Lecture Notes: Quantum Gates, Measurement, and Entanglement

Introduction

This document provides structured notes based on a lecture covering fundamental concepts in quantum mechanics relevant to quantum computation and information. We begin by discussing the physical realization of quantum gates using controllable interactions, focusing on spin systems. Subsequently, we delve into the formalism of quantum measurement, including general measurements, projective measurements, and POVMs, highlighting the probabilistic nature and state collapse inherent in the process. We explore the consequences of measurement, such as the Heisenberg Uncertainty Principle and the distinguishability of quantum states. The concept of phase factors and their significance is clarified. We then move to composite systems, introducing the tensor product structure and the crucial phenomenon of entanglement, illustrated with Bell states. Finally, we examine an application of entanglement in the superdense coding protocol, demonstrating how quantum resources can enhance communication capacity.

Physical Realization of Quantum Gates

Quantum gates, representing unitary operations on qubits, are the fundamental components of quantum algorithms. Their physical implementation requires precise control over quantum systems. This is achieved by engineering specific interactions, often using external fields, to guide the time evolution of the qubit state according to a desired unitary transformation.

Interaction Hamiltonians: Spin Qubit Example

A common physical realization of a qubit uses the spin state of a spin-\(\frac{1}{2}\) particle, such as an electron or an atomic nucleus. Such particles possess an intrinsic angular momentum \(\mathbf{S}\) and an associated magnetic dipole moment \(\mathbf{\mu}\). \[\mathbf{\mu} = \gamma \mathbf{S}\] where \(\gamma\) is the gyromagnetic ratio, a constant specific to the particle. For a spin-\(\frac{1}{2}\) system, the spin angular momentum operator is related to the vector of Pauli matrices \(\bm{\sigma}= (\hat{\sigma}_x, \hat{\sigma}_y, \hat{\sigma}_z)\) by \(\mathbf{S} = \frac{\hbar}{2} \bm{\sigma}\). Thus, the magnetic moment operator is: \[\mathbf{\mu} = \gamma \frac{\hbar}{2} \bm{\sigma}\]

When this particle is placed in an external magnetic field \(\mathbf{B}\), its energy is described by the interaction Hamiltonian: \[\hat{H} = - \mathbf{\mu} \cdot \mathbf{B} = - \gamma \frac{\hbar}{2} \bm{\sigma}\cdot \mathbf{B} \label{eq:spin_hamiltonian}\] This Hamiltonian dictates how the spin interacts with the field. Analogous to a classical compass needle aligning with the Earth’s magnetic field, the quantum spin experiences a torque that tends to align it with \(\mathbf{B}\) to minimize the interaction energy. By controlling the direction and strength of \(\mathbf{B}\), we can control the Hamiltonian governing the qubit’s evolution.

Unitary Evolution and Gate Implementation via Magnetic Fields

The evolution of a closed quantum system governed by a time-independent Hamiltonian \(\hat{H}\) is described by a unitary operator \(\hat{U}(t)\): \[\ket{\psi(t)} = \hat{U}(t) \ket{\psi(0)} \quad \text{where} \quad \hat{U}(t) = e^{-i \hat{H} t / \hbar}\] This operator represents the quantum gate implemented by the interaction over time \(t\).

Consider applying a static magnetic field along the x-axis: \(\mathbf{B} = B_x \hat{\mathbf{x}}\). The Hamiltonian ([eq:spin_hamiltonian]) simplifies to: \[\hat{H}_x = - \gamma \frac{\hbar}{2} B_x \hat{\sigma}_x\] Let us define an angular frequency (related to the Larmor frequency) \(\omega = \frac{\gamma B_x}{2}\). The Hamiltonian can be written as \(\hat{H}_x = - \hbar \omega \hat{\sigma}_x\). The corresponding evolution operator is: \[\hat{U}_x(t) = e^{-i \hat{H}_x t / \hbar} = e^{-i (-\hbar \omega \hat{\sigma}_x) t / \hbar} = e^{i \omega t \hat{\sigma}_x}\]

The sign convention in the Hamiltonian \(\hat{H} = -\mathbf{\mu} \cdot \mathbf{B}\) is standard. The resulting evolution operator \(e^{+i \omega t \hat{\sigma}_x}\) corresponds to a rotation around the \(+x\) axis. Some sources might use different conventions.

