Source code for simulaqron.virtual_node.qutip_simulator

#
# Copyright (c) 2017, Stephanie Wehner and Axel Dahlberg
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
# 1. Redistributions of source code must retain the above copyright
#    notice, this list of conditions and the following disclaimer.
# 2. Redistributions in binary form must reproduce the above copyright
#    notice, this list of conditions and the following disclaimer in the
#    documentation and/or other materials provided with the distribution.
# 3. All advertising materials mentioning features or use of this software
#    must display the following acknowledgement:
#    This product includes software developed by Stephanie Wehner, QuTech.
# 4. Neither the name of the QuTech organization nor the
#    names of its contributors may be used to endorse or promote products
#    derived from this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY <COPYRIGHT HOLDER> ''AS IS'' AND ANY
# EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
# WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL <COPYRIGHT HOLDER> BE LIABLE FOR ANY
# DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
# (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
# LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
# ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
# SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
import cmath
import logging
import math
from typing import Tuple, List

import numpy as np
from qutip import Qobj

try:
    import qutip as qp
    import qutip.qip.operations.gates as gate_ops

    # qutip-qip >= 0.3 renamed/moved gate expansion helpers.
    # Patch the old names as shims so the rest of this file needs no changes.
    if not hasattr(gate_ops, "gate_expand_1toN"):
        from qutip_qip.operations import expand_operator as _expand_op
        gate_ops.gate_expand_1toN = lambda U, N, t: _expand_op(U, N, t)

    if not hasattr(qp, "gate_expand_2toN"):
        from qutip_qip.operations import expand_operator as _expand_op
        qp.gate_expand_2toN = lambda U, N, t1, t2: _expand_op(U, N, [t1, t2])

except ImportError:
    raise RuntimeError("If you want to use the qutip backend you need to install simulaqron "
                       "with the optional dependencies: 'pip install simulaqron[opt]'")

from simulaqron.virtual_node.basics import QuantumEngine, QuantumError, NoQubitError


