Quantum Information Processing and Quantum Error Correction by Ivan Djordjevic by Ivan Djordjevic - Read Online

Book Preview

Quantum Information Processing and Quantum Error Correction - Ivan Djordjevic

You've reached the end of this preview. Sign up to read more!
Page 1 of 1

Index

Chapter 1

Introduction

Chapter Outline

1.1 Photon Polarization

1.2 The Concept of the Qubit

1.3 Spin-1/2 Systems

1.4 Quantum Gates and Quantum Information Processing

1.5 Quantum Teleportation

1.6 Quantum Error Correction Concepts

1.7 Quantum Key Distribution (QKD)

1.8 Organization of the Book

References

This chapter introduces quantum information processing (QIP) and quantum error correction coding (QECC) [1–11, 13–16] concepts. Quantum information is related to the use of quantum mechanics concepts to perform information processing and transmission of information. QIP is an exciting research area with numerous applications, including quantum key distribution (QKD), quantum teleportation, quantum computing, quantum lithography, and quantum memories. This area is currently experiencing intensive development and, given the novelty of underlying concepts, it could become of interest to a broad range of scientists, not only those involved in research. Moore’s law claims that the number of chips that can be etched on a single chip doubles every 18 months, leading to doubling of memory and doubling of computational speed. On this basis, extrapolation of this law until 2020 indicates that feature size might fall below 10 nm, and at this point, as the individual properties of atoms and electrons start to dominate, Moore’s law will cease to be valid. Therefore, the ever-increasing demands in miniaturization of electronics will eventually lead to a point when quantum effects become important. Given this fact, it seems that a much broader range of scientists will have to study QIP much sooner than it appears right now. Note that as the multicore architecture is becoming a prevailing high-performance chip design approach, improving computational speed can be achieved even without reducing the feature size through parallelization. Therefore, due to multicore processor architectures, the need for QIP could be prolonged for a certain amount of time. Another turning down point could be the fact that, despite intensive development of quantum algorithms, the number of available quantum algorithms is still small compared to that of classical algorithms. Until recently, it was widely believed that quantum computation would never become a reality. However, recent advances in various quantum gate implementations, as well as the proof of the accuracy threshold theorem, have given rise to optimism that quantum computers will soon become reality.

The fundamental features of QIP are different from those of classical computing and can be broken down into three categories: (1) linear superposition; (2) entanglement; (3) quantum parallelism. Below we provide some basic details of these features:

1. Linear superposition. Contrary to the classical bit, a quantum bit or qubit can take not only two discrete values 0 and 1, but also all possible linear combinations of them. This is a consequence of a fundamental property of quantum states: It is possible to construct a linear superposition of quantum state |0〉 and quantum state |1〉.

2. Entanglement. At a quantum level it appears that two quantum objects can form a single entity, even when they are well separated from each other. Any attempt to consider this entity as a combination of two independent quantum objects given by the tensor product of quantum states fails, unless the possibility of signal propagation at superluminal speed is allowed. These quantum objects that cannot be decomposed into a tensor product of individual independent quantum objects are called entangled quantum objects. Given the fact that arbitrary quantum states cannot be copied, which is the consequence of the no-cloning theorem, communication at superluminal speed is not possible, and as a consequence the entangled quantum states cannot be written as the tensor product of independent quantum states. Moreover, it can be shown that the amount of information contained in an entangled state of N qubits grows exponentially instead of linearly, which is the case for classical bits.

3. Quantum parallelism. This makes it possible to perform a large number of operations in parallel, which represents a key difference from classical computing. Namely, in classical computing it is possible to know the internal status of the computer. On the other hand, because of the no-cloning theorem, it is not possible to know the current state of a quantum computer. This property has led to the development of the Shor factorization algorithm, which can be used to crack the Rivest–Shamir–Adleman (RSA) encryption protocol. Some other important quantum algorithms include: the Grover search algorithm, which is used to perform a search for an entry in an unstructured database; the quantum Fourier transform, which is the basis for a number of different algorithms; and Simon’s algorithm. These algorithms are the subject of Chapter 5. A quantum computer is able to encode all input strings of length N simultaneously into a single computational step. In other words, the quantum computer is able simultaneously to pursue 2N classical paths, indicating that a quantum computer is significantly more powerful than a classical one.