Using the property \((\hat{\sigma}_x)^2 = \mathds{1}\) and the Taylor expansion of the exponential (or Euler’s formula for Pauli matrices \(e^{i \theta \hat{\sigma}_k} = \cos(\theta) \mathds{1}+ i \sin(\theta) \hat{\sigma}_k\)), we get: \[\hat{U}_x(t) = \cos(\omega t) \mathds{1}+ i \sin(\omega t) \hat{\sigma}_x \label{eq:rotation_x}\] This operator represents a rotation of the qubit state vector around the x-axis of the Bloch sphere by an angle \(\theta = 2\omega t = \gamma B_x t\). By precisely controlling the duration \(t\) for which the magnetic field \(B_x\) is applied, we can implement any desired rotation around the x-axis. Similarly, fields along y or z axes implement rotations \(\hat{U}_y(t) = e^{i \omega_y t \hat{\sigma}_y}\) and \(\hat{U}_z(t) = e^{i \omega_z t \hat{\sigma}_z}\).

To implement a Pauli-X gate (up to a global phase), we need a rotation by \(\pi\) around the x-axis. This requires the angle \(2\omega t = \pi\), or \(\omega t = \pi/2\). We apply the field \(B_x\) for a time \(t = \frac{\pi}{2\omega} = \frac{\pi}{\gamma B_x}\). The evolution operator becomes: \[\hat{U}_x(t_{\pi/2}) = \cos(\pi/2) \mathds{1}+ i \sin(\pi/2) \hat{\sigma}_x= 0 \cdot \mathds{1}+ i \cdot 1 \cdot \hat{\sigma}_x= i \hat{\sigma}_x\] Applying this to an arbitrary qubit state \(\ket{\psi} = \alpha \ket{0} + \beta \ket{1}\): \[\hat{U}_x(t_{\pi/2}) \ket{\psi} = i \hat{\sigma}_x(\alpha \ket{0} + \beta \ket{1}) = i (\alpha \hat{\sigma}_x\ket{0} + \beta \hat{\sigma}_x\ket{1}) = i (\alpha \ket{1} + \beta \ket{0})\] This operation swaps the amplitudes of the \(\ket{0}\) and \(\ket{1}\) states, which is the action of the Pauli-X (NOT) gate. The factor \(i\) is a global phase factor, which does not affect measurement outcomes and is often disregarded when considering the logical action of the gate.

Remark. By combining rotations around different axes (e.g., X, Y, Z), any arbitrary single-qubit unitary operation can be constructed. For universal quantum computation, we also need multi-qubit gates, such as the Controlled-NOT (CNOT) gate. Implementing these requires controllable interactions *between* qubits. In physical systems like trapped ions or superconducting circuits, this might involve mediating the interaction via a shared bus (like a phonon mode or microwave cavity) or direct coupling (e.g., dipole-dipole interaction). In Nuclear Magnetic Resonance (NMR) quantum computing, complex pulse sequences of radio-frequency magnetic fields are used to manipulate spins and engineer effective interactions, allowing for sophisticated gate implementations. The precise control achievable in NMR has made it a valuable testbed for quantum control techniques.

Quantum Measurement

Measurement occupies a central and somewhat enigmatic role in quantum mechanics. Unlike classical measurements, which passively reveal pre-existing properties, quantum measurements are inherently probabilistic and actively influence the state of the system being measured, a phenomenon known as state collapse or wave function collapse.

General Formalism of Measurement (Measurement Operators)

The most general description of a quantum measurement process is given by a set of measurement operators \(\{\hat{M}_m\}\), acting on the Hilbert space \(\mathcal{H}\) of the system. The index \(m\) labels the possible distinct outcomes of the measurement.

Measurement Operators and Probabilities

If the state of the quantum system immediately before the measurement is described by the normalized state vector \(\ket{\psi} \in \mathcal{H}\), the probability \(p(m)\) of obtaining outcome \(m\) is given by the Born rule: \[p(m) = \bra{\psi} \hat{M}_m^\dagger \hat{M}_m \ket{\psi} \label{eq:prob_measurement}\] where \(\hat{M}_m^\dagger\) is the adjoint (Hermitian conjugate) of \(\hat{M}_m\). The operator \(\hat{E}_m = \hat{M}_m^\dagger \hat{M}_m\) is positive semi-definite (meaning \(\bra{\phi} \hat{E}_m \ket{\phi} \ge 0\) for all \(\ket{\phi}\)). This ensures that the probability \(p(m)\) is always non-negative. We can also write the probability as the squared norm of the state vector acted upon by the measurement operator: \[p(m) = \left\| \hat{M}_m \ket{\psi} \right\|^2 \ge 0\]

