def test_composed_op_to_circuit(self):
"""
Test if unitary ComposedOp transpile to circuit and represents expected operator.
Test if to_circuit on non-unitary ListOp raises exception.
"""
x = np.array([[0.0, 1.0], [1.0, 0.0]]) # Pauli X as numpy array
y = np.array([[0.0, -1.0j], [1.0j, 0.0]]) # Pauli Y as numpy array
m1 = np.array([[0, 0, 1, 0], [0, 0, 0, -1], [0, 0, 0, 0], [0, 0, 0, 0]]) # non-unitary
m2 = np.array([[0, 0, 0, 0], [0, 0, 0, 0], [1, 0, 0, 0], [0, -1, 0, 0]]) # non-unitary
m_op1 = MatrixOp(m1)
m_op2 = MatrixOp(m2)
pm1 = (X ^ Y) ^ m_op1 # non-unitary TensoredOp
pm2 = (X ^ Y) ^ m_op2 # non-unitary TensoredOp
self.assertRaises(ExtensionError, pm1.to_circuit)
self.assertRaises(ExtensionError, pm2.to_circuit)
summed_op = pm1 + pm2 # unitary SummedOp([TensoredOp, TensoredOp])
circuit = summed_op.to_circuit() # should transpile without any exception
# same algorithm that leads to summed_op above, but using only arrays and matrix operations
unitary = np.kron(np.kron(x, y), m1 + m2)
self.assertTrue(Operator(unitary).equiv(circuit))
def test_list_op_to_circuit(self):
"""Test if unitary ListOps transpile to circuit. """
# generate unitary matrices of dimension 2,4,8, seed is fixed
np.random.seed(233423)
u2 = unitary_group.rvs(2)
u4 = unitary_group.rvs(4)
u8 = unitary_group.rvs(8)
# pauli matrices as numpy.arrays
x = np.array([[0.0, 1.0], [1.0, 0.0]])
y = np.array([[0.0, -1.0j], [1.0j, 0.0]])
z = np.array([[1.0, 0.0], [0.0, -1.0]])
# create MatrixOp and CircuitOp out of matrices
op2 = MatrixOp(u2)
op4 = MatrixOp(u4)
op8 = MatrixOp(u8)
c2 = op2.to_circuit_op()
# algorithm using only matrix operations on numpy.arrays
xu4 = np.kron(x, u4)
zc2 = np.kron(z, u2)
zc2y = np.kron(zc2, y)
matrix = np.matmul(xu4, zc2y)
matrix = np.matmul(matrix, u8)
matrix = np.kron(matrix, u2)
operator = Operator(matrix)
# same algorithm as above, but using PrimitiveOps
list_op = ((X ^ op4) @ (Z ^ c2 ^ Y) @ op8) ^ op2
circuit = list_op.to_circuit()
# verify that ListOp.to_circuit() outputs correct quantum circuit
self.assertTrue(operator.equiv(circuit), "ListOp.to_circuit() outputs wrong circuit!")
def test_op_to_circuit_with_parameters(self):
"""On parametrized SummedOp, to_matrix_op returns ListOp, instead of MatrixOp. To avoid
the infinite recursion, AquaError is raised. """
m1 = np.array([[0, 0, 1, 0], [0, 0, 0, -1], [0, 0, 0, 0], [0, 0, 0, 0]]) # non-unitary
m2 = np.array([[0, 0, 0, 0], [0, 0, 0, 0], [1, 0, 0, 0], [0, -1, 0, 0]]) # non-unitary
op1_with_param = MatrixOp(m1, Parameter('alpha')) # non-unitary
op2_with_param = MatrixOp(m2, Parameter('beta')) # non-unitary
summed_op_with_param = op1_with_param + op2_with_param # unitary
self.assertRaises(AquaError, summed_op_with_param.to_circuit) # should raise Aqua error
def test_to_matrix(self):
"""to matrix text """
np.testing.assert_array_equal(X.to_matrix(),
Operator.from_label('X').data)
np.testing.assert_array_equal(Y.to_matrix(),
Operator.from_label('Y').data)
np.testing.assert_array_equal(Z.to_matrix(),
Operator.from_label('Z').data)
op1 = Y + H
np.testing.assert_array_almost_equal(op1.to_matrix(),
Y.to_matrix() + H.to_matrix())
op2 = op1 * .5
np.testing.assert_array_almost_equal(op2.to_matrix(),
op1.to_matrix() * .5)
op3 = (4 - .6j) * op2
np.testing.assert_array_almost_equal(op3.to_matrix(),
op2.to_matrix() * (4 - .6j))
op4 = op3.tensor(X)
np.testing.assert_array_almost_equal(
op4.to_matrix(), np.kron(op3.to_matrix(), X.to_matrix()))
op5 = op4.compose(H ^ I)
np.testing.assert_array_almost_equal(
op5.to_matrix(), np.dot(op4.to_matrix(), (H ^ I).to_matrix()))
op6 = op5 + PrimitiveOp(Operator.from_label('+r').data)
np.testing.assert_array_almost_equal(
op6.to_matrix(),
op5.to_matrix() + Operator.from_label('+r').data)
param = Parameter("α")
m = np.array([[0, -1j], [1j, 0]])
op7 = MatrixOp(m, param)
np.testing.assert_array_equal(op7.to_matrix(), m * param)
param = Parameter("β")
op8 = PauliOp(primitive=Pauli(label="Y"), coeff=param)
np.testing.assert_array_equal(op8.to_matrix(), m * param)
param = Parameter("γ")
qc = QuantumCircuit(1)
qc.h(0)
op9 = CircuitOp(qc, coeff=param)
m = np.array([[1, 1], [1, -1]]) / np.sqrt(2)
np.testing.assert_array_equal(op9.to_matrix(), m * param)
def test_summedop_equals(self):
"""Test SummedOp.equals """
ops = [Z, CircuitOp(ZGate()), MatrixOp([[1, 0], [0, -1]]), Zero, Minus]
sum_op = sum(ops + [ListOp(ops)])
self.assertEqual(sum_op, sum_op)
self.assertEqual(sum_op + sum_op, 2 * sum_op)
self.assertEqual(sum_op + sum_op + sum_op, 3 * sum_op)
ops2 = [Z, CircuitOp(ZGate()), MatrixOp([[1, 0], [0, 1]]), Zero, Minus]
sum_op2 = sum(ops2 + [ListOp(ops)])
self.assertNotEqual(sum_op, sum_op2)
self.assertEqual(sum_op2, sum_op2)
sum_op3 = sum(ops)
self.assertNotEqual(sum_op, sum_op3)
self.assertNotEqual(sum_op2, sum_op3)
self.assertEqual(sum_op3, sum_op3)
def invalid_input(self):
"""Test invalid input raises an error."""
op = Z
with self.subTest('alpha < 0'):
with self.assertRaises(ValueError):
_ = CVaRMeasurement(op, alpha=-0.2)
with self.subTest('alpha > 1'):
with self.assertRaises(ValueError):
_ = CVaRMeasurement(op, alpha=12.3)
with self.subTest('Single pauli operator not diagonal'):
op = Y
with self.assertRaises(AquaError):
_ = CVaRMeasurement(op)
with self.subTest('Summed pauli operator not diagonal'):
op = X ^ Z + Z ^ I
with self.assertRaises(AquaError):
_ = CVaRMeasurement(op)
with self.subTest('List operator not diagonal'):
op = ListOp([X ^ Z, Z ^ I])
with self.assertRaises(AquaError):
_ = CVaRMeasurement(op)
with self.subTest('Matrix operator not diagonal'):
op = MatrixOp([[1, 1], [0, 1]])
with self.assertRaises(AquaError):
_ = CVaRMeasurement(op)
def test_matrix_op_conversions(self):
"""Test to reveal QiskitError when to_instruction or to_circuit method is called on
parametrized matrix op."""
m = np.array([[0, 0, 1, 0], [0, 0, 0, -1], [1, 0, 0, 0], [0, -1, 0, 0]])
matrix_op = MatrixOp(m, Parameter('beta'))
for method in ['to_instruction', 'to_circuit']:
with self.subTest(method):
# QiskitError: multiplication of Operator with ParameterExpression isn't implemented
self.assertRaises(QiskitError, getattr(matrix_op, method))
def test_summed_op_equals(self):
"""Test corner cases of SummedOp's equals function."""
with self.subTest('multiplicative factor'):
self.assertEqual(2 * X, X + X)
with self.subTest('commutative'):
self.assertEqual(X + Z, Z + X)
with self.subTest('circuit and paulis'):
z = CircuitOp(ZGate())
self.assertEqual(Z + z, z + Z)
with self.subTest('matrix op and paulis'):
z = MatrixOp([[1, 0], [0, -1]])
self.assertEqual(Z + z, z + Z)
with self.subTest('matrix multiplicative'):
z = MatrixOp([[1, 0], [0, -1]])
self.assertEqual(2 * z, z + z)
with self.subTest('parameter coefficients'):
expr = Parameter('theta')
z = MatrixOp([[1, 0], [0, -1]])
self.assertEqual(expr * z, expr * z)
with self.subTest('different coefficient types'):
expr = Parameter('theta')
z = MatrixOp([[1, 0], [0, -1]])
self.assertNotEqual(expr * z, 2 * z)
with self.subTest('additions aggregation'):
z = MatrixOp([[1, 0], [0, -1]])
a = z + z + Z
b = 2 * z + Z
c = z + Z + z
self.assertEqual(a, b)
self.assertEqual(b, c)
self.assertEqual(a, c)
def test_summed_op_equals(self):
"""Test corner cases of SummedOp's equals function."""
with self.subTest('multiplicative factor'):
self.assertEqual(2 * X, X + X)
with self.subTest('commutative'):
self.assertEqual(X + Z, Z + X)
with self.subTest('circuit and paulis'):
z = CircuitOp(ZGate())
self.assertEqual(Z + z, z + Z)
with self.subTest('matrix op and paulis'):
z = MatrixOp([[1, 0], [0, -1]])
self.assertEqual(Z + z, z + Z)
def test_invalid_primitive(self):
"""Test invalid MatrixOp construction"""
msg = "MatrixOp can only be instantiated with " \
"['list', 'ndarray', 'spmatrix', 'Operator'], not "
with self.assertRaises(TypeError) as cm:
_ = MatrixOp('invalid')
self.assertEqual(str(cm.exception), msg + "'str'")
with self.assertRaises(TypeError) as cm:
_ = MatrixOp(MatrixOperator(np.eye(2)))
self.assertEqual(str(cm.exception), msg + "'MatrixOperator'")
with self.assertRaises(TypeError) as cm:
_ = MatrixOp(None)
self.assertEqual(str(cm.exception), msg + "'NoneType'")
with self.assertRaises(TypeError) as cm:
_ = MatrixOp(2.0)
self.assertEqual(str(cm.exception), msg + "'float'")
def test_primitive_op_to_matrix(self):
"""Test to reveal TypeError: multiplication of 'complex' and 'Parameter' is not
implemented, which is raised on PrimitiveOps with parameter, when to_matrix is called. """
# MatrixOp
m = np.array([[0, 0, 1, 0], [0, 0, 0, -1], [1, 0, 0, 0], [0, -1, 0, 0]])
matrix_op = MatrixOp(m, Parameter('beta'))
# PauliOp
pauli_op = PauliOp(primitive=Pauli(label='XYZ'), coeff=Parameter('beta'))
self.assertRaises(TypeError, pauli_op.to_matrix)
# CircuitOp
qc = QuantumCircuit(2)
qc.cx(0, 1)
circuit_op = CircuitOp(qc, coeff=Parameter('alpha'))
for operator in [matrix_op, pauli_op, circuit_op]:
with self.subTest(operator):
self.assertRaises(TypeError, operator.to_matrix)
from qiskit.circuit.library import HGate
from qiskit.aqua.operators import MatrixOp, ComposedOp, SummedOp, X, Z, I, OperatorStateFn, PrimitiveOp, ListOp
if __name__ == '__main__':
op = MatrixOp([[1, 0], [0, -1]])
initial_str = str(op)
indented_str = op._indent(initial_str)
starts_with_indent = indented_str.startswith(op.INDENTATION)
assert starts_with_indent
indented_str_content = (indented_str[len(op.INDENTATION):] +
op.INDENTATION).split("\n{}".format(
op.INDENTATION))
print(indented_str)
print(initial_str)
print(indented_str_content)
print(initial_str.split("\n"))
assert indented_str_content == initial_str.split("\n")
assert "\t\tString\n\t\tto indent\n" == ListOp._indent(
"String\nto indent\n", indentation="\t\t")
print(
ComposedOp([
SummedOp([
ComposedOp([
OperatorStateFn(SummedOp(
[0.18093119978423136 * Z, -1.052373245772859 * I],
abelian=True),
is_measurement=True),
PrimitiveOp(HGate())
class TestOpConstruction(QiskitAquaTestCase):
"""Operator Construction tests."""
