjugated( ) and not self.is_transposed(): return PowerMatrixGate.conj(self, inplace=inplace) else: return PowerMatrixGate.T(self, inplace=inplace) def adj(self, *, inplace: bool = False) ‑> SelfAdjointUnitaryGate-
Apply adjunction to self.matrix().
Expand source code
def adj(self, *, inplace: bool = False) -> SelfAdjointUnitaryGate: """ Apply adjunction to self.matrix(). """ if self.power == 1: return self if inplace else deepcopy(self) else: return PowerMatrixGate.adj(self, inplace=inplace) def conj(self, *, inplace: bool = False) ‑> SelfAdjointUnitaryGate-
Apply conjugation to self.matrix().
Expand source code
def conj(self, *, inplace: bool = False) -> SelfAdjointUnitaryGate: """ Apply conjugation to self.matrix(). """ if self.power == 1 and not self.is_conjugated() and self.is_transposed( ): return PowerMatrixGate.T(self, inplace=inplace) else: return PowerMatrixGate.conj(self, inplace=inplace)
Inherited members
class StochasticGate-
Basic features.
Expand source code
class StochasticGate(__Base__): passAncestors
- hybridq.base.base.__Base__
Subclasses
- hybridq.gate.gate._StochasticGate
class UnitaryGate (*args, **kwargs)-
Class representing a single matrix gate.
Expand source code
class UnitaryGate(PowerMatrixGate, skip_requirements=True): def set_power(self, power: any, *, inplace: bool = False) -> UnitaryGate: # If power is negative, just apply the absolute value and apply conj # and T as well. if isnumber(power) and power < 0: if power != -1: self = PowerMatrixGate.set_power(self, -power, inplace=inplace) self = self.adj(inplace=inplace) # Apply power else: self = PowerMatrixGate.set_power(self, power, inplace=inplace) return self def unitary(self, *args, **kwargs) -> np.ndarray: """ Alias for `self.matrix`. """ from hybridq.utils import DeprecationWarning from warnings import warn # Warn that `self.matrix` should be used instead of `self.unitary` warn("Since '0.7.0', 'self.matrix' should be used instead " "of the less general 'self.unitary'") # Call self.matrix return self.matrix(*args, **kwargs)Ancestors
- PowerMatrixGate
- PowerGate
- hybridq.base.base.__Base__
Subclasses
Methods
def unitary(self, *args, **kwargs) ‑> numpy.ndarray-
Alias for
self.matrix.Expand source code
def unitary(self, *args, **kwargs) -> np.ndarray: """ Alias for `self.matrix`. """ from hybridq.utils import DeprecationWarning from warnings import warn # Warn that `self.matrix` should be used instead of `self.unitary` warn("Since '0.7.0', 'self.matrix' should be used instead " "of the less general 'self.unitary'") # Call self.matrix return self.matrix(*args, **kwargs)
Inherited members