Post-Measurement State Collapse and Normalization

If the measurement yields outcome \(m\), the state of the system immediately after the measurement is no longer \(\ket{\psi}\) but collapses to a new state \(\ket{\psi_m'}\) given by: \[\ket{\psi_m'} = \frac{\hat{M}_m \ket{\psi}}{\sqrt{p(m)}} = \frac{\hat{M}_m \ket{\psi}}{\left\| \hat{M}_m \ket{\psi} \right\|} \label{eq:post_measurement_state}\] The division by \(\sqrt{p(m)}\) (the norm of \(\hat{M}_m \ket{\psi}\)) ensures that the post-measurement state \(\ket{\psi_m'}\) is properly normalized (\(\left\|\ket{\psi_m'}\right\|=1\)). This abrupt change from \(\ket{\psi}\) to \(\ket{\psi_m'}\) is the "collapse of the wave function". It is a non-unitary, probabilistic process postulated in quantum mechanics, representing the update of the system’s description based on the information gained from the measurement.

Completeness Relation

The set of all possible outcomes must cover all possibilities, meaning the probabilities must sum to unity for any initial state \(\ket{\psi}\): \[\sum_m p(m) = \sum_m \bra{\psi} \hat{M}_m^\dagger \hat{M}_m \ket{\psi} = \bra{\psi} \left( \sum_m \hat{M}_m^\dagger \hat{M}_m \right) \ket{\psi} = 1\] This condition must hold for all possible normalized states \(\ket{\psi}\). This implies that the measurement operators must satisfy the completeness relation: \[\sum_m \hat{M}_m^\dagger \hat{M}_m = \mathds{1} \label{eq:completeness}\] where \(\mathds{1}\) is the identity operator on the Hilbert space \(\mathcal{H}\). This ensures the conservation of probability.

Destructive vs. Non-destructive Measurements

Measurements can be broadly classified based on their effect on the measured system’s existence:

• Destructive Measurement: The measurement process inherently destroys or absorbs the system being measured, preventing its further use or evolution. A common example is photon detection using a photomultiplier tube or a photographic plate, where the photon’s energy is absorbed to trigger a signal, annihilating the photon itself.

• Non-destructive Measurement (QND): The system survives the measurement interaction, allowing for subsequent operations or measurements. The state collapses according to [eq:post_measurement_state], but the system itself persists. An example is measuring the spin component of an atom using the Stern-Gerlach apparatus, where the atom is deflected based on its spin but is not absorbed. Quantum Non-Demolition (QND) measurements are a specific type designed to measure an observable without disturbing its value for subsequent measurements.

The general formalism involving post-measurement states ([eq:post_measurement_state]) primarily describes non-destructive measurements. In many quantum computing contexts, particularly measurements at the end of a circuit, the measurement might be effectively destructive as the qubits are not reused.

Projective Measurements

A particularly important and common type of quantum measurement is the projective measurement, which is directly associated with physical observables represented by Hermitian operators.

Projectors as Measurement Operators

Projective measurements are described by a set of measurement operators \(\{\hat{P}_m\}\) that are orthogonal projectors. A projector \(\hat{P}_m\) projects vectors onto a specific subspace \(\mathcal{H}_m\) of the total Hilbert space \(\mathcal{H}\). These projectors satisfy the following properties:

1. Hermiticity: \(\hat{P}_m^\dagger = \hat{P}_m\) (Projectors are Hermitian operators).

2. Idempotency: \(\hat{P}_m^2 = \hat{P}_m\) (Projecting twice is the same as projecting once).

3. Orthogonality: The subspaces corresponding to distinct outcomes \(m \ne n\) are orthogonal, meaning \(\mathcal{H}_m \perp \mathcal{H}_n\). This implies \(\hat{P}_m \hat{P}_n = \hat{0}\) (the zero operator) for \(m \ne n\). This can be combined with idempotency as \(\hat{P}_m \hat{P}_n = \delta_{mn} \hat{P}_m\).

For projective measurements, the completeness relation ([eq:completeness]) takes the form: \[\sum_m \hat{P}_m^\dagger \hat{P}_m = \sum_m \hat{P}_m^2 = \sum_m \hat{P}_m = \mathds{1} \label{eq:projector_completeness}\] This signifies that the orthogonal subspaces \(\{\mathcal{H}_m\}\) onto which the projectors project span the entire Hilbert space \(\mathcal{H}\) (\(\bigoplus_m \mathcal{H}_m = \mathcal{H}\)).

