def create_input_gate(self, x, uin_type):
# Encode x into quantum state
# x = 2-dim. variables, [-1,1]
u = QuantumCircuit(self.nqubit)
angle_y = np.arcsin(x)
angle_z = np.arccos(x**2)
if uin_type == 0:
for i in range(self.nqubit):
u.add_RY_gate(i, angle_y[i])
u.add_RZ_gate(i, angle_z[i])
elif uin_type == 1:
#for d in range(2):
for i in range(self.nqubit):
u.add_H_gate(i)
u.add_RY_gate(i, angle_y[i])
u.add_RZ_gate(i, angle_z[i])
# KT: add second order expansion
for i in range(self.nqubit - 1):
for j in range(i + 1, self.nqubit):
angle_z2 = np.arccos(x[i] * x[j])
u.add_CNOT_gate(i, j)
u.add_RZ_gate(j, angle_z2)
u.add_CNOT_gate(i, j)
return u
def U_in(x):
U = QuantumCircuit(nqubit)
angle_y = np.arcsin(x)
angle_z = np.arccos(x**2)
for i in range(nqubit):
U.add_RY_gate(i, angle_y)
U.add_RZ_gate(i, angle_z)
return U
def main():
import numpy as np
n_qubit = 2
obs = Observable(n_qubit)
initial_state = QuantumState(n_qubit)
obs.add_operator(1, "Z 0 Z 1")
circuit_list = []
p_list = [0.02, 0.04, 0.06, 0.08]
#prepare circuit list
for p in p_list:
circuit = QuantumCircuit(n_qubit)
circuit.add_H_gate(0)
circuit.add_RY_gate(1, np.pi / 6)
circuit.add_CNOT_gate(0, 1)
circuit.add_gate(
Probabilistic([p / 4, p / 4, p / 4],
[X(0), Y(0), Z(0)])) #depolarizing noise
circuit.add_gate(
Probabilistic([p / 4, p / 4, p / 4],
[X(1), Y(1), Z(1)])) #depolarizing noise
circuit_list.append(circuit)
#get mitigated output
mitigated, non_mitigated_array, fit_coefs = error_mitigation_extrapolate_linear(
circuit_list,
p_list,
initial_state,
obs,
n_circuit_sample=100000,
return_full=True)
#plot the result
p = np.linspace(0, max(p_list), 100)
plt.plot(p,
fit_coefs[0] * p + fit_coefs[1],
linestyle="--",
label="linear fit")
plt.scatter(p_list, non_mitigated_array, label="un-mitigated")
plt.scatter(0, mitigated, label="mitigated output")
#prepare the clean result
state = QuantumState(n_qubit)
circuit = QuantumCircuit(n_qubit)
circuit.add_H_gate(0)
circuit.add_RY_gate(1, np.pi / 6)
circuit.add_CNOT_gate(0, 1)
circuit.update_quantum_state(state)
plt.scatter(0, obs.get_expectation_value(state), label="True output")
plt.xlabel("error rate")
plt.ylabel("expectation value")
plt.legend()
plt.show()
def create_input_gate(self, x):
# 単一のxをエンコードするゲートを作成する関数
# xは入力特徴量(2次元)
# xの要素は[-1, 1]の範囲内
u = QuantumCircuit(self.nqubit)
angle_y = np.arcsin(x)
angle_z = np.arccos(x**2)
for i in range(self.nqubit):
if i % 2 == 0:
u.add_RY_gate(i, angle_y[0])
u.add_RZ_gate(i, angle_z[0])
else:
u.add_RY_gate(i, angle_y[1])
u.add_RZ_gate(i, angle_z[1])
return u
def create_input_gate(self, x):
# Encode x into quantum state
# x = 2-dim. variables, [-1,1]
u = QuantumCircuit(self.nqubit)
angle_y = np.arcsin(x)
angle_z = np.arccos(x**2)
for i in range(self.nqubit):
'''
if i % 2 == 0:
u.add_RY_gate(i, angle_y[0])
u.add_RZ_gate(i, angle_z[0])
else:
u.add_RY_gate(i, angle_y[1])
u.add_RZ_gate(i, angle_z[1])
'''
u.add_RY_gate(i, angle_y[i])
u.add_RZ_gate(i, angle_z[i])
return u
def _try_append_gate(self, op: ops.GateOperation,
qulacs_circuit: qulacs.QuantumCircuit,
indices: np.array):
# One qubit gate
if isinstance(op.gate, ops.pauli_gates._PauliX):