def test_pauli_primitives(self):
""" from to file test """
newop = X ^ Y ^ Z ^ I
self.assertEqual(newop.primitive, Pauli(label='XYZI'))
kpower_op = (Y ^ 5) ^ (I ^ 3)
self.assertEqual(kpower_op.primitive, Pauli(label='YYYYYIII'))
kpower_op2 = (Y ^ I) ^ 4
self.assertEqual(kpower_op2.primitive, Pauli(label='YIYIYIYI'))
# Check immutability
self.assertEqual(X.primitive, Pauli(label='X'))
self.assertEqual(Y.primitive, Pauli(label='Y'))
self.assertEqual(Z.primitive, Pauli(label='Z'))
self.assertEqual(I.primitive, Pauli(label='I'))
def test_composed_eval(self):
""" Test eval of ComposedOp """
self.assertAlmostEqual(Minus.eval('1'), -.5**.5)
def test_evals(self):
""" evals test """
# pylint: disable=no-member
# TODO: Think about eval names
self.assertEqual(Z.eval('0').eval('0'), 1)
self.assertEqual(Z.eval('1').eval('0'), 0)
self.assertEqual(Z.eval('0').eval('1'), 0)
self.assertEqual(Z.eval('1').eval('1'), -1)
self.assertEqual(X.eval('0').eval('0'), 0)
self.assertEqual(X.eval('1').eval('0'), 1)
self.assertEqual(X.eval('0').eval('1'), 1)
self.assertEqual(X.eval('1').eval('1'), 0)
self.assertEqual(Y.eval('0').eval('0'), 0)
self.assertEqual(Y.eval('1').eval('0'), -1j)
self.assertEqual(Y.eval('0').eval('1'), 1j)
self.assertEqual(Y.eval('1').eval('1'), 0)
# Check that Pauli logic eval returns same as matrix logic
self.assertEqual(PrimitiveOp(Z.to_matrix()).eval('0').eval('0'), 1)
self.assertEqual(PrimitiveOp(Z.to_matrix()).eval('1').eval('0'), 0)
self.assertEqual(PrimitiveOp(Z.to_matrix()).eval('0').eval('1'), 0)
self.assertEqual(PrimitiveOp(Z.to_matrix()).eval('1').eval('1'), -1)
self.assertEqual(PrimitiveOp(X.to_matrix()).eval('0').eval('0'), 0)
self.assertEqual(PrimitiveOp(X.to_matrix()).eval('1').eval('0'), 1)
self.assertEqual(PrimitiveOp(X.to_matrix()).eval('0').eval('1'), 1)
self.assertEqual(PrimitiveOp(X.to_matrix()).eval('1').eval('1'), 0)
self.assertEqual(PrimitiveOp(Y.to_matrix()).eval('0').eval('0'), 0)
self.assertEqual(PrimitiveOp(Y.to_matrix()).eval('1').eval('0'), -1j)
self.assertEqual(PrimitiveOp(Y.to_matrix()).eval('0').eval('1'), 1j)
self.assertEqual(PrimitiveOp(Y.to_matrix()).eval('1').eval('1'), 0)
pauli_op = Z ^ I ^ X ^ Y
mat_op = PrimitiveOp(pauli_op.to_matrix())
full_basis = list(
map(''.join, itertools.product('01', repeat=pauli_op.num_qubits)))
for bstr1, bstr2 in itertools.product(full_basis, full_basis):
# print('{} {} {} {}'.format(bstr1, bstr2, pauli_op.eval(bstr1, bstr2),
# mat_op.eval(bstr1, bstr2)))
np.testing.assert_array_almost_equal(
pauli_op.eval(bstr1).eval(bstr2),
mat_op.eval(bstr1).eval(bstr2))
gnarly_op = SummedOp([(H ^ I ^ Y).compose(X ^ X ^ Z).tensor(Z),
PrimitiveOp(Operator.from_label('+r0I')), 3 *
(X ^ CX ^ T)],
coeff=3 + .2j)
gnarly_mat_op = PrimitiveOp(gnarly_op.to_matrix())
full_basis = list(
map(''.join, itertools.product('01', repeat=gnarly_op.num_qubits)))
for bstr1, bstr2 in itertools.product(full_basis, full_basis):
np.testing.assert_array_almost_equal(
gnarly_op.eval(bstr1).eval(bstr2),
gnarly_mat_op.eval(bstr1).eval(bstr2))
def test_circuit_construction(self):
""" circuit construction test """
hadq2 = H ^ I
cz = hadq2.compose(CX).compose(hadq2)
qc = QuantumCircuit(2)
qc.append(cz.primitive, qargs=range(2))
ref_cz_mat = PrimitiveOp(CZGate()).to_matrix()
np.testing.assert_array_almost_equal(cz.to_matrix(), ref_cz_mat)
def test_io_consistency(self):
""" consistency test """
new_op = X ^ Y ^ I
label = 'XYI'
# label = new_op.primitive.to_label()
self.assertEqual(str(new_op.primitive), label)
np.testing.assert_array_almost_equal(new_op.primitive.to_matrix(),
Operator.from_label(label).data)
self.assertEqual(new_op.primitive, Pauli(label=label))
x_mat = X.primitive.to_matrix()
y_mat = Y.primitive.to_matrix()
i_mat = np.eye(2, 2)
np.testing.assert_array_almost_equal(
new_op.primitive.to_matrix(), np.kron(np.kron(x_mat, y_mat),
i_mat))
hi = np.kron(H.to_matrix(), I.to_matrix())
hi2 = Operator.from_label('HI').data
hi3 = (H ^ I).to_matrix()
np.testing.assert_array_almost_equal(hi, hi2)
np.testing.assert_array_almost_equal(hi2, hi3)
xy = np.kron(X.to_matrix(), Y.to_matrix())
xy2 = Operator.from_label('XY').data
xy3 = (X ^ Y).to_matrix()
np.testing.assert_array_almost_equal(xy, xy2)
np.testing.assert_array_almost_equal(xy2, xy3)
# Check if numpy array instantiation is the same as from Operator
matrix_op = Operator.from_label('+r')
np.testing.assert_array_almost_equal(
PrimitiveOp(matrix_op).to_matrix(),
PrimitiveOp(matrix_op.data).to_matrix())
# Ditto list of lists
np.testing.assert_array_almost_equal(
PrimitiveOp(matrix_op.data.tolist()).to_matrix(),
PrimitiveOp(matrix_op.data).to_matrix())
# TODO make sure this works once we resolve endianness mayhem
# qc = QuantumCircuit(3)
# qc.x(2)
# qc.y(1)
# from qiskit import BasicAer, QuantumCircuit, execute
# unitary = execute(qc, BasicAer.get_backend('unitary_simulator')).result().get_unitary()
# np.testing.assert_array_almost_equal(new_op.primitive.to_matrix(), unitary)
def test_to_matrix(self):
"""to matrix text """
np.testing.assert_array_equal(X.to_matrix(),
Operator.from_label('X').data)
np.testing.assert_array_equal(Y.to_matrix(),
Operator.from_label('Y').data)
np.testing.assert_array_equal(Z.to_matrix(),
Operator.from_label('Z').data)
op1 = Y + H
np.testing.assert_array_almost_equal(op1.to_matrix(),
Y.to_matrix() + H.to_matrix())
op2 = op1 * .5
np.testing.assert_array_almost_equal(op2.to_matrix(),
op1.to_matrix() * .5)
op3 = (4 - .6j) * op2
np.testing.assert_array_almost_equal(op3.to_matrix(),
op2.to_matrix() * (4 - .6j))
op4 = op3.tensor(X)
np.testing.assert_array_almost_equal(
op4.to_matrix(), np.kron(op3.to_matrix(), X.to_matrix()))
op5 = op4.compose(H ^ I)
np.testing.assert_array_almost_equal(
op5.to_matrix(), np.dot(op4.to_matrix(), (H ^ I).to_matrix()))
op6 = op5 + PrimitiveOp(Operator.from_label('+r').data)
np.testing.assert_array_almost_equal(
op6.to_matrix(),
op5.to_matrix() + Operator.from_label('+r').data)
def test_matrix_to_instruction(self):
"""Test MatrixOp.to_instruction yields an Instruction object."""