Example: Measuring a Qubit in the Computational Basis

Consider the standard measurement of a qubit in the computational basis \(\{\ket{0}, \ket{1}\}\). This corresponds to measuring the observable represented by the Pauli-Z operator \(\hat{\sigma}_z= \ket{0}\bra{0} - \ket{1}\bra{1}\). The possible outcomes are the eigenvalues +1 (associated with state \(\ket{0}\)) and -1 (associated with state \(\ket{1}\)). The measurement operators are the projectors onto the corresponding eigenstates: \[\begin{aligned} \hat{P}_0 &= \ket{0}\bra{0} \quad (\text{Projector onto the subspace spanned by } \ket{0}) \\ \hat{P}_1 &= \ket{1}\bra{1} \quad (\text{Projector onto the subspace spanned by } \ket{1})\end{aligned}\] These operators clearly satisfy Hermiticity, Idempotency, Orthogonality (\(\hat{P}_0 \hat{P}_1 = 0\)), and Completeness (\(\hat{P}_0 + \hat{P}_1 = \ket{0}\bra{0} + \ket{1}\bra{1} = \mathds{1}\)).

Let the qubit be in an arbitrary superposition state \(\ket{\psi} = \alpha \ket{0} + \beta \ket{1}\), where \(|\alpha|^2 + |\beta|^2 = 1\). The probability of obtaining outcome 0 (corresponding to projector \(\hat{P}_0\)) is: \[\begin{aligned} p(0) &= \bra{\psi} \hat{P}_0^\dagger \hat{P}_0 \ket{\psi} = \bra{\psi} \hat{P}_0 \ket{\psi} \\ &= (\alpha^* \bra{0} + \beta^* \bra{1}) (\ket{0}\bra{0}) (\alpha \ket{0} + \beta \ket{1}) \\ &= (\alpha^* \bra{0} + \beta^* \bra{1}) (\alpha \ket{0}\braket{0|0} + \beta \ket{0}\braket{0|1}) \\ &= (\alpha^* \bra{0} + \beta^* \bra{1}) (\alpha \ket{0}) \\ &= \alpha^* \alpha \braket{0|0} + \beta^* \alpha \braket{1|0} \\ &= |\alpha|^2 \cdot 1 + \beta^* \alpha \cdot 0 = |\alpha|^2\end{aligned}\] Similarly, the probability of obtaining outcome 1 (corresponding to projector \(\hat{P}_1\)) is: \[\begin{aligned} p(1) &= \bra{\psi} \hat{P}_1^\dagger \hat{P}_1 \ket{\psi} = \bra{\psi} \hat{P}_1 \ket{\psi} \\ &= (\alpha^* \bra{0} + \beta^* \bra{1}) (\ket{1}\bra{1}) (\alpha \ket{0} + \beta \ket{1}) \\ &= (\alpha^* \bra{0} + \beta^* \bra{1}) (\beta \ket{1}) \\ &= \alpha^* \beta \braket{0|1} + \beta^* \beta \braket{1|1} \\ &= \alpha^* \beta \cdot 0 + |\beta|^2 \cdot 1 = |\beta|^2\end{aligned}\] These are the standard probabilities associated with measuring a qubit in the computational basis. As required, \(p(0) + p(1) = |\alpha|^2 + |\beta|^2 = 1\).

State Collapse after Projective Measurement

Applying the general rule for the post-measurement state ([eq:post_measurement_state]) to projective measurements: If the measurement yields outcome \(m\) (corresponding to projector \(\hat{P}_m\)), the state collapses to: \[\ket{\psi_m'} = \frac{\hat{P}_m \ket{\psi}}{\sqrt{p(m)}} = \frac{\hat{P}_m \ket{\psi}}{\left\|\hat{P}_m \ket{\psi}\right\|}\] In the qubit example:

• If outcome 0 is obtained (\(p(0) = |\alpha|^2\)): \[\ket{\psi_0'} = \frac{\hat{P}_0 \ket{\psi}}{\sqrt{|\alpha|^2}} = \frac{\ket{0}\braket{0|\psi}}{|\alpha|} = \frac{\ket{0}(\alpha \braket{0|0} + \beta \braket{0|1})}{|\alpha|} = \frac{\alpha \ket{0}}{|\alpha|}\] The state collapses to \(\ket{0}\), up to a phase factor \(\alpha/|\alpha|\). Since global phases are unobservable, the effective state after measuring 0 is \(\ket{0}\).