Although the QIP has opened up some fascinating perspectives, as indicated above, there are certain limitations that need to be overcome before QIP becomes a commercial reality. The first is related to the number of existing quantum algorithms, whose number is significantly lower than that of classical algorithms. The second problem is related to physical implementation issues. There are many potential technologies, such as nuclear magnetic resonance (NMR), ion traps, cavity quantum electrodynamics, photonics, quantum dots, and superconducting technologies, to mention just a few. Nevertheless, it is not clear which technology will prevail. Regarding quantum teleportation, most probably the photonic implementation will prevail. On the other hand, for quantum computing applications there are many potential technologies that compete with each other. Moreover, presently the number of qubits that can be manipulated is of the order of tens, well below that needed for meaningful quantum computation, which is of the order of thousands. Another problem, which can be considered as the major difficulty, is related to decoherence. Decoherence is related to the interaction of qubits with environments that blur the fragile superposition states. It also introduces errors, indicating that the quantum register should be sufficiently isolated from the environment so that only few random errors occur occasionally, which can be corrected by QECC techniques. One of the most powerful applications of quantum error correction is the protection of quantum information as it dynamically undergoes quantum computation. Imperfect quantum gates affect quantum computation by introducing errors in computed data. Moreover, the imperfect control gates introduce errors in processed sequences since wrong operations can be applied. The QECC scheme now needs to deal not only with errors introduced by quantum channels, but also with errors introduced by imperfect quantum gates during the encoding/decoding process. Because of this, the reliability of data processed by quantum computers is not a priori guaranteed by QECC. The reason is threefold: (i) the gates used for encoders and decoders are composed of imperfect gates, including controlled imperfect gates; (ii) the syndrome extraction applies unitary operators to entangle ancillary qubits with code block; and (iii) the error recovery action requires the use of controlled operation to correct for these errors. Nevertheless, it can be shown that arbitrary good quantum error protection can be achieved even with imperfect gates, provided that the error probability per gate is below a certain threshold; this claim is known as the accuracy threshold theorem and will be discussed in Chapter 11, together with various fault-tolerant concepts.

This introductory chapter is organized as follows. In Section 1.1, photon polarization is described as it represents the simplest and most natural connection to QIP. In the same section, some basic concepts of quantum mechanics are introduced, such as the concept of state. The Dirac notation is also introduced, and will be used throughout the book. In Section 1.2, the concept of the qubit is formally introduced and its geometric interpretation given. In Section 1.3, another interesting example of a representation of qubits, the spin-1/2 system, is provided. Section 1.4 covers basic quantum gates and QIP fundamentals. The basic concepts of quantum teleportation are introduced in Section 1.5. Section 1.6 considers the basic QECC concepts. Finally, Section 1.7 is devoted to the quantum key distribution (QKD), also known as quantum cryptography.

1.1 Photon Polarization

The electric/magnetic field of plane linearly polarized waves is described as follows [12]:

(1.1)

where E (H) denotes electric (magnetic) field, p denotes the polarization orientation, r = xex + yey + zez is the position vector, and k = kxex + kyey + kzez denotes the wave propagation vector whose magnitude is k = 2π/λ (λ is the operating wavelength). For the x-polarization waves (p = ex, k = kez), Eq. (1.1) becomes

(1.2)

while for y-polarization (p = ey, k = kez) it becomes

(1.3)

where δ is the relative phase difference between the two orthogonal waves. The resultant wave can be obtained by combining (1.2) and (1.3) as follows:

(1.4)

The linearly polarized wave is obtained by setting the phase difference to an integer multiple of 2π

(1.5)

By ignoring the time-dependent term, we can represent the linear polarization as shown in the elliptical polarization is obtained. From Eqs (1.2) and (1.3), by eliminating the time-dependent term we obtain the following equation of the ellipse:

(1.6)

which is shown in , the equation of the ellipse becomes

(1.7)

FIGURE 1.1 Various forms of polarization: (a) linear polarization; (b) elliptic polarization; (c) circular polarization.