[docs]class QutipEngine(QuantumEngine): """ Basic quantum engine which uses QuTip. Works with density matrices and in principle allows full quantum dynamics via QuTip. Subsequently, this is quite slow. """ def __init__(self, node: str, num: int, maxQubits: int = 10): """ Initializes the Qutip engine. :param node: Node name this register is started from. :type node: str :param num: Number of this register. :type num: int :param maxQubits: Maximum number of qubits this engine will support. :type maxQubits: int """ super().__init__(node=node, num=num, maxQubits=maxQubits) # We start with no active qubits self.activeQubits = 0 self.qubitReg = qp.Qobj()
[docs] def add_fresh_qubit(self) -> int: """ Add a new qubit initialized in the :math:`|0>` state. :return: The ID of the new qubit allocated. :rtype: int """ # Prepare a clean qubit state in |0> v = qp.basis(2, 0) newQubit = v * v.dag() num = self.add_qubit(newQubit) return num
[docs] def add_qubit(self, newQubit): """ Add new qubit in the state described by the density matrix newQubit. :param newQubit: The density matrix of the new qubit. :return: The ID of the new qubit allocated. :rtype: int """ # Check if we are still allowed to add qubits if self.activeQubits >= self.maxQubits: raise NoQubitError("No more qubits available in register.") # Append to the existing state at the end if self.activeQubits > 0: self.qubitReg = qp.tensor(self.qubitReg, newQubit) else: self.qubitReg = newQubit # Index number of that qubit num = self.activeQubits # Increment the number of qubits self.activeQubits = self.activeQubits + 1 return num
[docs] def remove_qubit(self, qubitNum: int): """ Removes the qubit with the desired number qubitNum. :param qubitNum: Qubit number :type qubitNum: int """ if (qubitNum + 1) > self.activeQubits: raise QuantumError("No such qubit to remove") # Check if this the only qubit if self.activeQubits == 1: self.activeQubits = 0 self.qubitReg = qp.Qobj() return # Compute the list of qubits to keep keepList = [] for j in range(self.activeQubits): if j != qubitNum: keepList.append(j) # Trace out this qubit by taking the partial trace self.qubitReg = self.qubitReg.ptrace(keepList) # Update the number of qubits self.activeQubits = self.activeQubits - 1
[docs] def get_qubits_RI(self, qList: List[int]) -> Tuple[List[float], List[float]]: """ Retrieves the qubits in the list and returns the result as a list divided into a real and imaginary part. Twisted only likes to send real values lists, not complex ones. :param qList: List of qubits to retrieve, e.g. [1, 4] :type qList: List[int] :return: The qubits states real and imaginary parts. :rtype: Tuple[List[float], List[float]] """ rho = self.get_qubits(qList) Re = rho.full().real.tolist() Im = rho.full().imag.tolist() return (Re, Im)
[docs] def get_register_RI(self) -> Tuple[List[float], List[float]]: """ Retrieves the entire register in real and imaginary parts and returns the result as a list. Twisted only likes to send real valued lists, not complex ones. :return: The qubit states real and imaginary parts. :rtype: Tuple[List[float], List[float]] """ Re = self.qubitReg.full().real.tolist() Im = self.qubitReg.full().imag.tolist() return Re, Im
[docs] def get_density_matrix_RI(self) -> Tuple[Tuple[float], Tuple[float]]: """ Retrieves the density matrix of the qubit as a real and imaginary part. Twisted only likes to send real valued lists, not complex ones. :return: The qubit density matrix real and imaginary parts. :rtype: Tuple[List[float], List[float]] """ Re = self.qubitReg.full().real.tolist() Im = self.qubitReg.full().imag.tolist() return Re, Im # Qutip uses density matrices as the internal representation, so we don't need # to compute the outer product to get the result real_part, im_part = self.get_register_RI() return tuple(*real_part), tuple(*im_part)
[docs] def apply_H(self, qubitNum): """ Applies a Hadamard gate to the qubits with number qubitNum. :param qubitNum: Qubit number :type qubitNum: int """ f = math.sqrt(2) H = qp.Qobj([[1 / f, 1 / f], [1 / f, -1 / f]], dims=[[2], [2]]) self.apply_onequbit_gate(H, qubitNum)
[docs] def apply_K(self, qubitNum): """ Applies a K gate to the qubits with number qubitNum. Maps computational basis to Y eigenbasis. :param qubitNum: Qubit number :type qubitNum: int """ f = math.sqrt(2) i = complex(0, 1) K = qp.Qobj([[1 / f, -i / f], [i / f, -1 / f]], dims=[[2], [2]]) self.apply_onequbit_gate(K, qubitNum)