• If outcome 1 is obtained (\(p(1) = |\beta|^2\)): \[\ket{\psi_1'} = \frac{\hat{P}_1 \ket{\psi}}{\sqrt{|\beta|^2}} = \frac{\ket{1}\braket{1|\psi}}{|\beta|} = \frac{\ket{1}(\alpha \braket{1|0} + \beta \braket{1|1})}{|\beta|} = \frac{\beta \ket{1}}{|\beta|}\] The state collapses effectively to \(\ket{1}\).

The measurement forces the system into one of the eigenstates corresponding to the measured outcome.

Remark. The choice of projectors \(\{\hat{P}_m\}\) defines the measurement basis. Measuring in the computational basis \(\{\ket{0}, \ket{1}\}\) uses \(\hat{P}_0 = \ket{0}\bra{0}, \hat{P}_1 = \ket{1}\bra{1}\). Measuring in a different basis, like the Hadamard basis \(\{\ket{+}, \ket{-}\}\) where \(\ket{\pm} = (\ket{0} \pm \ket{1})/\sqrt{2}\), would involve projectors \(\hat{P}_+ = \ket{+}\bra{+}\) and \(\hat{P}_- = \ket{-}\bra{-}\). The outcomes and probabilities would differ according to the projections onto these basis states. For \(\ket{\psi} = \alpha \ket{0} + \beta \ket{1}\), \(p(+) = |\braket{+|\psi}|^2 = |\frac{\alpha+\beta}{\sqrt{2}}|^2\) and \(p(-) = |\braket{-|\psi}|^2 = |\frac{\alpha-\beta}{\sqrt{2}}|^2\).

Remark. The concept of "collapse" remains one of the most profound and debated aspects of quantum theory. It signifies a departure from the smooth, deterministic unitary evolution described by the Schrödinger equation. The Copenhagen interpretation accepts collapse as a fundamental postulate, often drawing a boundary (the "Heisenberg cut") between the quantum system and the classical measurement apparatus. However, defining this boundary is problematic, as the apparatus itself should ultimately be describable by quantum mechanics. Alternative interpretations, such as the Many-Worlds Interpretation (4.3.5), attempt to explain the appearance of collapse without modifying the underlying unitary dynamics, typically at the cost of other counter-intuitive features (like branching universes). Despite these interpretational debates, the mathematical formalism of measurement operators and state collapse provides a consistent and experimentally verified framework for predicting measurement outcomes.

Distinguishing Quantum States

A crucial question in quantum information is whether we can reliably determine the state of a quantum system if it is known to be one of several possibilities. The answer depends critically on whether the possible states are orthogonal.

Case 1: Distinguishing Orthogonal States

Suppose Alice prepares a system in one state from a set of mutually orthogonal states \(\{\ket{\psi_1}, \ket{\psi_2}, \dots, \ket{\psi_n}\}\), satisfying \(\braket{\psi_i|\psi_j} = \delta_{ij}\). She sends the chosen state, say \(\ket{\psi_j}\), to Bob.

Bob can determine which state Alice sent with certainty by performing a projective measurement defined by the projectors \(\hat{P}_i = \ket{\psi_i}\bra{\psi_i}\). These projectors form a valid measurement set if the states span the whole space or if we add a projector onto the remaining orthogonal subspace. When Bob receives the state \(\ket{\psi_j}\) and performs this measurement, the probability of obtaining outcome \(i\) (corresponding to projector \(\hat{P}_i\)) is: \[p(i | \text{state is } \ket{\psi_j}) = \bra{\psi_j} \hat{P}_i^\dagger \hat{P}_i \ket{\psi_j} = \bra{\psi_j} \hat{P}_i \ket{\psi_j} = \bra{\psi_j} (\ket{\psi_i}\bra{\psi_i}) \ket{\psi_j} = |\braket{\psi_i|\psi_j}|^2 = |\delta_{ij}|^2 = \delta_{ij}\] This means:

• If \(i = j\), the probability is \(p(j | \ket{\psi_j}) = 1\).

• If \(i \ne j\), the probability is \(p(i | \ket{\psi_j}) = 0\).