matop = (H ^ 3).to_matrix_op()
with self.subTest('assert to_instruction returns Instruction'):
self.assertIsInstance(matop.to_instruction(), Instruction)
matop = ((H ^ 3) + (Z ^ 3)).to_matrix_op()
with self.subTest('matrix operator is not unitary'):
with self.assertRaises(ExtensionError):
matop.to_instruction()
def test_adjoint(self):
""" adjoint test """
gnarly_op = 3 * (H ^ I ^ Y).compose(X ^ X ^ Z).tensor(T ^ Z) + \
PrimitiveOp(Operator.from_label('+r0IX').data)
np.testing.assert_array_almost_equal(
np.conj(np.transpose(gnarly_op.to_matrix())),
gnarly_op.adjoint().to_matrix())
def test_primitive_strings(self):
""" get primitives test """
self.assertEqual(X.primitive_strings(), {'Pauli'})
gnarly_op = 3 * (H ^ I ^ Y).compose(X ^ X ^ Z).tensor(T ^ Z) + \
PrimitiveOp(Operator.from_label('+r0IX').data)
self.assertEqual(gnarly_op.primitive_strings(),
{'QuantumCircuit', 'Matrix'})
def test_to_pauli_op(self):
""" Test to_pauli_op method """
gnarly_op = 3 * (H ^ I ^ Y).compose(X ^ X ^ Z).tensor(T ^ Z) + \
PrimitiveOp(Operator.from_label('+r0IX').data)
mat_op = gnarly_op.to_matrix_op()
pauli_op = gnarly_op.to_pauli_op()
self.assertIsInstance(pauli_op, SummedOp)
for p in pauli_op:
self.assertIsInstance(p, PauliOp)
np.testing.assert_array_almost_equal(mat_op.to_matrix(),
pauli_op.to_matrix())
def test_circuit_permute(self):
r""" Test the CircuitOp's .permute method """
perm = range(7)[::-1]
c_op = (((CX ^ 3) ^ X) @ (H ^ 7) @ (X ^ Y ^ Z ^ I ^ X ^ X ^ X)
@ (Y ^ (CX ^ 3)) @ (X ^ Y ^ Z ^ I ^ X ^ X ^ X))
c_op_perm = c_op.permute(perm)
self.assertNotEqual(c_op, c_op_perm)
c_op_id = c_op_perm.permute(perm)
self.assertEqual(c_op, c_op_id)
def test_summed_op_reduce(self):
"""Test SummedOp"""
sum_op = (X ^ X * 2) + (Y ^ Y) # type: SummedOp
with self.subTest('SummedOp test 1'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op],
['XX', 'YY'])
self.assertListEqual([op.coeff for op in sum_op], [2, 1])
sum_op = (X ^ X * 2) + (Y ^ Y)
sum_op += Y ^ Y
with self.subTest('SummedOp test 2-a'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op],
['XX', 'YY', 'YY'])
self.assertListEqual([op.coeff for op in sum_op], [2, 1, 1])
sum_op = sum_op.collapse_summands()
with self.subTest('SummedOp test 2-b'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op],
['XX', 'YY'])
self.assertListEqual([op.coeff for op in sum_op], [2, 2])
sum_op = (X ^ X * 2) + (Y ^ Y)
sum_op += (Y ^ Y) + (X ^ X * 2)
with self.subTest('SummedOp test 3-a'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op],
['XX', 'YY', 'YY', 'XX'])
self.assertListEqual([op.coeff for op in sum_op], [2, 1, 1, 2])
sum_op = sum_op.reduce()
with self.subTest('SummedOp test 3-b'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op],
['XX', 'YY'])
self.assertListEqual([op.coeff for op in sum_op], [4, 2])
sum_op = SummedOp([X ^ X * 2, Y ^ Y], 2)
with self.subTest('SummedOp test 4-a'):
self.assertEqual(sum_op.coeff, 2)
self.assertListEqual([str(op.primitive) for op in sum_op],
['XX', 'YY'])
self.assertListEqual([op.coeff for op in sum_op], [2, 1])
sum_op = sum_op.collapse_summands()
with self.subTest('SummedOp test 4-b'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op],
['XX', 'YY'])
self.assertListEqual([op.coeff for op in sum_op], [4, 2])
sum_op = SummedOp([X ^ X * 2, Y ^ Y], 2)
sum_op += Y ^ Y
with self.subTest('SummedOp test 5-a'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op],
['XX', 'YY', 'YY'])
self.assertListEqual([op.coeff for op in sum_op], [4, 2, 1])
sum_op = sum_op.collapse_summands()
with self.subTest('SummedOp test 5-b'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op],
['XX', 'YY'])
self.assertListEqual([op.coeff for op in sum_op], [4, 3])
sum_op = SummedOp([X ^ X * 2, Y ^ Y], 2)
sum_op += (X ^ X) * 2 + (Y ^ Y)
with self.subTest('SummedOp test 6-a'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op],
['XX', 'YY', 'XX', 'YY'])
self.assertListEqual([op.coeff for op in sum_op], [4, 2, 2, 1])
sum_op = sum_op.collapse_summands()
with self.subTest('SummedOp test 6-b'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op],
['XX', 'YY'])
self.assertListEqual([op.coeff for op in sum_op], [6, 3])
sum_op = SummedOp([X ^ X * 2, Y ^ Y], 2)
sum_op += sum_op
with self.subTest('SummedOp test 7-a'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op],
['XX', 'YY', 'XX', 'YY'])
self.assertListEqual([op.coeff for op in sum_op], [4, 2, 4, 2])
sum_op = sum_op.collapse_summands()
with self.subTest('SummedOp test 7-b'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op],
['XX', 'YY'])
self.assertListEqual([op.coeff for op in sum_op], [8, 4])
sum_op = SummedOp([X ^ X * 2, Y ^ Y], 2) + SummedOp([X ^ X * 2, Z ^ Z],
3)
with self.subTest('SummedOp test 8-a'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op],
['XX', 'YY', 'XX', 'ZZ'])
self.assertListEqual([op.coeff for op in sum_op], [4, 2, 6, 3])
sum_op = sum_op.collapse_summands()
with self.subTest('SummedOp test 8-b'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op],
['XX', 'YY', 'ZZ'])
self.assertListEqual([op.coeff for op in sum_op], [10, 2, 3])
def test_summed_op_equals(self):
"""Test corner cases of SummedOp's equals function."""
with self.subTest('multiplicative factor'):
self.assertEqual(2 * X, X + X)
with self.subTest('commutative'):
self.assertEqual(X + Z, Z + X)
with self.subTest('circuit and paulis'):
z = CircuitOp(ZGate())
self.assertEqual(Z + z, z + Z)
with self.subTest('matrix op and paulis'):
z = MatrixOp([[1, 0], [0, -1]])
self.assertEqual(Z + z, z + Z)
def test_circuit_compose_register_independent(self):
"""Test that CircuitOp uses combines circuits independent of the register.
I.e. that is uses ``QuantumCircuit.compose`` over ``combine`` or ``extend``.
"""
op = Z ^ 2
qr = QuantumRegister(2, 'my_qr')
circuit = QuantumCircuit(qr)
composed = op.compose(CircuitOp(circuit))
self.assertEqual(composed.num_qubits, 2)
@data(Z, CircuitOp(ZGate()), MatrixOp([[1, 0], [0, -1]]))
def test_op_hashing(self, op):
"""Regression test against faulty set comparison.
Set comparisons rely on a hash table which requires identical objects to have identical
hashes. Thus, the PrimitiveOp.__hash__ should support this requirement.
"""
self.assertEqual(set([2 * op]), set([2 * op]))
class TestOpConstruction(QiskitAquaTestCase):
"""Operator Construction tests."""
def test_pauli_primitives(self):
""" from to file test """
newop = X ^ Y ^ Z ^ I
self.assertEqual(newop.primitive, Pauli(label='XYZI'))
kpower_op = (Y ^ 5) ^ (I ^ 3)
self.assertEqual(kpower_op.primitive, Pauli(label='YYYYYIII'))
kpower_op2 = (Y ^ I) ^ 4
self.assertEqual(kpower_op2.primitive, Pauli(label='YIYIYIYI'))
# Check immutability
self.assertEqual(X.primitive, Pauli(label='X'))
self.assertEqual(Y.primitive, Pauli(label='Y'))
self.assertEqual(Z.primitive, Pauli(label='Z'))
self.assertEqual(I.primitive, Pauli(label='I'))
def test_composed_eval(self):
""" Test eval of ComposedOp """
self.assertAlmostEqual(Minus.eval('1'), -.5 ** .5)
def test_evals(self):
""" evals test """
# pylint: disable=no-member
# TODO: Think about eval names
self.assertEqual(Z.eval('0').eval('0'), 1)
self.assertEqual(Z.eval('1').eval('0'), 0)
self.assertEqual(Z.eval('0').eval('1'), 0)
self.assertEqual(Z.eval('1').eval('1'), -1)
self.assertEqual(X.eval('0').eval('0'), 0)
self.assertEqual(X.eval('1').eval('0'), 1)
self.assertEqual(X.eval('0').eval('1'), 1)
self.assertEqual(X.eval('1').eval('1'), 0)
self.assertEqual(Y.eval('0').eval('0'), 0)
self.assertEqual(Y.eval('1').eval('0'), -1j)
self.assertEqual(Y.eval('0').eval('1'), 1j)
self.assertEqual(Y.eval('1').eval('1'), 0)
with self.assertRaises(ValueError):
Y.eval('11')
with self.assertRaises(ValueError):
(X ^ Y).eval('1111')
with self.assertRaises(ValueError):
Y.eval((X ^ X).to_matrix_op())
# Check that Pauli logic eval returns same as matrix logic
self.assertEqual(PrimitiveOp(Z.to_matrix()).eval('0').eval('0'), 1)
self.assertEqual(PrimitiveOp(Z.to_matrix()).eval('1').eval('0'), 0)
self.assertEqual(PrimitiveOp(Z.to_matrix()).eval('0').eval('1'), 0)
self.assertEqual(PrimitiveOp(Z.to_matrix()).eval('1').eval('1'), -1)
self.assertEqual(PrimitiveOp(X.to_matrix()).eval('0').eval('0'), 0)
self.assertEqual(PrimitiveOp(X.to_matrix()).eval('1').eval('0'), 1)
self.assertEqual(PrimitiveOp(X.to_matrix()).eval('0').eval('1'), 1)
self.assertEqual(PrimitiveOp(X.to_matrix()).eval('1').eval('1'), 0)
self.assertEqual(PrimitiveOp(Y.to_matrix()).eval('0').eval('0'), 0)
self.assertEqual(PrimitiveOp(Y.to_matrix()).eval('1').eval('0'), -1j)
self.assertEqual(PrimitiveOp(Y.to_matrix()).eval('0').eval('1'), 1j)
self.assertEqual(PrimitiveOp(Y.to_matrix()).eval('1').eval('1'), 0)
pauli_op = Z ^ I ^ X ^ Y
mat_op = PrimitiveOp(pauli_op.to_matrix())
full_basis = list(map(''.join, itertools.product('01', repeat=pauli_op.num_qubits)))
for bstr1, bstr2 in itertools.product(full_basis, full_basis):
# print('{} {} {} {}'.format(bstr1, bstr2, pauli_op.eval(bstr1, bstr2),
# mat_op.eval(bstr1, bstr2)))
np.testing.assert_array_almost_equal(pauli_op.eval(bstr1).eval(bstr2),
mat_op.eval(bstr1).eval(bstr2))
gnarly_op = SummedOp([(H ^ I ^ Y).compose(X ^ X ^ Z).tensor(Z),
PrimitiveOp(Operator.from_label('+r0I')),
3 * (X ^ CX ^ T)], coeff=3 + .2j)
gnarly_mat_op = PrimitiveOp(gnarly_op.to_matrix())
full_basis = list(map(''.join, itertools.product('01', repeat=gnarly_op.num_qubits)))
for bstr1, bstr2 in itertools.product(full_basis, full_basis):
np.testing.assert_array_almost_equal(gnarly_op.eval(bstr1).eval(bstr2),
gnarly_mat_op.eval(bstr1).eval(bstr2))
def test_circuit_construction(self):
""" circuit construction test """
hadq2 = H ^ I
cz = hadq2.compose(CX).compose(hadq2)
qc = QuantumCircuit(2)
qc.append(cz.primitive, qargs=range(2))
ref_cz_mat = PrimitiveOp(CZGate()).to_matrix()
np.testing.assert_array_almost_equal(cz.to_matrix(), ref_cz_mat)
def test_io_consistency(self):
""" consistency test """
new_op = X ^ Y ^ I
label = 'XYI'
# label = new_op.primitive.to_label()
self.assertEqual(str(new_op.primitive), label)
np.testing.assert_array_almost_equal(new_op.primitive.to_matrix(),
Operator.from_label(label).data)
self.assertEqual(new_op.primitive, Pauli(label=label))
x_mat = X.primitive.to_matrix()
y_mat = Y.primitive.to_matrix()
i_mat = np.eye(2, 2)
np.testing.assert_array_almost_equal(new_op.primitive.to_matrix(),
np.kron(np.kron(x_mat, y_mat), i_mat))
hi = np.kron(H.to_matrix(), I.to_matrix())
hi2 = Operator.from_label('HI').data
hi3 = (H ^ I).to_matrix()
np.testing.assert_array_almost_equal(hi, hi2)
np.testing.assert_array_almost_equal(hi2, hi3)
xy = np.kron(X.to_matrix(), Y.to_matrix())
xy2 = Operator.from_label('XY').data
xy3 = (X ^ Y).to_matrix()
np.testing.assert_array_almost_equal(xy, xy2)
np.testing.assert_array_almost_equal(xy2, xy3)
# Check if numpy array instantiation is the same as from Operator
matrix_op = Operator.from_label('+r')
np.testing.assert_array_almost_equal(PrimitiveOp(matrix_op).to_matrix(),
PrimitiveOp(matrix_op.data).to_matrix())
# Ditto list of lists
np.testing.assert_array_almost_equal(PrimitiveOp(matrix_op.data.tolist()).to_matrix(),
PrimitiveOp(matrix_op.data).to_matrix())
# TODO make sure this works once we resolve endianness mayhem
# qc = QuantumCircuit(3)
# qc.x(2)
# qc.y(1)
# from qiskit import BasicAer, QuantumCircuit, execute
# unitary = execute(qc, BasicAer.get_backend('unitary_simulator')).result().get_unitary()
# np.testing.assert_array_almost_equal(new_op.primitive.to_matrix(), unitary)
def test_to_matrix(self):
"""to matrix text """
np.testing.assert_array_equal(X.to_matrix(), Operator.from_label('X').data)
np.testing.assert_array_equal(Y.to_matrix(), Operator.from_label('Y').data)
np.testing.assert_array_equal(Z.to_matrix(), Operator.from_label('Z').data)
op1 = Y + H
np.testing.assert_array_almost_equal(op1.to_matrix(), Y.to_matrix() + H.to_matrix())
op2 = op1 * .5
np.testing.assert_array_almost_equal(op2.to_matrix(), op1.to_matrix() * .5)
op3 = (4 - .6j) * op2
np.testing.assert_array_almost_equal(op3.to_matrix(), op2.to_matrix() * (4 - .6j))
op4 = op3.tensor(X)
np.testing.assert_array_almost_equal(op4.to_matrix(),
np.kron(op3.to_matrix(), X.to_matrix()))
op5 = op4.compose(H ^ I)
np.testing.assert_array_almost_equal(op5.to_matrix(), np.dot(op4.to_matrix(),
(H ^ I).to_matrix()))
op6 = op5 + PrimitiveOp(Operator.from_label('+r').data)
np.testing.assert_array_almost_equal(
op6.to_matrix(), op5.to_matrix() + Operator.from_label('+r').data)
def test_circuit_op_to_matrix(self):
""" test CircuitOp.to_matrix """
qc = QuantumCircuit(1)
qc.rz(1.0, 0)
qcop = CircuitOp(qc)
np.testing.assert_array_almost_equal(
qcop.to_matrix(), scipy.linalg.expm(-0.5j * Z.to_matrix()))
def test_matrix_to_instruction(self):
"""Test MatrixOp.to_instruction yields an Instruction object."""