, the equation of the ellipse becomes the circle:

(1.8)

and the corresponding polarization is known as circular polarization (see Figure 1.1c). A right circularly polarized wave is obtained for δ = π/2 + 2:

(1.9)

Otherwise, for δ = −π/2 + 2, the polarization is known as left circularly polarized.

Very often, the Jones vector representation of a polarization wave is used:

(1.10)

where k is the power-splitting ratio between states of polarizations (SOPs), with the complex phasor term being typically omitted in practice.

Another interesting representation is the Stokes vector representation:

(1.11)

where the parameter S0 is related to the optical intensity by

(1.12a)

The parameter S1 > 0 is related to the preference for horizontal polarization and is defined by

(1.12b)

The parameter S2 > 0 is related to the preference for π/4 SOP:

(1.12c)

Finally, the parameter S3 > 0 is related to the preference for right-circular polarization and is defined by

(1.12d)

The parameter S0 is related to other Stokes parameters by

(1.13)

The degree of polarization is defined by

(1.14)

For p = 1 the polarization does not change with time. The Stokes vector can be represented in terms of Jones vector parameters as

(1.15)

After the normalization with respect to S0, the normalized Stokes parameters are given by

(1.16)

If the normalized Stokes parameters are used, the polarization state can be represented as a point on a Poincaré sphere, as shown in Figure 1.2. The points located at the opposite sides of the line crossing the center represent the orthogonal polarizations.

FIGURE 1.2 Representation of polarization state as a point on a Poincaré sphere.

The polarization ellipse is very often represented in terms of the ellipticity and the azimuth, which are illustrated in Figure 1.3. The ellipticity is defined by the ratio of half-axis lengths. The corresponding angle is called the ellipticity angle and is denoted by ε. Small ellipticity means that the polarization ellipse is highly elongated, while for zero elipticity the polarization is linear. For ε = ±π/4, the polarization is circular. For ε > 0 the polarization is right-elliptical. On the other hand, the azimuth angle η defines the orientation of the main axis of the ellipse with respect to Ex.

FIGURE 1.3 The ellipticity and azimuth of the polarization ellipse.

The polarization ellipse parameters can be related to the Jones vector parameters by

(1.17)

Finally, the parameters of the polarization ellipse can be related to the Stokes vector parameters by

(1.18)

and the corresponding geometrical interpretation is provided in Figures 1.2 and 1.3.

Let us now observe the polarizer–analyzer ensemble, shown in Figure 1.4. When an electromagnetic wave passes through the polarizer, it can be represented as a vector in the xOy plane transverse to the propagation direction, as given by Eq. (1.5), where the angle θ Eq. (1.5) can be rewritten as

(1.19)

If θ = 0 rad, the light is polarized along the x-axis, while for θ = π/2 rad it is polarized along the y-axis. Natural light is unpolarized as it represents incoherent superposition of 50% of the light polarized along the x-axis and 50% of light polarized along the y-axis. After the analyzer, whose axis makes an angle ϕ with respect to the x, the output electric field is given by

(1.20)

The intensity of the analyzer output field is given by

(1.21)

which is commonly referred to as the Malus law.

FIGURE 1.4 The polarizer–analyzer ensemble for study of photon polarization.

Decomposition of polarization by a birefringent plate is now considered (see Figure 1.5). Experiments show that photodetectors PDx and PDy are never triggered simultaneously, which indicates that an entire photon reaches either PDx or PDy (a photon never splits). Therefore, the corresponding probabilities that a photon is detected by photodetectors PDx and PDy can be determined by

(1.22)

If the total number of photons is N, the number of detected photons in x-polarization will be Nx Ncos² θ and the number of detected photons in y-polarization will be Ny Nsin² θ. In the limit, as N → ∞ we would expect the Malus law to be obtained.