Bob will measure outcome \(j\) with 100

If Alice agrees to send either \(\ket{0}\) or \(\ket{1}\) (which are orthogonal), Bob can measure in the computational basis (\(\{\hat{P}_0, \hat{P}_1\}\)). If he measures 0, he knows Alice sent \(\ket{0}\); if he measures 1, he knows Alice sent \(\ket{1}\). This is analogous to classical communication where distinct signals can be perfectly distinguished.

Case 2: Distinguishing Non-Orthogonal States

The situation changes dramatically if the set of possible states prepared by Alice, \(\{\ket{\phi_1}, \ket{\phi_2}, \dots, \ket{\phi_n}\}\), contains at least one pair of states that are not orthogonal. That is, there exists some \(i \ne j\) such that their inner product \(\braket{\phi_i|\phi_j} \ne 0\).

It is impossible to perfectly distinguish between two or more non-orthogonal quantum states using any single quantum measurement.

Proof. Sketch of Proof. Assume, for contradiction, that a measurement described by operators \(\{\hat{M}_m\}\) exists that can perfectly distinguish the non-orthogonal states \(\{\ket{\phi_i}\}\). Let the outcome \(m=i\) correspond to identifying the state as \(\ket{\phi_i}\). For perfect distinction, two conditions must hold: 1. If the state is \(\ket{\phi_i}\), the measurement must yield outcome \(i\) with probability 1: \(p(i | \ket{\phi_i}) = \bra{\phi_i} \hat{M}_i^\dagger \hat{M}_i \ket{\phi_i} = 1\). 2. If the state is \(\ket{\phi_i}\), the measurement must yield any other outcome \(j \ne i\) with probability 0: \(p(j | \ket{\phi_i}) = \bra{\phi_i} \hat{M}_j^\dagger \hat{M}_j \ket{\phi_i} = \left\|\hat{M}_j \ket{\phi_i}\right\|^2 = 0\). This implies \(\hat{M}_j \ket{\phi_i} = \ket{0}\) (the null vector) for all \(j \ne i\).

Now, consider two non-orthogonal states \(\ket{\phi_i}\) and \(\ket{\phi_k}\) (\(i \ne k\)), so their inner product \(\braket{\phi_k|\phi_i} \ne 0\). Using the completeness relation \(\sum_m \hat{M}_m^\dagger \hat{M}_m = \mathds{1}\), we can write: \[\begin{aligned} \braket{\phi_k|\phi_i} &= \bra{\phi_k} \mathds{1}\ket{\phi_i} = \bra{\phi_k} \left( \sum_m \hat{M}_m^\dagger \hat{M}_m \right) \ket{\phi_i} \\ &= \sum_m \bra{\phi_k} \hat{M}_m^\dagger \hat{M}_m \ket{\phi_i} \\ &= \sum_m (\hat{M}_m \ket{\phi_k})^\dagger (\hat{M}_m \ket{\phi_i}) \end{aligned}\] Consider the term \(\hat{M}_m \ket{\phi_i}\). From condition 2, this term is zero unless \(m=i\). Similarly, the term \(\hat{M}_m \ket{\phi_k}\) is zero unless \(m=k\). Since \(i \ne k\), for any given \(m\), at least one of the vectors \((\hat{M}_m \ket{\phi_k})\) or \((\hat{M}_m \ket{\phi_i})\) must be the null vector. Therefore, the product \((\hat{M}_m \ket{\phi_k})^\dagger (\hat{M}_m \ket{\phi_i})\) is zero for all \(m\). This implies that the sum is zero: \(\braket{\phi_k|\phi_i} = 0\). This contradicts our initial assumption that \(\ket{\phi_i}\) and \(\ket{\phi_k}\) are non-orthogonal (\(\braket{\phi_k|\phi_i} \ne 0\)). Thus, the initial assumption that a perfect distinguishing measurement exists must be false. ◻

The impossibility of perfectly distinguishing non-orthogonal states is deeply connected to the no-cloning theorem, which states that an unknown quantum state cannot be perfectly copied. If we could perfectly distinguish non-orthogonal states, we could, in principle, build a device that identifies the state and then prepares multiple copies of it, violating the no-cloning theorem.

This fundamental limitation has significant implications for quantum communication and cryptography. While non-orthogonal states cannot be distinguished perfectly, strategies exist (often using POVMs) to distinguish them probabilistically or unambiguously with a chance of failure (inconclusive outcome).