matop = (H ^ 3).to_matrix_op()
with self.subTest('assert to_instruction returns Instruction'):
self.assertIsInstance(matop.to_instruction(), Instruction)
matop = ((H ^ 3) + (Z ^ 3)).to_matrix_op()
with self.subTest('matrix operator is not unitary'):
with self.assertRaises(ExtensionError):
matop.to_instruction()
def test_adjoint(self):
""" adjoint test """
gnarly_op = 3 * (H ^ I ^ Y).compose(X ^ X ^ Z).tensor(T ^ Z) + \
PrimitiveOp(Operator.from_label('+r0IX').data)
np.testing.assert_array_almost_equal(np.conj(np.transpose(gnarly_op.to_matrix())),
gnarly_op.adjoint().to_matrix())
def test_primitive_strings(self):
""" get primitives test """
self.assertEqual(X.primitive_strings(), {'Pauli'})
gnarly_op = 3 * (H ^ I ^ Y).compose(X ^ X ^ Z).tensor(T ^ Z) + \
PrimitiveOp(Operator.from_label('+r0IX').data)
self.assertEqual(gnarly_op.primitive_strings(), {'QuantumCircuit', 'Matrix'})
def test_to_pauli_op(self):
""" Test to_pauli_op method """
gnarly_op = 3 * (H ^ I ^ Y).compose(X ^ X ^ Z).tensor(T ^ Z) + \
PrimitiveOp(Operator.from_label('+r0IX').data)
mat_op = gnarly_op.to_matrix_op()
pauli_op = gnarly_op.to_pauli_op()
self.assertIsInstance(pauli_op, SummedOp)
for p in pauli_op:
self.assertIsInstance(p, PauliOp)
np.testing.assert_array_almost_equal(mat_op.to_matrix(), pauli_op.to_matrix())
def test_circuit_permute(self):
r""" Test the CircuitOp's .permute method """
perm = range(7)[::-1]
c_op = (((CX ^ 3) ^ X) @
(H ^ 7) @
(X ^ Y ^ Z ^ I ^ X ^ X ^ X) @
(Y ^ (CX ^ 3)) @
(X ^ Y ^ Z ^ I ^ X ^ X ^ X))
c_op_perm = c_op.permute(perm)
self.assertNotEqual(c_op, c_op_perm)
c_op_id = c_op_perm.permute(perm)
self.assertEqual(c_op, c_op_id)
def test_summed_op_reduce(self):
"""Test SummedOp"""
sum_op = (X ^ X * 2) + (Y ^ Y) # type: SummedOp
with self.subTest('SummedOp test 1'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op], ['XX', 'YY'])
self.assertListEqual([op.coeff for op in sum_op], [2, 1])
sum_op = (X ^ X * 2) + (Y ^ Y)
sum_op += Y ^ Y
with self.subTest('SummedOp test 2-a'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op], ['XX', 'YY', 'YY'])
self.assertListEqual([op.coeff for op in sum_op], [2, 1, 1])
sum_op = sum_op.collapse_summands()
with self.subTest('SummedOp test 2-b'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op], ['XX', 'YY'])
self.assertListEqual([op.coeff for op in sum_op], [2, 2])
sum_op = (X ^ X * 2) + (Y ^ Y)
sum_op += (Y ^ Y) + (X ^ X * 2)
with self.subTest('SummedOp test 3-a'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op], ['XX', 'YY', 'YY', 'XX'])
self.assertListEqual([op.coeff for op in sum_op], [2, 1, 1, 2])
sum_op = sum_op.reduce()
with self.subTest('SummedOp test 3-b'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op], ['XX', 'YY'])
self.assertListEqual([op.coeff for op in sum_op], [4, 2])
sum_op = SummedOp([X ^ X * 2, Y ^ Y], 2)
with self.subTest('SummedOp test 4-a'):
self.assertEqual(sum_op.coeff, 2)
self.assertListEqual([str(op.primitive) for op in sum_op], ['XX', 'YY'])
self.assertListEqual([op.coeff for op in sum_op], [2, 1])
sum_op = sum_op.collapse_summands()
with self.subTest('SummedOp test 4-b'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op], ['XX', 'YY'])
self.assertListEqual([op.coeff for op in sum_op], [4, 2])
sum_op = SummedOp([X ^ X * 2, Y ^ Y], 2)
sum_op += Y ^ Y
with self.subTest('SummedOp test 5-a'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op], ['XX', 'YY', 'YY'])
self.assertListEqual([op.coeff for op in sum_op], [4, 2, 1])
sum_op = sum_op.collapse_summands()
with self.subTest('SummedOp test 5-b'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op], ['XX', 'YY'])
self.assertListEqual([op.coeff for op in sum_op], [4, 3])
sum_op = SummedOp([X ^ X * 2, Y ^ Y], 2)
sum_op += (X ^ X) * 2 + (Y ^ Y)
with self.subTest('SummedOp test 6-a'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op], ['XX', 'YY', 'XX', 'YY'])
self.assertListEqual([op.coeff for op in sum_op], [4, 2, 2, 1])
sum_op = sum_op.collapse_summands()
with self.subTest('SummedOp test 6-b'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op], ['XX', 'YY'])
self.assertListEqual([op.coeff for op in sum_op], [6, 3])
sum_op = SummedOp([X ^ X * 2, Y ^ Y], 2)
sum_op += sum_op
with self.subTest('SummedOp test 7-a'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op], ['XX', 'YY', 'XX', 'YY'])
self.assertListEqual([op.coeff for op in sum_op], [4, 2, 4, 2])
sum_op = sum_op.collapse_summands()
with self.subTest('SummedOp test 7-b'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op], ['XX', 'YY'])
self.assertListEqual([op.coeff for op in sum_op], [8, 4])
sum_op = SummedOp([X ^ X * 2, Y ^ Y], 2) + SummedOp([X ^ X * 2, Z ^ Z], 3)
with self.subTest('SummedOp test 8-a'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op], ['XX', 'YY', 'XX', 'ZZ'])
self.assertListEqual([op.coeff for op in sum_op], [4, 2, 6, 3])
sum_op = sum_op.collapse_summands()
with self.subTest('SummedOp test 8-b'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op], ['XX', 'YY', 'ZZ'])
self.assertListEqual([op.coeff for op in sum_op], [10, 2, 3])
def test_compose_op_of_different_dim(self):
"""
Test if smaller operator expands to correct dim when composed with bigger operator.
Test if PrimitiveOps compose methods are consistent.