FIGURE 1.5 Polarization decomposition by a birefringent plate. PD, photodetector.

Let us now study the polarization decomposition and recombination by means of birefringent plates, as illustrated in Figure 1.6. Classical physics prediction of the total probability of a photon passing the polarizer–analyzer ensemble is given by

(1.23)

which is inconsistent with the Malus law, given by Eq. (1.21). In order to reconstruct the results from wave optics, it is necessary to introduce into quantum mechanics the concept of probability amplitude, that α is detected as β, which is denoted as a(α → β), and it is a complex number. The probability is obtained as the squared magnitude of probability amplitude:

(1.24)

The relevant probability amplitudes relating to Figure 1.6 are

(1.25)

The basic principle of quantum mechanics is to sum up the probability amplitudes for indistinguishable paths:

(1.26)

The corresponding total probability is

(1.27)

and this result is consistent with the Malus law!

FIGURE 1.6 Polarization decomposition and recombination by a birefringent plate.

Based on the previous discussion, the state vector of the photon polarization is given by

(1.28)

where ψx is related to x-polarization and ψy to y-polarization, with the normalization condition as follows:

(1.29)

In this representation, the x- and y-polarization photons can be represented by

(1.30)

and the right and left circular polarization photons are represented by

(1.31)

In Eqs (1.28)–(1.31) we used Dirac notation to denote the column vectors (kets). In Dirac notation, with each column vector (ket) |ψ〉, we associate a row vector (bra) 〈ψ| as follows:

(1.32)

The scalar (dot) product of ket |ϕ〉 and bra 〈ψ| is defined by bracket as follows:

(1.33)

The normalization condition can be expressed in terms of scalar product by

(1.34)

Based on (1.30) and (1.31), it is evident that

Because the vectors |x〉 and |y〉 are orthogonal, their dot product is zero:

(1.35)

and they form the basis. Any state vector |ψ〉 can be written as a linear superposition of basis kets as follows:

(1.36)

We can use now (1.34) and (1.36) to derive an important relation in quantum mechanics, known as the completeness relation. The projections of state vector |ψ〉 along basis vectors |x〉 and |y〉 are given by

(1.37)

By substituting (1.37) into (1.36) we obtain:

(1.38)

and from the right side of Eq. (1.38) we derive the completeness relation:

(1.39)

The probability that the photon in state |ψ〉 will pass the x-polaroid is given by

(1.40)

The probability amplitude of the photon in state |ψ〉 to pass the x-polaroid is

(1.41)

Let |ϕ〉 and |ψ〉 be two physical states. The probability amplitude of finding ϕ in ψ, denoted as a(ϕ → ψ), is given by

(1.42)

and the probability of ϕ passing the ψ test is given by

(1.43)

1.2 The Concept of the Qubit

Based on the previous section, it can be concluded that the quantum bit, also known as the qubit, lies in a two-dimensional Hilbert space H, isomorphic to C² space, where C is the complex number space, and can be represented as

(1.44)

where the |0〉 and |1〉 states are computational basis (CB) states, and |ψ〉 is a superposition state. If we perform the measurement of a qubit, we will get |0〉 with probability |α|² and |1〉 with probability of |β|². Measurement changes the state of a qubit from a superposition of |0〉 and |1〉 to the specific state consistent with the measurement result. If we parametrize the probability amplitudes α and β as follows:

(1.45)

where θ is a polar angle and ϕ is an azimuthal angle, we can geometrically represent the qubit by a Bloch sphere (or a Poincaré sphere for the photon), as illustrated in Figure 1.7. (Note that the Bloch sphere in Figure 1.7 is a little different from the Poincaré sphere in Figure 1.2.) Bloch vector coordinates are given by (cos ϕ sin θ, sin ϕ sin θ, cos θ). This Bloch vector representation is related to the CB by

(1.46)

where 0 ≤ θ π and 0 ≤ ϕ < 2π. The north and south poles correspond to computational |0〉 (|x〉-polarization) and |1〉 (|y〉-polarization) basis kets respectively. Other important bases are the diagonal basis related to the CB by

(1.47)

and the circular basis {|R〉,|L〉}, related to the CB as follows:

(1.48)

FIGURE 1.7 Block (Poincaré) sphere representation of a single qubit.