qulacs_circuit.add_X_gate(indices[0])
elif isinstance(op.gate, ops.pauli_gates._PauliY):
qulacs_circuit.add_Y_gate(indices[0])
elif isinstance(op.gate, ops.pauli_gates._PauliZ):
qulacs_circuit.add_Z_gate(indices[0])
elif isinstance(op.gate, ops.common_gates.HPowGate):
qulacs_circuit.add_H_gate(indices[0])
elif isinstance(op.gate, ops.common_gates.XPowGate):
qulacs_circuit.add_RX_gate(indices[0], -np.pi * op.gate._exponent)
elif isinstance(op.gate, ops.common_gates.YPowGate):
qulacs_circuit.add_RY_gate(indices[0], -np.pi * op.gate._exponent)
elif isinstance(op.gate, ops.common_gates.ZPowGate):
qulacs_circuit.add_RZ_gate(indices[0], -np.pi * op.gate._exponent)
elif isinstance(op.gate, ops.SingleQubitMatrixGate):
mat = op.gate._matrix
qulacs_circuit.add_dense_matrix_gate(indices[0], mat)
elif isinstance(op.gate, circuits.qasm_output.QasmUGate):
lmda = op.gate.lmda
theta = op.gate.theta
phi = op.gate.phi
gate = qulacs.gate.U3(indices[0], theta * np.pi, phi * np.pi,
lmda * np.pi)
qulacs_circuit.add_gate(gate)
# Two qubit gate
elif isinstance(op.gate, ops.common_gates.CNotPowGate):
if op.gate._exponent == 1.0:
qulacs_circuit.add_CNOT_gate(indices[0], indices[1])
else:
mat = _get_google_rotx(op.gate._exponent)
gate = qulacs.gate.DenseMatrix(indices[1], mat)
gate.add_control_qubit(indices[0], 1)
qulacs_circuit.add_gate(gate)
elif isinstance(op.gate, ops.common_gates.CZPowGate):
if op.gate._exponent == 1.0:
qulacs_circuit.add_CZ_gate(indices[0], indices[1])
else:
mat = _get_google_rotz(op.gate._exponent)
gate = qulacs.gate.DenseMatrix(indices[1], mat)
gate.add_control_qubit(indices[0], 1)
qulacs_circuit.add_gate(gate)
elif isinstance(op.gate, ops.common_gates.SwapPowGate):
if op.gate._exponent == 1.0:
qulacs_circuit.add_SWAP_gate(indices[0], indices[1])
else:
qulacs_circuit.add_dense_matrix_gate(indices, op._unitary_())
elif isinstance(op.gate, ops.parity_gates.XXPowGate):
qulacs_circuit.add_multi_Pauli_rotation_gate(
indices, [1, 1], -np.pi * op.gate._exponent)
elif isinstance(op.gate, ops.parity_gates.YYPowGate):
qulacs_circuit.add_multi_Pauli_rotation_gate(
indices, [2, 2], -np.pi * op.gate._exponent)
elif isinstance(op.gate, ops.parity_gates.ZZPowGate):
qulacs_circuit.add_multi_Pauli_rotation_gate(
indices, [3, 3], -np.pi * op.gate._exponent)
elif isinstance(op.gate, ops.TwoQubitMatrixGate):
indices.reverse()
mat = op.gate._matrix
qulacs_circuit.add_dense_matrix_gate(indices, mat)
# Three qubit gate
"""
# deprecated because these functions cause errors in gpu
elif isinstance(op.gate, ops.three_qubit_gates.CCXPowGate):
mat = _get_google_rotx(op.gate._exponent)
gate = qulacs.gate.DenseMatrix(indices[2], mat)
gate.add_control_qubit(indices[0],1)
gate.add_control_qubit(indices[1],1)
qulacs_circuit.add_gate(gate)
elif isinstance(op.gate, ops.three_qubit_gates.CCZPowGate):
mat = _get_google_rotz(op.gate._exponent)
gate = qulacs.gate.DenseMatrix(indices[2], mat)
gate.add_control_qubit(indices[0],1)
gate.add_control_qubit(indices[1],1)
qulacs_circuit.add_gate(gate)
"""
elif isinstance(op.gate, ops.three_qubit_gates.CSwapGate):
mat = np.zeros(shape=(4, 4))