"""
# PauliOps of different dim
xy_p = (X ^ Y)
xyz_p = (X ^ Y ^ Z)
pauli_op = xy_p @ xyz_p
expected_result = (I ^ I ^ Z)
self.assertEqual(pauli_op, expected_result)
# MatrixOps of different dim
xy_m = xy_p.to_matrix_op()
xyz_m = xyz_p.to_matrix_op()
matrix_op = xy_m @ xyz_m
self.assertEqual(matrix_op, expected_result.to_matrix_op())
# CircuitOps of different dim
xy_c = xy_p.to_circuit_op()
xyz_c = xyz_p.to_circuit_op()
circuit_op = xy_c @ xyz_c
self.assertTrue(np.array_equal(pauli_op.to_matrix(), matrix_op.to_matrix()))
self.assertTrue(np.allclose(pauli_op.to_matrix(), circuit_op.to_matrix(), rtol=1e-14))
self.assertTrue(np.allclose(matrix_op.to_matrix(), circuit_op.to_matrix(), rtol=1e-14))
def test_permute_on_primitive_op(self):
""" Test if permute methods of PrimitiveOps are consistent and work as expected. """
indices = [1, 2, 4]
# PauliOp
pauli_op = (X ^ Y ^ Z)
permuted_pauli_op = pauli_op.permute(indices)
expected_pauli_op = (X ^ I ^ Y ^ Z ^ I)
self.assertEqual(permuted_pauli_op, expected_pauli_op)
# CircuitOp
circuit_op = pauli_op.to_circuit_op()
permuted_circuit_op = circuit_op.permute(indices)
expected_circuit_op = expected_pauli_op.to_circuit_op()
self.assertEqual(permuted_circuit_op.primitive.__str__(),
expected_circuit_op.primitive.__str__())
# MatrixOp
matrix_op = pauli_op.to_matrix_op()
permuted_matrix_op = matrix_op.permute(indices)
expected_matrix_op = expected_pauli_op.to_matrix_op()
equal = np.allclose(permuted_matrix_op.to_matrix(), expected_matrix_op.to_matrix())
self.assertTrue(equal)
def test_permute_on_list_op(self):
""" Test if ListOp permute method is consistent with PrimitiveOps permute methods. """
op1 = (X ^ Y ^ Z).to_circuit_op()
op2 = (Z ^ X ^ Y)
# ComposedOp
indices = [1, 2, 0]
primitive_op = op1 @ op2
primitive_op_perm = primitive_op.permute(indices) # CircuitOp.permute
composed_op = ComposedOp([op1, op2])
composed_op_perm = composed_op.permute(indices)
# reduce the ListOp to PrimitiveOp
to_primitive = composed_op_perm.oplist[0] @ composed_op_perm.oplist[1]
# compare resulting PrimitiveOps
equal = np.allclose(primitive_op_perm.to_matrix(), to_primitive.to_matrix())
self.assertTrue(equal)
# TensoredOp
indices = [3, 5, 4, 0, 2, 1]
primitive_op = op1 ^ op2
primitive_op_perm = primitive_op.permute(indices)
tensored_op = TensoredOp([op1, op2])
tensored_op_perm = tensored_op.permute(indices)
# reduce the ListOp to PrimitiveOp
composed_oplist = tensored_op_perm.oplist
to_primitive = \
composed_oplist[0] @ (composed_oplist[1].oplist[0] ^ composed_oplist[1].oplist[1]) @ \
composed_oplist[2]
# compare resulting PrimitiveOps
equal = np.allclose(primitive_op_perm.to_matrix(), to_primitive.to_matrix())
self.assertTrue(equal)
# SummedOp
primitive_op = (X ^ Y ^ Z)
summed_op = SummedOp([primitive_op])
indices = [1, 2, 0]
primitive_op_perm = primitive_op.permute(indices) # PauliOp.permute
summed_op_perm = summed_op.permute(indices)
# reduce the ListOp to PrimitiveOp
to_primitive = summed_op_perm.oplist[0] @ primitive_op @ summed_op_perm.oplist[2]
# compare resulting PrimitiveOps
equal = np.allclose(primitive_op_perm.to_matrix(), to_primitive.to_matrix())
self.assertTrue(equal)
def test_expand_on_list_op(self):
""" Test if expanded ListOp has expected num_qubits. """
add_qubits = 3
# ComposedOp
composed_op = ComposedOp([(X ^ Y ^ Z), (H ^ T), (Z ^ X ^ Y ^ Z).to_matrix_op()])
expanded = composed_op._expand_dim(add_qubits)
self.assertEqual(composed_op.num_qubits + add_qubits, expanded.num_qubits)
# TensoredOp
tensored_op = TensoredOp([(X ^ Y), (Z ^ I)])
expanded = tensored_op._expand_dim(add_qubits)
self.assertEqual(tensored_op.num_qubits + add_qubits, expanded.num_qubits)
# SummedOp
summed_op = SummedOp([(X ^ Y), (Z ^ I ^ Z)])
expanded = summed_op._expand_dim(add_qubits)
self.assertEqual(summed_op.num_qubits + add_qubits, expanded.num_qubits)
def test_expand_on_state_fn(self):
""" Test if expanded StateFn has expected num_qubits. """
num_qubits = 3
add_qubits = 2
# case CircuitStateFn, with primitive QuantumCircuit
qc2 = QuantumCircuit(num_qubits)
qc2.cx(0, 1)
cfn = CircuitStateFn(qc2, is_measurement=True)
cfn_exp = cfn._expand_dim(add_qubits)
self.assertEqual(cfn_exp.num_qubits, add_qubits + num_qubits)
# case OperatorStateFn, with OperatorBase primitive, in our case CircuitStateFn
osfn = OperatorStateFn(cfn)
osfn_exp = osfn._expand_dim(add_qubits)
self.assertEqual(osfn_exp.num_qubits, add_qubits + num_qubits)
# case DictStateFn
dsfn = DictStateFn('1'*num_qubits, is_measurement=True)
self.assertEqual(dsfn.num_qubits, num_qubits)
dsfn_exp = dsfn._expand_dim(add_qubits)
self.assertEqual(dsfn_exp.num_qubits, num_qubits + add_qubits)
# case VectorStateFn
vsfn = VectorStateFn(np.ones(2**num_qubits, dtype=complex))
self.assertEqual(vsfn.num_qubits, num_qubits)
vsfn_exp = vsfn._expand_dim(add_qubits)
self.assertEqual(vsfn_exp.num_qubits, num_qubits + add_qubits)
def test_permute_on_state_fn(self):
""" Test if StateFns permute are consistent. """
num_qubits = 4
dim = 2**num_qubits
primitive_list = [1.0/(i+1) for i in range(dim)]
primitive_dict = {format(i, 'b').zfill(num_qubits): 1.0/(i+1) for i in range(dim)}
dict_fn = DictStateFn(primitive=primitive_dict, is_measurement=True)
vec_fn = VectorStateFn(primitive=primitive_list, is_measurement=True)
# check if dict_fn and vec_fn are equivalent
equivalent = np.allclose(dict_fn.to_matrix(), vec_fn.to_matrix())
self.assertTrue(equivalent)
# permute
indices = [2, 3, 0, 1]
permute_dict = dict_fn.permute(indices)
permute_vect = vec_fn.permute(indices)
equivalent = np.allclose(permute_dict.to_matrix(), permute_vect.to_matrix())
self.assertTrue(equivalent)
def test_compose_consistency(self):
"""Test if PrimitiveOp @ ComposedOp is consistent with ComposedOp @ PrimitiveOp."""
# PauliOp
op1 = (X ^ Y ^ Z)
op2 = (X ^ Y ^ Z)
op3 = (X ^ Y ^ Z).to_circuit_op()
comp1 = op1 @ ComposedOp([op2, op3])
comp2 = ComposedOp([op3, op2]) @ op1
self.assertListEqual(comp1.oplist, list(reversed(comp2.oplist)))
# CircitOp
op1 = op1.to_circuit_op()
op2 = op2.to_circuit_op()
op3 = op3.to_matrix_op()
comp1 = op1 @ ComposedOp([op2, op3])
comp2 = ComposedOp([op3, op2]) @ op1
self.assertListEqual(comp1.oplist, list(reversed(comp2.oplist)))
# MatrixOp
op1 = op1.to_matrix_op()
op2 = op2.to_matrix_op()
op3 = op3.to_pauli_op()
comp1 = op1 @ ComposedOp([op2, op3])
comp2 = ComposedOp([op3, op2]) @ op1
self.assertListEqual(comp1.oplist, list(reversed(comp2.oplist)))
def test_compose_with_indices(self):
""" Test compose method using its permutation feature."""
pauli_op = (X ^ Y ^ Z)
circuit_op = (T ^ H)
matrix_op = (X ^ Y ^ H ^ T).to_matrix_op()
evolved_op = EvolvedOp(matrix_op)
# composition of PrimitiveOps
num_qubits = 4
primitive_op = pauli_op @ circuit_op @ matrix_op
composed_op = pauli_op @ circuit_op @ evolved_op
self.assertEqual(primitive_op.num_qubits, num_qubits)
self.assertEqual(composed_op.num_qubits, num_qubits)
# with permutation
num_qubits = 5
indices = [1, 4]
permuted_primitive_op = evolved_op @ circuit_op.permute(indices) @ pauli_op @ matrix_op
composed_primitive_op = \
evolved_op @ pauli_op.compose(circuit_op, permutation=indices, front=True) @ matrix_op
self.assertTrue(np.allclose(permuted_primitive_op.to_matrix(),
composed_primitive_op.to_matrix()))
self.assertEqual(num_qubits, permuted_primitive_op.num_qubits)
# ListOp
num_qubits = 6
tensored_op = TensoredOp([pauli_op, circuit_op])
summed_op = pauli_op + circuit_op.permute([2, 1])
composed_op = circuit_op @ evolved_op @ matrix_op
list_op = summed_op @ composed_op.compose(tensored_op, permutation=[1, 2, 3, 5, 4],
front=True)
self.assertEqual(num_qubits, list_op.num_qubits)
num_qubits = 4
circuit_fn = CircuitStateFn(primitive=circuit_op.primitive, is_measurement=True)
operator_fn = OperatorStateFn(primitive=circuit_op ^ circuit_op, is_measurement=True)
no_perm_op = circuit_fn @ operator_fn
self.assertEqual(no_perm_op.num_qubits, num_qubits)
indices = [0, 4]
perm_op = operator_fn.compose(circuit_fn, permutation=indices, front=True)
self.assertEqual(perm_op.num_qubits, max(indices) + 1)
# StateFn
num_qubits = 3
dim = 2**num_qubits
vec = [1.0/(i+1) for i in range(dim)]
dic = {format(i, 'b').zfill(num_qubits): 1.0/(i+1) for i in range(dim)}
is_measurement = True
op_state_fn = OperatorStateFn(matrix_op, is_measurement=is_measurement) # num_qubit = 4
vec_state_fn = VectorStateFn(vec, is_measurement=is_measurement) # 3
dic_state_fn = DictStateFn(dic, is_measurement=is_measurement) # 3
circ_state_fn = CircuitStateFn(circuit_op.to_circuit(), is_measurement=is_measurement) # 2
composed_op = op_state_fn @ vec_state_fn @ dic_state_fn @ circ_state_fn
self.assertEqual(composed_op.num_qubits, op_state_fn.num_qubits)
# with permutation
perm = [2, 4, 6]
composed = \
op_state_fn @ dic_state_fn.compose(vec_state_fn, permutation=perm, front=True) @ \
circ_state_fn
self.assertEqual(composed.num_qubits, max(perm) + 1)
def test_summed_op_equals(self):
"""Test corner cases of SummedOp's equals function."""
with self.subTest('multiplicative factor'):
self.assertEqual(2 * X, X + X)
with self.subTest('commutative'):
self.assertEqual(X + Z, Z + X)
with self.subTest('circuit and paulis'):
z = CircuitOp(ZGate())
self.assertEqual(Z + z, z + Z)
with self.subTest('matrix op and paulis'):
z = MatrixOp([[1, 0], [0, -1]])
self.assertEqual(Z + z, z + Z)
with self.subTest('matrix multiplicative'):
z = MatrixOp([[1, 0], [0, -1]])
self.assertEqual(2 * z, z + z)
with self.subTest('parameter coefficients'):
expr = Parameter('theta')
z = MatrixOp([[1, 0], [0, -1]])
self.assertEqual(expr * z, expr * z)
with self.subTest('different coefficient types'):
expr = Parameter('theta')
z = MatrixOp([[1, 0], [0, -1]])
self.assertNotEqual(expr * z, 2 * z)
with self.subTest('additions aggregation'):
z = MatrixOp([[1, 0], [0, -1]])
a = z + z + Z
b = 2 * z + Z
c = z + Z + z
self.assertEqual(a, b)
self.assertEqual(b, c)
self.assertEqual(a, c)
def test_circuit_compose_register_independent(self):
"""Test that CircuitOp uses combines circuits independent of the register.
I.e. that is uses ``QuantumCircuit.compose`` over ``combine`` or ``extend``.