1.3 Spin-1/2 Systems

In addition to the photon, an important realization of the qubit is a spin-1/2 system. Nuclear magnetic resonance (NMR) is based on the fact that the proton possesses a magnetic moment μ , takes only two values, and this property characterizes a spin-1/2 particle. Experimentally this was confirmed by the Stern–Gerlach experiment, shown in , where the corresponding basis kets represent the spin-up and spin-down states. The superposition state can be represented in terms of these bases as follows:

(1.49)

where |ψ+|² (|ψ−|²) denotes the probability of finding the system in the spin-up (spin-down) state. By using the trigonometric identity sin²(θ/2) + cos²(θ/2) = 1, the expansion coefficients ψ+ and ψ− can be expressed as follows:

(1.50)

so that

(1.51)

Therefore, the single superposition state of a spin-1/2 system can also be visualized as the point (θ,ϕ) on a unit sphere (Bloch sphere).

FIGURE 1.8 The Stern–Gerlach experiment.

1.4 Quantum Gates and Quantum Information Processing

In quantum mechanics, the primitive undefined concepts are physical system, observable, and state. The concept of state has been introduced in the previous sections. An observable, such as momentum and spin, can be represented by an operator, such as A, in the vector space in question. An operator, or gate, acts on a ket from the left, (A) · |α〉 = A|α〉, and results in another ket. A linear operator (gate) B can be expressed in terms of eigenkets {|a(n)〉} of a Hermitian operator A. (An operator A is said to be Hermitian .) The operator X is associated with a square matrix (albeit infinite in extent), whose elements are

(1.52)

and can explicitly be written as

(1.53)

to denote that operator X is represented by the matrix above.

Very important single-qubit gates are: the Hadamard gate H, the phase shift gate S, the π/8 (or T) gate, controlled-NOT (or CNOT) gate, and Pauli operators X, Y, Z. The Hadamard gate H, phase-shift gate, T gate, and CNOT gate have the following matrix representation in the computational basis (CB) {|0〉,|1〉}:

(1.54)

The Pauli operators, on the other hand, have the following matrix representation in the CB:

(1.55)

is given as follows:

(1.56)

So the action of an X gate is to introduce the bit flip, the action of a Z gate is to introduce the phase flip, and the action of a Y gate is to simultaneously introduce the bit and phase flips.

Several important single-, double-, and three-qubit gates are shown in Figure 1.9. The action of a single-qubit gate is to apply the operator U on qubit |ψ〉, which results in another qubit. A controlled-U gate conditionally applies the operator U on target qubit |ψ〉, when the control qubit |c〉 is in the |1〉 state. One particularly important controlled U-gate is the controlled-NOT (CNOT) gate. This gate flips the content of target qubit |t〉 when the control qubit |c〉 is in the |1〉 state. The purpose of the SWAP gate is to interchange the positions of two qubits, and can be implemented by using three CNOT gates, as shown in Figure 1.9d. Finally, the Toffoli gate represents the generalization of the CNOT gate, where two control qubits are used. The minimum set of gates that can be used to perform an arbitrary quantum computation algorithm is known as the universal set of gates. The most popular sets of universal quantum gates are {H, S, CNOT, Toffoli} gates, {H, S, π/8 (T), CNOT} gates, the Barenco gate [13], and the Deutsch gate [14]. By using these universal quantum gates, more complicated operations can be performed. As an illustration, in Figure 1.10 the Bell states (Einstein–Podolsky–Rosen (EPR) pairs) preparation circuit is shown; this is very important in quantum teleportation and QKD applications.

FIGURE 1.9 Important quantum gates and their actions: (a) single-qubit gate; (b) controlled- U gate; (c) CNOT gate; (d) SWAP gate; (e) Toffoli gate.

FIGURE 1.10 Bell states (EPR pairs) preparation circuit.