[docs] def apply_X(self, qubitNum): """ Applies a X gate to the qubits with number qubitNum. :param qubitNum: Qubit number :type qubitNum: int """ X = qp.Qobj([[0, 1], [1, 0]], dims=[[2], [2]]) self.apply_onequbit_gate(X, qubitNum)
[docs] def apply_Z(self, qubitNum): """ Applies a Z gate to the qubits with number qubitNum. :param qubitNum: Qubit number :type qubitNum: int """ Z = qp.Qobj([[1, 0], [0, -1]], dims=[[2], [2]]) self.apply_onequbit_gate(Z, qubitNum)
[docs] def apply_Y(self, qubitNum): """ Applies a Y gate to the qubits with number qubitNum. :param qubitNum: Qubit number :type qubitNum: int """ i = complex(0, 1) Y = qp.Qobj([[0, -i], [i, 0]], dims=[[2], [2]]) self.apply_onequbit_gate(Y, qubitNum)
[docs] def apply_T(self, qubitNum): """ Applies a T gate to the qubits with number qubitNum. :param qubitNum: Qubit number :type qubitNum: int """ i = complex(0, 1) Y = qp.Qobj([[1, 0], [0, cmath.exp(i * np.pi / 4)]], dims=[[2], [2]]) self.apply_onequbit_gate(Y, qubitNum)
[docs] def apply_rotation(self, qubitNum, n, a): """ Applies a rotation around the axis n with the angle a to qubit with number qubitNum. If n is zero a ValueError is raised. :param qubitNum: Qubit number :type qubitNum: int :param n: A tuple of three numbers specifying the rotation axis, e.g n=(1,0,0) :type n: Tuple[float, float, float] :param a: The rotation angle in radians. :type a: float """ nNorm = np.linalg.norm(n) if nNorm == 0: raise ValueError("Rotation vector n can't be 0") R = (-1j * a / (2 * nNorm) * (n[0] * qp.sigmax() + n[1] * qp.sigmay() + n[2] * qp.sigmaz())).expm() self.apply_onequbit_gate(R, qubitNum)
[docs] def apply_CNOT(self, qubitNum1: int, qubitNum2: int): """ Applies the CNOT to the qubit with the numbers qubitNum1 and qubitNum2. :param qubitNum1: Qubit number 1. :type qubitNum1: int :param qubitNum1: Qubit number 2. :type qubitNum1: int """ # Construct the CNOT matrix cnot = qp.Qobj([[1, 0, 0, 0], [0, 1, 0, 0], [0, 0, 0, 1], [0, 0, 1, 0]], dims=[[2, 2], [2, 2]]) # Apply it to the desired qubits self.apply_twoqubit_gate(cnot, qubitNum1, qubitNum2)
[docs] def apply_CPHASE(self, qubitNum1, qubitNum2): """ Applies the CPHASE to the qubit with the numbers qubitNum1 and qubitNum2. :param qubitNum1: Qubit number 1. :type qubitNum1: int :param qubitNum1: Qubit number 2. :type qubitNum1: int """ # Construct the CPHASE matrix cphase = qp.Qobj([[1, 0, 0, 0], [0, 1, 0, 0], [0, 0, 1, 0], [0, 0, 0, -1]], dims=[[2, 2], [2, 2]]) # Apply it to the desired qubits self.apply_twoqubit_gate(cphase, qubitNum1, qubitNum2)
[docs] def get_qubits(self, list: List[int]): """ Returns the qubits with numbers in list. :param list: List of qubits to retrieve. :type list: List[int] """ # Qutip distinguishes between system dimensionality and matrix dimensionality # so we need to make sure it knows we are talking about multiple qubits k = int(math.log2(self.qubitReg.shape[0])) dimL = [] for j in range(k): dimL.append(2) self.qubitReg.dims = [dimL, dimL] logging.debug("Dimensions %s", self.qubitReg.dims) return self.qubitReg.ptrace(list)
[docs] def apply_onequbit_gate(self, gateU: Qobj, qubitNum: int): """ Applies a unitary gate to the specified qubit. :param gateU: Unitary to apply as Qobj. :type gateU: Qobj :param qubitNum: The number of the qubit this gate is applied to :type qubitNum: int """ overallU = gate_ops.gate_expand_1toN(gateU, self.activeQubits, qubitNum) # Compute the overall unitary, identity everywhere with gateU at position qubitNum # Qutip distinguishes between system dimensionality and matrix dimensionality # so we need to make sure it knows we are talking about multiple qubits k = int(math.log2(overallU.shape[0])) dimL = [] for j in range(k): dimL.append(2) overallU.dims = [dimL, dimL] self.qubitReg.dims = [dimL, dimL] # Apply the unitary self.qubitReg = overallU * self.qubitReg * overallU.dag()
[docs] def apply_twoqubit_gate(self, gateU: Qobj, qubit1: int, qubit2: int): """ Applies a unitary gate to the two specified qubits. :param gateU: Unitary to apply as Qobj. :type gateU: Qobj :param qubit1: The first qubit :type qubit1: int :param qubit2: The second qubit :type qubit2: int """ # Construct the overall unitary overallU = qp.gate_expand_2toN(gateU, self.activeQubits, qubit1, qubit2) # Qutip distinguishes between system dimensionality and matrix dimensionality # so we need to make sure it knows we are talking about multiple qubits k = int(math.log2(overallU.shape[0])) dimL = [] for j in range(k): dimL.append(2) overallU.dims = [dimL, dimL] self.qubitReg.dims = [dimL, dimL] # Apply the unitary self.qubitReg = overallU * self.qubitReg * overallU.dag()