mat[0, 0] = 1
mat[1, 2] = 1
mat[2, 1] = 1
mat[3, 3] = 1
gate = qulacs.gate.DenseMatrix(indices[1:], mat)
gate.add_control_qubit(indices[0], 1)
qulacs_circuit.add_gate(gate)
# Misc
elif protocols.has_unitary(op):
indices.reverse()
mat = op._unitary_()
qulacs_circuit.add_dense_matrix_gate(indices, mat)
# Not unitary
else:
return False
return True
def create_input_gate(self, x, uin_type):
# Encode x into quantum state
# uin_type: unitary data-input type 0, 1, 20, 21, 30, 31, 40, 41, 50, 51, 60, 61
# x = 1dim. variables, [-1,1]
I_mat = np.eye(2, dtype=complex)
X_mat = X(0).get_matrix()
Y_mat = Y(0).get_matrix()
Z_mat = Z(0).get_matrix()
#make operators s.t. exp(i*theta * sigma^z_j@sigma^z_k) @:tensor product
def ZZ(u, theta, j, k):
u.add_CNOT_gate(j, k)
u.add_RZ_gate(k, -2 * theta * self.time_step)
u.add_CNOT_gate(j, k)
return u
def XX(u, theta, j, k):
u.add_H_gate(j)
u.add_H_gate(k)
ZZ(u, theta, j, k)
u.add_H_gate(j)
u.add_H_gate(k)
return u
def YY(u, theta, j, k):
u.add_U1_gate(j, -np.pi / 2.)
u.add_U1_gate(k, -np.pi / 2.)
XX(u, theta, j, k)
u.add_U1_gate(j, np.pi / 2.)
u.add_U1_gate(k, np.pi / 2.)
return u
theta = x
u = QuantumCircuit(self.nqubit)
angle_y = np.arcsin(x)
angle_z = np.arccos(x**2)
if uin_type == 0:
for i in range(self.nqubit):
u.add_RY_gate(i, angle_y[i])
u.add_RZ_gate(i, angle_z[i])
elif uin_type == 1:
#for d in range(2):
for i in range(self.nqubit):
u.add_H_gate(i)
u.add_RY_gate(i, angle_y[i])
u.add_RZ_gate(i, angle_z[i])
# KT: add second order expansion
for i in range(self.nqubit - 1):
for j in range(i + 1, self.nqubit):
angle_z2 = np.arccos(x[i] * x[j])
u.add_CNOT_gate(i, j)
u.add_RZ_gate(j, angle_z2)
u.add_CNOT_gate(i, j)
elif uin_type == 20:
for i in range(self.nqubit):
u.add_RX_gate(i, -2 * x[i] * self.time_step)
elif uin_type == 21:
ham = np.zeros((2**self.nqubit, 2**self.nqubit), dtype=complex)
for i in range(self.nqubit): # i runs 0 to nqubit-1
J_x = x[i]
print(x)
ham += J_x * make_fullgate([[i, X_mat]], self.nqubit)
## Build time-evolution operator by diagonalizing the Ising hamiltonian H*P = P*D <-> H = P*D*P^dagger
diag, eigen_vecs = np.linalg.eigh(ham)
time_evol_op = np.dot(
np.dot(eigen_vecs,
np.diag(np.exp(-1j * self.time_step * diag))),
eigen_vecs.T.conj()) # e^-iHT
# Convert to qulacs gate
time_evol_gate = DenseMatrix([i for i in range(self.nqubit)],
time_evol_op)
u.add_gate(time_evol_gate)
elif uin_type == 30:
#Ising hamiltonian with input coefficient
# nearest neighbor spin-conbination has interaction
for i in range(self.nqubit):
u.add_RX_gate(i, -2 * x[i] * self.time_step)
ZZ(u, theta[i] * theta[(i + 1) % self.nqubit], i, i + 1)
elif uin_type == 31:
ham = np.zeros((2**self.nqubit, 2**self.nqubit), dtype=complex)
for i in range(self.nqubit):
J_x = x[i]
ham += J_x * make_fullgate([[i, X_mat]], self.nqubit)
J_zz = x[i] * x[(i + 1) % self.nqubit]
ham += J_zz * make_fullgate(
[[i, Z_mat], [(i + 1) % self.nqubit, Z_mat]], self.nqubit)
diag, eigen_vecs = np.linalg.eigh(ham)
time_evol_op = np.dot(
np.dot(eigen_vecs,
np.diag(np.exp(-1j * self.time_step * diag))),
eigen_vecs.T.conj())
time_evol_gate = DenseMatrix([i for i in range(self.nqubit)],