"""
op = Z ^ 2
qr = QuantumRegister(2, 'my_qr')
circuit = QuantumCircuit(qr)
composed = op.compose(CircuitOp(circuit))
self.assertEqual(composed.num_qubits, 2)
def test_matrix_op_conversions(self):
"""Test to reveal QiskitError when to_instruction or to_circuit method is called on
parametrized matrix op."""
m = np.array([[0, 0, 1, 0], [0, 0, 0, -1], [1, 0, 0, 0], [0, -1, 0, 0]])
matrix_op = MatrixOp(m, Parameter('beta'))
for method in ['to_instruction', 'to_circuit']:
with self.subTest(method):
# QiskitError: multiplication of Operator with ParameterExpression isn't implemented
self.assertRaises(QiskitError, getattr(matrix_op, method))
def test_primitive_op_to_matrix(self):
"""Test to reveal TypeError: multiplication of 'complex' and 'Parameter' is not
implemented, which is raised on PrimitiveOps with parameter, when to_matrix is called. """
# MatrixOp
m = np.array([[0, 0, 1, 0], [0, 0, 0, -1], [1, 0, 0, 0], [0, -1, 0, 0]])
matrix_op = MatrixOp(m, Parameter('beta'))
# PauliOp
pauli_op = PauliOp(primitive=Pauli(label='XYZ'), coeff=Parameter('beta'))
self.assertRaises(TypeError, pauli_op.to_matrix)
# CircuitOp
qc = QuantumCircuit(2)
qc.cx(0, 1)
circuit_op = CircuitOp(qc, coeff=Parameter('alpha'))
for operator in [matrix_op, pauli_op, circuit_op]:
with self.subTest(operator):
self.assertRaises(TypeError, operator.to_matrix)
def test_list_op_to_circuit(self):
"""Test if unitary ListOps transpile to circuit. """
# generate unitary matrices of dimension 2,4,8, seed is fixed
np.random.seed(233423)
u2 = unitary_group.rvs(2)
u4 = unitary_group.rvs(4)
u8 = unitary_group.rvs(8)
# pauli matrices as numpy.arrays
x = np.array([[0.0, 1.0], [1.0, 0.0]])
y = np.array([[0.0, -1.0j], [1.0j, 0.0]])
z = np.array([[1.0, 0.0], [0.0, -1.0]])
# create MatrixOp and CircuitOp out of matrices
op2 = MatrixOp(u2)
op4 = MatrixOp(u4)
op8 = MatrixOp(u8)
c2 = op2.to_circuit_op()
# algorithm using only matrix operations on numpy.arrays
xu4 = np.kron(x, u4)
zc2 = np.kron(z, u2)
zc2y = np.kron(zc2, y)
matrix = np.matmul(xu4, zc2y)
matrix = np.matmul(matrix, u8)
matrix = np.kron(matrix, u2)
operator = Operator(matrix)
# same algorithm as above, but using PrimitiveOps
list_op = ((X ^ op4) @ (Z ^ c2 ^ Y) @ op8) ^ op2
circuit = list_op.to_circuit()
# verify that ListOp.to_circuit() outputs correct quantum circuit
self.assertTrue(operator.equiv(circuit), "ListOp.to_circuit() outputs wrong circuit!")
def test_composed_op_to_circuit(self):
"""
Test if unitary ComposedOp transpile to circuit and represents expected operator.
Test if to_circuit on non-unitary ListOp raises exception.
"""
x = np.array([[0.0, 1.0], [1.0, 0.0]]) # Pauli X as numpy array
y = np.array([[0.0, -1.0j], [1.0j, 0.0]]) # Pauli Y as numpy array
m1 = np.array([[0, 0, 1, 0], [0, 0, 0, -1], [0, 0, 0, 0], [0, 0, 0, 0]]) # non-unitary
m2 = np.array([[0, 0, 0, 0], [0, 0, 0, 0], [1, 0, 0, 0], [0, -1, 0, 0]]) # non-unitary
m_op1 = MatrixOp(m1)
m_op2 = MatrixOp(m2)
pm1 = (X ^ Y) ^ m_op1 # non-unitary TensoredOp
pm2 = (X ^ Y) ^ m_op2 # non-unitary TensoredOp
self.assertRaises(ExtensionError, pm1.to_circuit)
self.assertRaises(ExtensionError, pm2.to_circuit)
summed_op = pm1 + pm2 # unitary SummedOp([TensoredOp, TensoredOp])
circuit = summed_op.to_circuit() # should transpile without any exception
# same algorithm that leads to summed_op above, but using only arrays and matrix operations
unitary = np.kron(np.kron(x, y), m1 + m2)
self.assertTrue(Operator(unitary).equiv(circuit))
def test_op_to_circuit_with_parameters(self):
"""On parametrized SummedOp, to_matrix_op returns ListOp, instead of MatrixOp. To avoid
the infinite recursion, AquaError is raised. """
m1 = np.array([[0, 0, 1, 0], [0, 0, 0, -1], [0, 0, 0, 0], [0, 0, 0, 0]]) # non-unitary
m2 = np.array([[0, 0, 0, 0], [0, 0, 0, 0], [1, 0, 0, 0], [0, -1, 0, 0]]) # non-unitary
op1_with_param = MatrixOp(m1, Parameter('alpha')) # non-unitary
op2_with_param = MatrixOp(m2, Parameter('beta')) # non-unitary
summed_op_with_param = op1_with_param + op2_with_param # unitary
self.assertRaises(AquaError, summed_op_with_param.to_circuit) # should raise Aqua error
def test_permute_list_op_with_inconsistent_num_qubits(self):
"""Test if permute raises error if ListOp contains operators with different num_qubits."""
list_op = ListOp([X, X ^ X])
self.assertRaises(AquaError, list_op.permute, [0, 1])
@data(Z, CircuitOp(ZGate()), MatrixOp([[1, 0], [0, -1]]))
def test_op_indent(self, op):
"""Test that indentation correctly adds INDENTATION at the beginning of each line"""
initial_str = str(op)
indented_str = op._indent(initial_str)
starts_with_indent = indented_str.startswith(op.INDENTATION)
self.assertTrue(starts_with_indent)
indented_str_content = (
indented_str[len(op.INDENTATION):]
).split("\n{}".format(op.INDENTATION))
self.assertListEqual(indented_str_content, initial_str.split("\n"))
def test_composed_op_immutable_under_eval(self):
"""Test ``ComposedOp.eval`` does not change the operator instance."""
op = 2 * ComposedOp([X])
_ = op.eval()
# previous bug: after op.eval(), op was 2 * ComposedOp([2 * X])
self.assertEqual(op, 2 * ComposedOp([X]))
def test_op_parameters(self):
"""Test that Parameters are stored correctly"""
phi = Parameter('φ')
theta = ParameterVector(name='θ',
length=2)
qc = QuantumCircuit(2)
qc.rz(phi, 0)
qc.rz(phi, 1)
for i in range(2):
qc.rx(theta[i], i)
qc.h(0)
qc.x(1)
l = Parameter('λ')
op = PrimitiveOp(qc,
coeff=l)
params = set([phi, l, *theta.params])
self.assertEqual(params, op.parameters)
self.assertEqual(params, StateFn(op).parameters)
self.assertEqual(params, StateFn(qc, coeff=l).parameters)
def test_list_op_parameters(self):
"""Test that Parameters are stored correctly in a List Operator"""
lam = Parameter('λ')
phi = Parameter('φ')
omega = Parameter('ω')
mat_op = PrimitiveOp([[0, 1],
[1, 0]],
coeff=omega)
qc = QuantumCircuit(1)
qc.rx(phi, 0)
qc_op = PrimitiveOp(qc)
op1 = SummedOp([mat_op, qc_op])
params = [phi, omega]
self.assertEqual(op1.parameters, set(params))
# check list nesting case
op2 = PrimitiveOp([[1, 0],
[0, -1]],
coeff=lam)
list_op = ListOp([op1, op2])
params.append(lam)
self.assertEqual(list_op.parameters, set(params))
@data(VectorStateFn([1, 0]),
DictStateFn({'0': 1}),
CircuitStateFn(QuantumCircuit(1)),
OperatorStateFn(I),
OperatorStateFn(MatrixOp([[1, 0], [0, 1]])),
OperatorStateFn(CircuitOp(QuantumCircuit(1))))
def test_statefn_eval(self, op):
"""Test calling eval on StateFn returns the statevector."""
expected = Statevector([1, 0])
self.assertEqual(op.eval().primitive, expected)
def test_to_circuit_op(self):
"""Test to_circuit_op method."""
vector = np.array([2, 2])
vsfn = VectorStateFn([1, 1], coeff=2)
dsfn = DictStateFn({'0': 1, '1': 1}, coeff=2)
for sfn in [vsfn, dsfn]:
np.testing.assert_array_almost_equal(
sfn.to_circuit_op().eval().primitive.data, vector
)
def test_invalid_primitive(self):
"""Test invalid MatrixOp construction"""
msg = "MatrixOp can only be instantiated with " \
"['list', 'ndarray', 'spmatrix', 'Operator'], not "
with self.assertRaises(TypeError) as cm:
_ = MatrixOp('invalid')
self.assertEqual(str(cm.exception), msg + "'str'")
with self.assertRaises(TypeError) as cm:
_ = MatrixOp(MatrixOperator(np.eye(2)))
self.assertEqual(str(cm.exception), msg + "'MatrixOperator'")
with self.assertRaises(TypeError) as cm:
_ = MatrixOp(None)
self.assertEqual(str(cm.exception), msg + "'NoneType'")
with self.assertRaises(TypeError) as cm:
_ = MatrixOp(2.0)
self.assertEqual(str(cm.exception), msg + "'float'")
def test_summedop_equals(self):
"""Test SummedOp.equals """
ops = [Z, CircuitOp(ZGate()), MatrixOp([[1, 0], [0, -1]]), Zero, Minus]
sum_op = sum(ops + [ListOp(ops)])
self.assertEqual(sum_op, sum_op)
self.assertEqual(sum_op + sum_op, 2 * sum_op)
self.assertEqual(sum_op + sum_op + sum_op, 3 * sum_op)
ops2 = [Z, CircuitOp(ZGate()), MatrixOp([[1, 0], [0, 1]]), Zero, Minus]
sum_op2 = sum(ops2 + [ListOp(ops)])
self.assertNotEqual(sum_op, sum_op2)
self.assertEqual(sum_op2, sum_op2)
sum_op3 = sum(ops)
self.assertNotEqual(sum_op, sum_op3)
self.assertNotEqual(sum_op2, sum_op3)
self.assertEqual(sum_op3, sum_op3)
class TestOpConstruction(QiskitAquaTestCase):
"""Operator Construction tests."""