[docs] def measure_qubit_inplace(self, qubitNum: int): """ Measures the desired qubit in the standard basis. This returns the classical outcome. The quantum register is in the post-measurement state corresponding to the obtained outcome. :param qubitNum: The number of the qubit to measure. :type qubitNum: int """ # Check we have such a qubit... if (qubitNum + 1) > self.activeQubits: raise QuantumError("No such qubit to be measured.") # Construct the two measurement operators, and put them at the right position v0 = qp.basis(2, 0) P0 = v0 * v0.dag() M0 = gate_ops.gate_expand_1toN(P0, self.activeQubits, qubitNum) v1 = qp.basis(2, 1) P1 = v1 * v1.dag() M1 = gate_ops.gate_expand_1toN(P1, self.activeQubits, qubitNum) # Compute the success probabilities obj = M0 * self.qubitReg p0 = obj.tr().real obj = M1 * self.qubitReg p1 = obj.tr().real # Clamp and renormalize to handle tiny negative values that can arise # from floating-point rounding after multi-qubit gate sequences. p0 = max(0.0, p0) p1 = max(0.0, p1) total = p0 + p1 if total > 0: p0 /= total p1 /= total else: p0, p1 = 0.5, 0.5 # Sample the measurement outcome from these probabilities outcome = np.random.choice([0, 1], p=[p0, p1]).item() # Compute the post-measurement state, getting rid of the measured qubit if outcome == 0: self.qubitReg = M0 * self.qubitReg * M0.dag() / p0 else: self.qubitReg = M1 * self.qubitReg * M1.dag() / p1 # return measurement outcome return outcome
[docs] def measure_qubit(self, qubitNum: int): """ Measures the desired qubit in the standard basis. This returns the classical outcome and deletes the qubit. :param qubitNum: The number of the qubit to measure. :type qubitNum: int """ outcome = self.measure_qubit_inplace(qubitNum) self.remove_qubit(qubitNum) return outcome
[docs] def replace_qubit(self, qubitNum: int, state): """ Replaces the qubit at position qubitNum with the one given by state. :param qubitNum: Qubit to be replaced :type qubitNum: int :param state: New state to write in the place of the old qubit. :type state: Any """ # Remove the qubit currently there by tracing it out self.remove_qubit(qubitNum) # Tensor on the new qubit at the end self.add_qubit(state) # Put the new qubit in the correct position qList = list(range(self.activeQubits)) qList[qubitNum] = self.activeQubits qList[self.activeQubits - 1] = qubitNum self.qubitReg.permute(qList)
[docs] def absorb(self, other): """ Absorb the qubits from the other engine into this one. This is done by tensoring the state at the end. :param other: The other qubit to absorb. :type other: int """ # Check whether there is space newNum = self.activeQubits + other.activeQubits if newNum > self.maxQubits: raise QuantumError("Cannot merge: qubits exceed the maximum available.\n") # Check whether there are in fact qubits to tensor up.... if self.activeQubits == 0: self.qubitReg = other.qubitReg elif other.activeQubits != 0: self.qubitReg = qp.tensor(self.qubitReg, other.qubitReg) self.activeQubits = newNum
[docs] def absorb_parts(self, R, I, activeQ): """ Absorb the qubits, given in pieces :param R: Real part of the qubit state as a list. :type R: List[float] :param I: Imaginary part as a list. :type I: List[float] :param activeQ: Active number of qubits """ # Convert the real and imaginary parts given as lists into a qutip object M = I for s in range(len(I)): for t in range(len(I)): M[s][t] = R[s][t] + I[s][t] * 1j qt = qp.Qobj(M) # Check whether there is space newNum = self.activeQubits + activeQ if newNum > self.maxQubits: raise QuantumError("Cannot merge: qubits exceed the maximum available.\n") # Check whether there are in fact qubits to tensor up.... if self.activeQubits == 0: self.qubitReg = qt elif qt.shape[0] != 0: self.qubitReg = qp.tensor(self.qubitReg, qt) self.activeQubits = newNum # Qutip distinguishes between system dimensionality and matrix dimensionality # so we need to make sure it knows we are talking about multiple qubits k = int(math.log2(self.qubitReg.shape[0])) dimL = [] for j in range(k): dimL.append(2) self.qubitReg.dims = [dimL, dimL]