time_evol_op)
u.add_gate(time_evol_gate)
elif uin_type == 40:
#Ising hamiltonian with input coefficient
# every two possible spin-conbination has interaction
for i in range(self.nqubit):
u.add_RX_gate(i, -2 * x[i] * self.time_step)
for j in range(i + 1, self.nqubit):
ZZ(u, theta[i] * theta[j], i, j)
elif uin_type == 41:
ham = np.zeros((2**self.nqubit, 2**self.nqubit), dtype=complex)
for i in range(self.nqubit):
J_x = x[i]
ham += J_x * make_fullgate([[i, X_mat]], self.nqubit)
for j in range(i + 1, self.nqubit):
J_ij = x[i] * x[j]
ham += J_ij * make_fullgate([[i, Z_mat], [j, Z_mat]],
self.nqubit)
diag, eigen_vecs = np.linalg.eigh(ham)
time_evol_op = np.dot(
np.dot(eigen_vecs,
np.diag(np.exp(-1j * self.time_step * diag))),
eigen_vecs.T.conj())
time_evol_gate = DenseMatrix([i for i in range(self.nqubit)],
time_evol_op)
u.add_gate(time_evol_gate)
elif uin_type == 50:
#Heisenberg hamiltonian with input coefficient
# nearest neighbor spin-conbination has interaction
for i in range(self.nqubit):
u.add_RX_gate(i, -2 * x[i] * self.time_step)
XX(u, theta[i] * theta[(i + 1) % self.nqubit], i, i + 1)
YY(u, theta[i] * theta[(i + 1) % self.nqubit], i, i + 1)
ZZ(u, theta[i] * theta[(i + 1) % self.nqubit], i, i + 1)
elif uin_type == 51:
ham = np.zeros((2**self.nqubit, 2**self.nqubit), dtype=complex)
for i in range(self.nqubit):
J_x = x[i]
ham += J_x * make_fullgate([[i, X_mat]], self.nqubit)
J_xx = x[i] * x[(i + 1) % self.nqubit]
J_yy = x[i] * x[(i + 1) % self.nqubit]
J_zz = x[i] * x[(i + 1) % self.nqubit]
ham += J_xx * make_fullgate(
[[i, X_mat], [(i + 1) % self.nqubit, X_mat]], self.nqubit)
ham += J_yy * make_fullgate(
[[i, Y_mat], [(i + 1) % self.nqubit, Y_mat]], self.nqubit)
ham += J_xx * make_fullgate(
[[i, Z_mat], [(i + 1) % self.nqubit, Z_mat]], self.nqubit)
diag, eigen_vecs = np.linalg.eigh(ham)
time_evol_op = np.dot(
np.dot(eigen_vecs,
np.diag(np.exp(-1j * self.time_step * diag))),
eigen_vecs.T.conj())
time_evol_gate = DenseMatrix([i for i in range(self.nqubit)],
time_evol_op)
u.add_gate(time_evol_gate)
elif uin_type == 60:
#Heisenberg hamiltonian with input coefficient
# every two possible spin-conbination has interaction
for i in range(self.nqubit):
u.add_RX_gate(i, -2 * x[i] * self.time_step)
for j in range(i + 1, self.nqubit):
XX(u, theta[i] * theta[j], i, j)
YY(u, theta[i] * theta[j], i, j)
ZZ(u, theta[i] * theta[j], i, j)
elif uin_type == 61:
ham = np.zeros((2**self.nqubit, 2**self.nqubit), dtype=complex)
for i in range(self.nqubit):
J_x = x[i]
ham += J_x * make_fullgate([[i, X_mat]], self.nqubit)
for j in range(i + 1, self.nqubit):
J_xx = x[i] * x[j]
J_yy = x[i] * x[j]
J_zz = x[i] * x[j]
ham += J_xx * make_fullgate([[i, X_mat], [j, X_mat]],
self.nqubit)
ham += J_yy * make_fullgate([[i, Y_mat], [j, Y_mat]],
self.nqubit)
ham += J_xx * make_fullgate([[i, Z_mat], [j, Z_mat]],
self.nqubit)
diag, eigen_vecs = np.linalg.eigh(ham)
time_evol_op = np.dot(
np.dot(eigen_vecs,
np.diag(np.exp(-1j * self.time_step * diag))),
eigen_vecs.T.conj())
time_evol_gate = DenseMatrix([i for i in range(self.nqubit)],
time_evol_op)
u.add_gate(time_evol_gate)
else:
pass
return u