So far, single-, double-, and triple-qubit quantum gates have been considered. An arbitrary quantum state of K qubits has the form Σs αs |s〉, where s runs over all binary strings of length K. Therefore, there are 2K complex coefficients, all independent except for the normalization constraint:

(1.57)

For example, the state α00|00〉 + α01|01〉 + α10|10〉 + α11|11〉 (with |α00|² + |α01|² + |α10|² + |α11|² = 1) is the general two-qubit state (we use |00〉 to denote the tensor product |0〉 ⊗ |0〉). The multiple qubits can be entangled so that they cannot be decomposed into two separate states. For example, the Bell state or EPR pair (|00〉 + |11〉)/√2 cannot be written in terms of tensor product |ψ1〉|ψ2〉 = (α1|0〉 + β1|1〉) ⊗ (α2|0〉 + β2|1〉) = α1 α2|00〉 + α1 β2|01〉 + β1 α2|10〉 + β1 β2|11〉, because in order to do so it must be α1 α2 = β1 β2 = 1/√2, while α1 β2 = β1 α2 = 0, which a priori has no reason to be valid. This state can be obtained by using the circuit shown in Figure 1.10, for the two-qubit input state |00〉. For more details on quantum gates and algorithms, the interested reader is referred to Chapters 3 and 5 respectively.

1.5 Quantum Teleportation

Quantum teleportation [17] is a technique to transfer quantum information from source to destination by employing entangled states. That is, in quantum teleportation, the entanglement in a Bell state (EPR pair) is used to transport an arbitrary quantum state |ψ〉 between two distant observers A and B (often called Alice and Bob), as illustrated in Figure 1.11. The quantum teleportation system employs three qubits: qubit 1 is an arbitrary state to be teleported, while qubits 2 and 3 are in a Bell state |B00〉 = (|00〉 + |11〉)/√2. Let the state to be teleported be denoted by |ψ〉 = a|0〉 + b|1〉. The input to the circuit shown in Figure 1.11 is therefore |ψ〉|B00〉, and can be rewritten as

(1.58)

The CNOT gate is then applied with the first qubit serving as control and the second qubit as target, which transforms (1.58) into

(1.59)

In the next stage, the Hadamard gate is applied to the first qubit, which maps |0〉 to (|0〉 + |1〉)/√2 and |1〉 to (|0〉 − |1〉)/√2, so that the overall transformation of (1.59) is as follows:

(1.60)

The measurements are performed on qubits 1 and 2, and based on the results of measurements, denoted respectively as a and b, the controlled-X (CNOT) and controlled-Z gates are applied conditionally to lead to the following content on qubit 3:

(1.61)

indicating that the arbitrary state |ψ〉 is teleported to the remote destination and can be found at the qubit 3 position.

FIGURE 1.11 The quantum teleportation principle.

1.6 Quantum Error Correction Concepts

The QIP relies on delicate superposition states, which are sensitive to interactions with the environment, resulting in decoherence. Moreover, the quantum gates are imperfect and the use of quantum error correction coding (QECC) is necessary to enable fault-tolerant computing and to deal with quantum errors [18–23]. QECC is also essential in quantum communication and quantum teleportation applications. The elements of quantum error correction codes are shown in Figure 1.12a. The (N,K) QECC code performs encoding of the quantum state of K qubits, specified by 2K complex coefficients αs, into a quantum state of N qubits, in such a way that errors can be detected and corrected, and all 2K complex coefficients can be perfectly restored, up to the global phase shift. This means that, from quantum mechanics (see Chapter 2), we know that two states |ψ〉 and ej θ|ψ〉 are equal up to a global phase shift as the results of measurement on both states are the same. A quantum error correction consists of four major steps: encoding, error detection, error recovery, and decoding. The sender (Alice) encodes quantum information in state |ψ〉 with the help of local ancillary qubits |0〉, and then sends the encoded qubits over a noisy quantum channel (say, a free-space optical channel or optical fiber). The receiver (Bob) performs multi-qubit measurement on all qubits to diagnose the channel error and performs a recovery unitary operation R to reverse the action of the channel. Quantum error correction is essentially more complicated than classical error correction. The difficulties in quantum error correction can be summarized as follows: (i) the no-cloning theorem indicates that it is impossible to make a copy of an arbitrary quantum state; (ii) quantum errors are continuous and a qubit can be in any superposition of the two bases states; and (iii) the measurements destroy the quantum information. The quantum error correction principles will be more evident after a simple example given below.