Positive Operator-Valued Measurements (POVMs)

POVMs generalize the concept of projective measurements and provide the most comprehensive framework for describing measurements allowed by quantum mechanics. They are particularly relevant when the post-measurement state is not of primary interest, or when dealing with tasks like optimally distinguishing non-orthogonal states.

Definition and Properties

A POVM is defined by a set of operators \(\{\hat{E}_m\}\), called POVM elements or effects, acting on the Hilbert space \(\mathcal{H}\) of the system. The index \(m\) labels the possible measurement outcomes. These operators must satisfy two conditions:

1. Positivity: Each \(\hat{E}_m\) must be a positive semi-definite operator. This means \(\bra{\phi} \hat{E}_m\ket{\phi} \ge 0\) for all states \(\ket{\phi} \in \mathcal{H}\). This ensures that the calculated probabilities are non-negative.

2. Completeness: The sum of all POVM elements must equal the identity operator on the Hilbert space: \[\sum_m \hat{E}_m = \mathds{1} \label{eq:povm_completeness}\] This ensures that the probabilities sum to one.

The probability \(p(m)\) of obtaining outcome \(m\) when the system is in state \(\ket{\psi}\) (or more generally, described by density operator \(\rho\)) is given by: \[p(m) = \bra{\psi} \hat{E}_m \ket{\psi} \quad (\text{for pure state } \ket{\psi})\] or, more generally: \[p(m) = \operatorname{Tr}(\hat{E}_m \rho) \quad (\text{for mixed state } \rho) \label{eq:povm_probability_general}\] Note that POVM elements \(\hat{E}_m\) are not required to be orthogonal projectors, nor do they need to commute with each other.

Relation to Measurement Operators (\(\hat{M}_m\))

The general measurement formalism using operators \(\{\hat{M}_m\}\) (2.1) is directly connected to the POVM formalism. Given a set of measurement operators \(\{\hat{M}_m\}\), we can define the corresponding POVM elements as: \[\hat{E}_m = \hat{M}_m^\dagger \hat{M}_m\] These operators \(\hat{E}_m\) are inherently positive semi-definite. Furthermore, the completeness relation for the measurement operators, \(\sum_m \hat{M}_m^\dagger \hat{M}_m = \mathds{1}\), directly translates into the completeness relation for the POVM elements, \(\sum_m \hat{E}_m = \mathds{1}\). The probability rule \(p(m) = \bra{\psi} \hat{M}_m^\dagger \hat{M}_m \ket{\psi}\) becomes identical to \(p(m) = \bra{\psi} \hat{E}_m \ket{\psi}\).

Conversely, given a valid set of POVM elements \(\{\hat{E}_m\}\), it is always possible to find a set of measurement operators \(\{\hat{M}_m\}\) such that \(\hat{E}_m = \hat{M}_m^\dagger \hat{M}_m\). However, this decomposition is not unique. A common choice is \(\hat{M}_m = \sqrt{\hat{E}_m}\), where \(\sqrt{\hat{E}_m}\) denotes the unique positive square root of the positive operator \(\hat{E}_m\). More generally, Neumark’s extension theorem shows that any POVM can be realized by introducing an auxiliary system (ancilla), performing a joint unitary operation on the system and ancilla, and then performing a standard projective measurement on the ancilla (or a combined system).

Remark. Projective measurements are a special case of POVMs. If the measurement operators are orthogonal projectors \(\{\hat{P}_m\}\), then the corresponding POVM elements are \(\hat{E}_m = \hat{P}_m^\dagger \hat{P}_m = \hat{P}_m^2 = \hat{P}_m\). In this case, the POVM elements are themselves the projectors.

Application: Strategies for Distinguishing Non-Orthogonal States

Since perfect distinction of non-orthogonal states is impossible (2.3.2), POVMs provide the mathematical tools to design optimal strategies for partial distinction. Two common strategies are:

1. Minimum Error Discrimination: Design a POVM that minimizes the average probability of making an incorrect identification.

2. Unambiguous State Discrimination (USD): Design a POVM where some outcomes identify the state with certainty, while other outcomes are explicitly "inconclusive". This strategy avoids errors but does not always succeed in identifying the state.

Let Alice send either \(\ket{\psi_1} = \ket{0}\) or \(\ket{\psi_2} = \ket{+} = \frac{1}{\sqrt{2}}(\ket{0} + \ket{1})\). Bob wants to identify the state unambiguously. Since \(\braket{\psi_1|\psi_2} = 1/\sqrt{2} \ne 0\), perfect discrimination is impossible. Bob can use a POVM with three outcomes:

• Outcome 1: "The state is definitely \(\ket{\psi_2}\)".