def test_pauli_primitives(self):
""" from to file test """
newop = X ^ Y ^ Z ^ I
self.assertEqual(newop.primitive, Pauli(label='XYZI'))
kpower_op = (Y ^ 5) ^ (I ^ 3)
self.assertEqual(kpower_op.primitive, Pauli(label='YYYYYIII'))
kpower_op2 = (Y ^ I) ^ 4
self.assertEqual(kpower_op2.primitive, Pauli(label='YIYIYIYI'))
# Check immutability
self.assertEqual(X.primitive, Pauli(label='X'))
self.assertEqual(Y.primitive, Pauli(label='Y'))
self.assertEqual(Z.primitive, Pauli(label='Z'))
self.assertEqual(I.primitive, Pauli(label='I'))
def test_composed_eval(self):
""" Test eval of ComposedOp """
self.assertAlmostEqual(Minus.eval('1'), -.5**.5)
def test_evals(self):
""" evals test """
# pylint: disable=no-member
# TODO: Think about eval names
self.assertEqual(Z.eval('0').eval('0'), 1)
self.assertEqual(Z.eval('1').eval('0'), 0)
self.assertEqual(Z.eval('0').eval('1'), 0)
self.assertEqual(Z.eval('1').eval('1'), -1)
self.assertEqual(X.eval('0').eval('0'), 0)
self.assertEqual(X.eval('1').eval('0'), 1)
self.assertEqual(X.eval('0').eval('1'), 1)
self.assertEqual(X.eval('1').eval('1'), 0)
self.assertEqual(Y.eval('0').eval('0'), 0)
self.assertEqual(Y.eval('1').eval('0'), -1j)
self.assertEqual(Y.eval('0').eval('1'), 1j)
self.assertEqual(Y.eval('1').eval('1'), 0)
with self.assertRaises(ValueError):
Y.eval('11')
with self.assertRaises(ValueError):
(X ^ Y).eval('1111')
with self.assertRaises(ValueError):
Y.eval((X ^ X).to_matrix_op())
# Check that Pauli logic eval returns same as matrix logic
self.assertEqual(PrimitiveOp(Z.to_matrix()).eval('0').eval('0'), 1)
self.assertEqual(PrimitiveOp(Z.to_matrix()).eval('1').eval('0'), 0)
self.assertEqual(PrimitiveOp(Z.to_matrix()).eval('0').eval('1'), 0)
self.assertEqual(PrimitiveOp(Z.to_matrix()).eval('1').eval('1'), -1)
self.assertEqual(PrimitiveOp(X.to_matrix()).eval('0').eval('0'), 0)
self.assertEqual(PrimitiveOp(X.to_matrix()).eval('1').eval('0'), 1)
self.assertEqual(PrimitiveOp(X.to_matrix()).eval('0').eval('1'), 1)
self.assertEqual(PrimitiveOp(X.to_matrix()).eval('1').eval('1'), 0)
self.assertEqual(PrimitiveOp(Y.to_matrix()).eval('0').eval('0'), 0)
self.assertEqual(PrimitiveOp(Y.to_matrix()).eval('1').eval('0'), -1j)
self.assertEqual(PrimitiveOp(Y.to_matrix()).eval('0').eval('1'), 1j)
self.assertEqual(PrimitiveOp(Y.to_matrix()).eval('1').eval('1'), 0)
pauli_op = Z ^ I ^ X ^ Y
mat_op = PrimitiveOp(pauli_op.to_matrix())
full_basis = list(
map(''.join, itertools.product('01', repeat=pauli_op.num_qubits)))
for bstr1, bstr2 in itertools.product(full_basis, full_basis):
# print('{} {} {} {}'.format(bstr1, bstr2, pauli_op.eval(bstr1, bstr2),
# mat_op.eval(bstr1, bstr2)))
np.testing.assert_array_almost_equal(
pauli_op.eval(bstr1).eval(bstr2),
mat_op.eval(bstr1).eval(bstr2))
gnarly_op = SummedOp([(H ^ I ^ Y).compose(X ^ X ^ Z).tensor(Z),
PrimitiveOp(Operator.from_label('+r0I')), 3 *
(X ^ CX ^ T)],
coeff=3 + .2j)
gnarly_mat_op = PrimitiveOp(gnarly_op.to_matrix())
full_basis = list(
map(''.join, itertools.product('01', repeat=gnarly_op.num_qubits)))
for bstr1, bstr2 in itertools.product(full_basis, full_basis):
np.testing.assert_array_almost_equal(
gnarly_op.eval(bstr1).eval(bstr2),
gnarly_mat_op.eval(bstr1).eval(bstr2))
def test_circuit_construction(self):
""" circuit construction test """
hadq2 = H ^ I
cz = hadq2.compose(CX).compose(hadq2)
qc = QuantumCircuit(2)
qc.append(cz.primitive, qargs=range(2))
ref_cz_mat = PrimitiveOp(CZGate()).to_matrix()
np.testing.assert_array_almost_equal(cz.to_matrix(), ref_cz_mat)
def test_io_consistency(self):
""" consistency test """
new_op = X ^ Y ^ I
label = 'XYI'
# label = new_op.primitive.to_label()
self.assertEqual(str(new_op.primitive), label)
np.testing.assert_array_almost_equal(new_op.primitive.to_matrix(),
Operator.from_label(label).data)
self.assertEqual(new_op.primitive, Pauli(label=label))
x_mat = X.primitive.to_matrix()
y_mat = Y.primitive.to_matrix()
i_mat = np.eye(2, 2)
np.testing.assert_array_almost_equal(
new_op.primitive.to_matrix(), np.kron(np.kron(x_mat, y_mat),
i_mat))
hi = np.kron(H.to_matrix(), I.to_matrix())
hi2 = Operator.from_label('HI').data
hi3 = (H ^ I).to_matrix()
np.testing.assert_array_almost_equal(hi, hi2)
np.testing.assert_array_almost_equal(hi2, hi3)
xy = np.kron(X.to_matrix(), Y.to_matrix())
xy2 = Operator.from_label('XY').data
xy3 = (X ^ Y).to_matrix()
np.testing.assert_array_almost_equal(xy, xy2)
np.testing.assert_array_almost_equal(xy2, xy3)
# Check if numpy array instantiation is the same as from Operator
matrix_op = Operator.from_label('+r')
np.testing.assert_array_almost_equal(
PrimitiveOp(matrix_op).to_matrix(),
PrimitiveOp(matrix_op.data).to_matrix())
# Ditto list of lists
np.testing.assert_array_almost_equal(
PrimitiveOp(matrix_op.data.tolist()).to_matrix(),
PrimitiveOp(matrix_op.data).to_matrix())
# TODO make sure this works once we resolve endianness mayhem
# qc = QuantumCircuit(3)
# qc.x(2)
# qc.y(1)
# from qiskit import BasicAer, QuantumCircuit, execute
# unitary = execute(qc, BasicAer.get_backend('unitary_simulator')).result().get_unitary()
# np.testing.assert_array_almost_equal(new_op.primitive.to_matrix(), unitary)
def test_to_matrix(self):
"""to matrix text """
np.testing.assert_array_equal(X.to_matrix(),
Operator.from_label('X').data)
np.testing.assert_array_equal(Y.to_matrix(),
Operator.from_label('Y').data)
np.testing.assert_array_equal(Z.to_matrix(),
Operator.from_label('Z').data)
op1 = Y + H
np.testing.assert_array_almost_equal(op1.to_matrix(),
Y.to_matrix() + H.to_matrix())
op2 = op1 * .5
np.testing.assert_array_almost_equal(op2.to_matrix(),
op1.to_matrix() * .5)
op3 = (4 - .6j) * op2
np.testing.assert_array_almost_equal(op3.to_matrix(),
op2.to_matrix() * (4 - .6j))
op4 = op3.tensor(X)
np.testing.assert_array_almost_equal(
op4.to_matrix(), np.kron(op3.to_matrix(), X.to_matrix()))
op5 = op4.compose(H ^ I)
np.testing.assert_array_almost_equal(
op5.to_matrix(), np.dot(op4.to_matrix(), (H ^ I).to_matrix()))
op6 = op5 + PrimitiveOp(Operator.from_label('+r').data)
np.testing.assert_array_almost_equal(
op6.to_matrix(),
op5.to_matrix() + Operator.from_label('+r').data)
def test_matrix_to_instruction(self):
"""Test MatrixOp.to_instruction yields an Instruction object."""
matop = (H ^ 3).to_matrix_op()
with self.subTest('assert to_instruction returns Instruction'):
self.assertIsInstance(matop.to_instruction(), Instruction)
matop = ((H ^ 3) + (Z ^ 3)).to_matrix_op()
with self.subTest('matrix operator is not unitary'):
with self.assertRaises(ExtensionError):
matop.to_instruction()
def test_adjoint(self):
""" adjoint test """
gnarly_op = 3 * (H ^ I ^ Y).compose(X ^ X ^ Z).tensor(T ^ Z) + \
PrimitiveOp(Operator.from_label('+r0IX').data)
np.testing.assert_array_almost_equal(
np.conj(np.transpose(gnarly_op.to_matrix())),
gnarly_op.adjoint().to_matrix())
def test_primitive_strings(self):
""" get primitives test """
self.assertEqual(X.primitive_strings(), {'Pauli'})
gnarly_op = 3 * (H ^ I ^ Y).compose(X ^ X ^ Z).tensor(T ^ Z) + \
PrimitiveOp(Operator.from_label('+r0IX').data)
self.assertEqual(gnarly_op.primitive_strings(),
{'QuantumCircuit', 'Matrix'})
def test_to_pauli_op(self):
""" Test to_pauli_op method """
gnarly_op = 3 * (H ^ I ^ Y).compose(X ^ X ^ Z).tensor(T ^ Z) + \
PrimitiveOp(Operator.from_label('+r0IX').data)
mat_op = gnarly_op.to_matrix_op()
pauli_op = gnarly_op.to_pauli_op()
self.assertIsInstance(pauli_op, SummedOp)
for p in pauli_op:
self.assertIsInstance(p, PauliOp)
np.testing.assert_array_almost_equal(mat_op.to_matrix(),
pauli_op.to_matrix())
def test_circuit_permute(self):
r""" Test the CircuitOp's .permute method """
perm = range(7)[::-1]
c_op = (((CX ^ 3) ^ X) @ (H ^ 7) @ (X ^ Y ^ Z ^ I ^ X ^ X ^ X)
@ (Y ^ (CX ^ 3)) @ (X ^ Y ^ Z ^ I ^ X ^ X ^ X))
c_op_perm = c_op.permute(perm)
self.assertNotEqual(c_op, c_op_perm)
c_op_id = c_op_perm.permute(perm)
self.assertEqual(c_op, c_op_id)
def test_summed_op_reduce(self):
"""Test SummedOp"""
sum_op = (X ^ X * 2) + (Y ^ Y) # type: SummedOp
with self.subTest('SummedOp test 1'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op],
['XX', 'YY'])
self.assertListEqual([op.coeff for op in sum_op], [2, 1])
sum_op = (X ^ X * 2) + (Y ^ Y)
sum_op += Y ^ Y
with self.subTest('SummedOp test 2-a'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op],
['XX', 'YY', 'YY'])
self.assertListEqual([op.coeff for op in sum_op], [2, 1, 1])
sum_op = sum_op.collapse_summands()
with self.subTest('SummedOp test 2-b'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op],
['XX', 'YY'])
self.assertListEqual([op.coeff for op in sum_op], [2, 2])
sum_op = (X ^ X * 2) + (Y ^ Y)
sum_op += (Y ^ Y) + (X ^ X * 2)
with self.subTest('SummedOp test 3-a'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op],
['XX', 'YY', 'YY', 'XX'])
self.assertListEqual([op.coeff for op in sum_op], [2, 1, 1, 2])
sum_op = sum_op.reduce()
with self.subTest('SummedOp test 3-b'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op],
['XX', 'YY'])
self.assertListEqual([op.coeff for op in sum_op], [4, 2])
sum_op = SummedOp([X ^ X * 2, Y ^ Y], 2)
with self.subTest('SummedOp test 4-a'):
self.assertEqual(sum_op.coeff, 2)
self.assertListEqual([str(op.primitive) for op in sum_op],
['XX', 'YY'])
self.assertListEqual([op.coeff for op in sum_op], [2, 1])
sum_op = sum_op.collapse_summands()
with self.subTest('SummedOp test 4-b'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op],
['XX', 'YY'])
self.assertListEqual([op.coeff for op in sum_op], [4, 2])
sum_op = SummedOp([X ^ X * 2, Y ^ Y], 2)
sum_op += Y ^ Y
with self.subTest('SummedOp test 5-a'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op],
['XX', 'YY', 'YY'])
self.assertListEqual([op.coeff for op in sum_op], [4, 2, 1])
sum_op = sum_op.collapse_summands()
with self.subTest('SummedOp test 5-b'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op],
['XX', 'YY'])
self.assertListEqual([op.coeff for op in sum_op], [4, 3])
sum_op = SummedOp([X ^ X * 2, Y ^ Y], 2)
sum_op += (X ^ X) * 2 + (Y ^ Y)
with self.subTest('SummedOp test 6-a'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op],
['XX', 'YY', 'XX', 'YY'])
self.assertListEqual([op.coeff for op in sum_op], [4, 2, 2, 1])
sum_op = sum_op.collapse_summands()
with self.subTest('SummedOp test 6-b'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op],
['XX', 'YY'])
self.assertListEqual([op.coeff for op in sum_op], [6, 3])
sum_op = SummedOp([X ^ X * 2, Y ^ Y], 2)
sum_op += sum_op
with self.subTest('SummedOp test 7-a'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op],
['XX', 'YY', 'XX', 'YY'])
self.assertListEqual([op.coeff for op in sum_op], [4, 2, 4, 2])
sum_op = sum_op.collapse_summands()
with self.subTest('SummedOp test 7-b'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op],
['XX', 'YY'])
self.assertListEqual([op.coeff for op in sum_op], [8, 4])
sum_op = SummedOp([X ^ X * 2, Y ^ Y], 2) + SummedOp([X ^ X * 2, Z ^ Z],
3)
with self.subTest('SummedOp test 8-a'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op],
['XX', 'YY', 'XX', 'ZZ'])
self.assertListEqual([op.coeff for op in sum_op], [4, 2, 6, 3])
sum_op = sum_op.collapse_summands()
with self.subTest('SummedOp test 8-b'):
self.assertEqual(sum_op.coeff, 1)
self.assertListEqual([str(op.primitive) for op in sum_op],
['XX', 'YY', 'ZZ'])
self.assertListEqual([op.coeff for op in sum_op], [10, 2, 3])
def test_compose_consistency(self):
"""Checks if PrimitiveOp @ ComposedOp is consistent with ComposedOp @ PrimitiveOp."""