FIGURE 1.12 (a) Quantum error-correction principle. (b) The bit-flipping quantum channel model. (c) Three-qubit flip-error correction code encoder.

Assume we want to send a single qubit |ψ〉 = α|0〉 + β|1〉 through the quantum channel in which during transmission the transmitted qubit can be flipped toX|ψ〉 = β|0〉 + α|1〉 with probability p. Such a quantum channel is called a bit-flip channel and it can be described as shown in Figure 1.12b. The three-qubit flip code sends the same qubit three times, and therefore represents the repetition code (if the control qubit is |1〉 the target qubit gets flipped, otherwise it stays unchanged). The output of the first CNOT gate is used as input to the second CNOT gate in which the second ancillary qubit (the third qubit) is controlled by the information qubit (the first qubit), so that the corresponding encoder output is obtained as

. With this code, we are able to correct a single qubit-flip error, which occurs with probability (1 − p)³ + 3p(1 − p)² = 1 − 3p² + 2p³. Therefore, the probability of an error remaining uncorrected or wrongly corrected with this code is 3p² − 2p³. It is clear from Figure 1.12c that the three-qubit flip-code encoder is a systematic encoder in which the information qubit is unchanged, and the ancillary qubits are used to impose the encoding operation and create the parity qubits (the output qubits 2 and 3).

Let us assume that a qubit flip occurred on the first qubit, leading to received quantum word |ψr〉 = α|100〉 + β|011〉. In order to identify the error it is necessary to perform the measurements on the observables Z1 Z2 and Z2 Z3, where the subscript denotes the index of qubit on which a given Pauli gate is applied. The result of the measurement is the eigenvalue ±1, and corresponding eigenvectors are two valid codewords, namely |000〉 and |111〉. The observables can be represented as follows:

(1.62)

, indicating that an error occurred on either the first or second qubit, but not on the second or third qubit. The intersection reveals that the first qubit was in error. By using this approach we can create the three-qubit look-up table (LUT), given in Table 1.1.

Table 1.1 The three-qubit flip-code LUT

Three-qubit flip-code error detection and error correction circuits are shown in Figure 1.13. The results of measurements on ancillaries (see Figure 1.13a) will determine the error syndrome [±1 ±1], and based on the LUT given in Table 1.1, we identify the error event and apply the corresponding Xi gate on the ith qubit being in error, and the error is corrected since X² = I. The control logic operation is described in Table 1.1. For example, if both outputs of the measurements circuits are −1, the operator X2 is activated. The last step is to perform decoding as shown in Figure 1.13b by simply reversing the order of elements in the corresponding encoder.

FIGURE 1.13 (a) Three-qubit flip-code error detection and error correction circuit. (b) Decoder circuit configuration.

1.7 Quantum Key Distribution (QKD)

The QKD exploits the principle of quantum mechanics in order to enable to demonstrably secure distribution of a private key between two remote destinations. Private key cryptography is much older than public key cryptosystems, commonly used today. In a private key cryptosystem Alice (sender) must have an encoding key, while Bob (receiver) must have a matching decoding key to decrypt the encoded message. The simplest private key cryptosystem is the Vernam cipher (one time pad), which operates as follows [1]: (i) Alice and Bob share n-bit key strings; (ii) Alice encodes her n-bit message by adding the message and the key together; and (iii) Bob decodes the information by subtracting the key from the received message. There are several drawbacks in this scheme: (i) secure distribution of the key as well as the length of the key must be at least as long as the message length; (ii) the key bits cannot be reused; and (iii) the keys must be delivered in advance, securely stored until use, and be destroyed after their