# PauliOp
op1 = (X ^ Y ^ Z)
op2 = (X ^ Y ^ Z)
op3 = (X ^ Y ^ Z).to_circuit_op()
comp1 = op1 @ ComposedOp([op2, op3])
comp2 = ComposedOp([op3, op2]) @ op1
self.assertListEqual(comp1.oplist, list(reversed(comp2.oplist)))
# CircitOp
op1 = op1.to_circuit_op()
op2 = op2.to_circuit_op()
op3 = op3.to_matrix_op()
comp1 = op1 @ ComposedOp([op2, op3])
comp2 = ComposedOp([op3, op2]) @ op1
self.assertListEqual(comp1.oplist, list(reversed(comp2.oplist)))
# MatrixOp
op1 = op1.to_matrix_op()
op2 = op2.to_matrix_op()
op3 = op3.to_pauli_op()
comp1 = op1 @ ComposedOp([op2, op3])
comp2 = ComposedOp([op3, op2]) @ op1
self.assertListEqual(comp1.oplist, list(reversed(comp2.oplist)))
def test_summed_op_equals(self):
"""Test corner cases of SummedOp's equals function."""
with self.subTest('multiplicative factor'):
self.assertEqual(2 * X, X + X)
with self.subTest('commutative'):
self.assertEqual(X + Z, Z + X)
with self.subTest('circuit and paulis'):
z = CircuitOp(ZGate())
self.assertEqual(Z + z, z + Z)
with self.subTest('matrix op and paulis'):
z = MatrixOp([[1, 0], [0, -1]])
self.assertEqual(Z + z, z + Z)
with self.subTest('matrix multiplicative'):
z = MatrixOp([[1, 0], [0, -1]])
self.assertEqual(2 * z, z + z)
with self.subTest('parameter coefficients'):
expr = Parameter('theta')
z = MatrixOp([[1, 0], [0, -1]])
self.assertEqual(expr * z, expr * z)
with self.subTest('different coefficient types'):
expr = Parameter('theta')
z = MatrixOp([[1, 0], [0, -1]])
self.assertNotEqual(expr * z, 2 * z)
with self.subTest('additions aggregation'):
z = MatrixOp([[1, 0], [0, -1]])
a = z + z + Z
b = 2 * z + Z
c = z + Z + z
self.assertEqual(a, b)
self.assertEqual(b, c)
self.assertEqual(a, c)
def test_circuit_compose_register_independent(self):
"""Test that CircuitOp uses combines circuits independent of the register.
I.e. that is uses ``QuantumCircuit.compose`` over ``combine`` or ``extend``.
"""
op = Z ^ 2
qr = QuantumRegister(2, 'my_qr')
circuit = QuantumCircuit(qr)
composed = op.compose(CircuitOp(circuit))
self.assertEqual(composed.num_qubits, 2)
def test_matrix_op_conversions(self):
"""Test to reveal QiskitError when to_instruction or to_circuit method is called on
parametrized matrix op."""
m = np.array([[0, 0, 1, 0], [0, 0, 0, -1], [1, 0, 0, 0], [0, -1, 0,
0]])
matrix_op = MatrixOp(m, Parameter('beta'))
for method in ['to_instruction', 'to_circuit']:
with self.subTest(method):
# QiskitError: multiplication of Operator with ParameterExpression isn't implemented
self.assertRaises(QiskitError, getattr(matrix_op, method))
def test_primitive_op_to_matrix(self):
"""Test to reveal TypeError: multiplication of 'complex' and 'Parameter' is not
implemented, which is raised on PrimitiveOps with parameter, when to_matrix is called. """
# MatrixOp
m = np.array([[0, 0, 1, 0], [0, 0, 0, -1], [1, 0, 0, 0], [0, -1, 0,
0]])
matrix_op = MatrixOp(m, Parameter('beta'))
# PauliOp
pauli_op = PauliOp(primitive=Pauli(label='XYZ'),
coeff=Parameter('beta'))
self.assertRaises(TypeError, pauli_op.to_matrix)
# CircuitOp
qc = QuantumCircuit(2)
qc.cx(0, 1)
circuit_op = CircuitOp(qc, coeff=Parameter('alpha'))
for operator in [matrix_op, pauli_op, circuit_op]:
with self.subTest(operator):
self.assertRaises(TypeError, operator.to_matrix)
def test_list_op_to_circuit(self):
"""Test if unitary ListOps transpile to circuit. """
# generate unitary matrices of dimension 2,4,8, seed is fixed
np.random.seed(233423)
u2 = unitary_group.rvs(2)
u4 = unitary_group.rvs(4)
u8 = unitary_group.rvs(8)
# pauli matrices as numpy.arrays
x = np.array([[0.0, 1.0], [1.0, 0.0]])
y = np.array([[0.0, -1.0j], [1.0j, 0.0]])
z = np.array([[1.0, 0.0], [0.0, -1.0]])
# create MatrixOp and CircuitOp out of matrices
op2 = MatrixOp(u2)
op4 = MatrixOp(u4)
op8 = MatrixOp(u8)
c2 = op2.to_circuit_op()
# algorithm using only matrix operations on numpy.arrays
xu4 = np.kron(x, u4)
zc2 = np.kron(z, u2)
zc2y = np.kron(zc2, y)
matrix = np.matmul(xu4, zc2y)
matrix = np.matmul(matrix, u8)
matrix = np.kron(matrix, u2)
operator = Operator(matrix)
# same algorithm as above, but using PrimitiveOps
list_op = ((X ^ op4) @ (Z ^ c2 ^ Y) @ op8) ^ op2
circuit = list_op.to_circuit()
# verify that ListOp.to_circuit() outputs correct quantum circuit
self.assertTrue(operator.equiv(circuit),
"ListOp.to_circuit() outputs wrong circuit!")
def test_composed_op_to_circuit(self):
"""
Test if unitary ComposedOp transpile to circuit and represents expected operator.
Test if to_circuit on non-unitary ListOp raises exception.
"""
x = np.array([[0.0, 1.0], [1.0, 0.0]]) # Pauli X as numpy array
y = np.array([[0.0, -1.0j], [1.0j, 0.0]]) # Pauli Y as numpy array
m1 = np.array([[0, 0, 1, 0], [0, 0, 0, -1], [0, 0, 0, 0],
[0, 0, 0, 0]]) # non-unitary
m2 = np.array([[0, 0, 0, 0], [0, 0, 0, 0], [1, 0, 0, 0],
[0, -1, 0, 0]]) # non-unitary
m_op1 = MatrixOp(m1)
m_op2 = MatrixOp(m2)
pm1 = (X ^ Y) ^ m_op1 # non-unitary TensoredOp
pm2 = (X ^ Y) ^ m_op2 # non-unitary TensoredOp
self.assertRaises(ExtensionError, pm1.to_circuit)
self.assertRaises(ExtensionError, pm2.to_circuit)
summed_op = pm1 + pm2 # unitary SummedOp([TensoredOp, TensoredOp])
circuit = summed_op.to_circuit(
) # should transpile without any exception
# same algorithm that leads to summed_op above, but using only arrays and matrix operations
unitary = np.kron(np.kron(x, y), m1 + m2)
self.assertTrue(Operator(unitary).equiv(circuit))
def test_op_to_circuit_with_parameters(self):
"""On parametrized SummedOp, to_matrix_op returns ListOp, instead of MatrixOp. To avoid
the infinite recursion, AquaError is raised. """
m1 = np.array([[0, 0, 1, 0], [0, 0, 0, -1], [0, 0, 0, 0],
[0, 0, 0, 0]]) # non-unitary
m2 = np.array([[0, 0, 0, 0], [0, 0, 0, 0], [1, 0, 0, 0],
[0, -1, 0, 0]]) # non-unitary
op1_with_param = MatrixOp(m1, Parameter('alpha')) # non-unitary
op2_with_param = MatrixOp(m2, Parameter('beta')) # non-unitary
summed_op_with_param = op1_with_param + op2_with_param # unitary
self.assertRaises(
AquaError,
summed_op_with_param.to_circuit) # should raise Aqua error
@data(Z, CircuitOp(ZGate()), MatrixOp([[1, 0], [0, -1]]))
def test_op_hashing(self, op):
"""Regression test against faulty set comparison.
Set comparisons rely on a hash table which requires identical objects to have identical
hashes. Thus, the PrimitiveOp.__hash__ should support this requirement.
"""
self.assertEqual(set([2 * op]), set([2 * op]))
@data(Z, CircuitOp(ZGate()), MatrixOp([[1, 0], [0, -1]]))
def test_op_indent(self, op):
"""Test that indentation correctly adds INDENTATION at the beginning of each line"""
initial_str = str(op)
indented_str = op._indent(initial_str)
starts_with_indent = indented_str.startswith(op.INDENTATION)
self.assertTrue(starts_with_indent)
indented_str_content = (indented_str[len(op.INDENTATION):]).split(
"\n{}".format(op.INDENTATION))
self.assertListEqual(indented_str_content, initial_str.split("\n"))
def test_composed_op_immutable_under_eval(self):
"""Test ``ComposedOp.eval`` does not change the operator instance."""
op = 2 * ComposedOp([X])
_ = op.eval()
# previous bug: after op.eval(), op was 2 * ComposedOp([2 * X])
self.assertEqual(op, 2 * ComposedOp([X]))
def test_op_parameters(self):
"""Test that Parameters are stored correctly"""
phi = Parameter('φ')
theta = ParameterVector(name='θ', length=2)
qc = QuantumCircuit(2)
qc.rz(phi, 0)
qc.rz(phi, 1)
for i in range(2):
qc.rx(theta[i], i)
qc.h(0)
qc.x(1)
l = Parameter('λ')
op = PrimitiveOp(qc, coeff=l)
params = set([phi, l, *theta.params])
self.assertEqual(params, op.parameters)
self.assertEqual(params, StateFn(op).parameters)
self.assertEqual(params, StateFn(qc, coeff=l).parameters)
def test_list_op_parameters(self):
"""Test that Parameters are stored correctly in a List Operator"""
lam = Parameter('λ')
phi = Parameter('φ')
omega = Parameter('ω')
mat_op = PrimitiveOp([[0, 1], [1, 0]], coeff=omega)
qc = QuantumCircuit(1)
qc.rx(phi, 0)
qc_op = PrimitiveOp(qc)
op1 = SummedOp([mat_op, qc_op])
params = [phi, omega]
self.assertEqual(op1.parameters, set(params))
# check list nesting case
op2 = PrimitiveOp([[1, 0], [0, -1]], coeff=lam)
list_op = ListOp([op1, op2])
params.append(lam)
self.assertEqual(list_op.parameters, set(params))
@data(VectorStateFn([1, 0]), DictStateFn({'0': 1}),
CircuitStateFn(QuantumCircuit(1)), OperatorStateFn(I),
OperatorStateFn(MatrixOp([[1, 0], [0, 1]])),
OperatorStateFn(CircuitOp(QuantumCircuit(1))))
def test_statefn_eval(self, op):
"""Test calling eval on StateFn returns the statevector."""
expected = Statevector([1, 0])
self.assertEqual(op.eval().primitive, expected)