• Outcome 2: "The state is definitely \(\ket{\psi_1}\)".

• Outcome 3: "Inconclusive".

To achieve this, the POVM element \(\hat{E}_1\) must yield zero probability if the state was \(\ket{\psi_1}\), and \(\hat{E}_2\) must yield zero probability if the state was \(\ket{\psi_2}\). This suggests choosing \(\hat{E}_1\) proportional to the projector onto the subspace orthogonal to \(\ket{\psi_1}\), and \(\hat{E}_2\) proportional to the projector onto the subspace orthogonal to \(\ket{\psi_2}\).

• Orthogonal to \(\ket{\psi_1} = \ket{0}\) is \(\ket{1}\). Projector: \(\hat{\Pi}_{\perp \psi_1} = \ket{1}\bra{1}\).

• Orthogonal to \(\ket{\psi_2} = \ket{+}\) is \(\ket{-} = \frac{1}{\sqrt{2}}(\ket{0} - \ket{1})\). Projector: \(\hat{\Pi}_{\perp \psi_2} = \ket{-}\bra{-}\).

A valid POVM for USD can be constructed as (related to the Ivanovic-Dieks-Peres limit): Let \(c = \frac{1}{1+|\braket{\psi_1|\psi_2}|} = \frac{1}{1+1/\sqrt{2}}\). \[\begin{aligned} \hat{E}_1 &= c \hat{\Pi}_{\perp \psi_1} = c \ket{1}\bra{1} \label{eq:usd_e1_corrected} \\ \hat{E}_2 &= c \hat{\Pi}_{\perp \psi_2} = c \ket{-}\bra{-} \label{eq:usd_e2_corrected} \\ \hat{E}_3 &= \mathds{1}- \hat{E}_1 - \hat{E}_2 \label{eq:usd_e3_corrected} \end{aligned}\] We need to verify that \(\hat{E}_3\) is positive semi-definite. \(\hat{E}_1 = c \begin{pmatrix} 0 & 0 \\ 0 & 1 \end{pmatrix}\) \(\hat{E}_2 = c \ket{-}\bra{-} = \frac{c}{2}(\ket{0} - \ket{1})(\bra{0} - \bra{1}) = \frac{c}{2}(\ket{0}\bra{0} - \ket{0}\bra{1} - \ket{1}\bra{0} + \ket{1}\bra{1}) = \frac{c}{2}\begin{pmatrix} 1 & -1 \\ -1 & 1 \end{pmatrix}\) \(\hat{E}_1 + \hat{E}_2 = \begin{pmatrix} c/2 & -c/2 \\ -c/2 & c/2 + c \end{pmatrix} = \begin{pmatrix} c/2 & -c/2 \\ -c/2 & 3c/2 \end{pmatrix}\) \(\hat{E}_3 = \mathds{1}- (\hat{E}_1 + \hat{E}_2) = \begin{pmatrix} 1 - c/2 & c/2 \\ c/2 & 1 - 3c/2 \end{pmatrix}\). With \(c = 1/(1+1/\sqrt{2}) = \sqrt{2}/(\sqrt{2}+1) = \sqrt{2}(\sqrt{2}-1) = 2-\sqrt{2} \approx 0.586\). \(1 - c/2 = 1 - (1-\frac{\sqrt{2}}{2}) = \frac{\sqrt{2}}{2} \approx 0.707\). \(1 - 3c/2 = 1 - 3(1-\frac{\sqrt{2}}{2}) = 1 - 3 + \frac{3\sqrt{2}}{2} = \frac{3\sqrt{2}}{2} - 2 \approx 0.121\). \(\hat{E}_3 = \begin{pmatrix} \sqrt{2}/2 & (2-\sqrt{2})/2 \\ (2-\sqrt{2})/2 & (3\sqrt{2}/2)-2 \end{pmatrix}\). The determinant is \((\frac{\sqrt{2}}{2})(\frac{3\sqrt{2}}{2}-2) - (\frac{2-\sqrt{2}}{2})^2 = (\frac{3}{2} - \sqrt{2}) - \frac{4 - 4\sqrt{2} + 2}{4} = (\frac{3}{2} - \sqrt{2}) - \frac{6 - 4\sqrt{2}}{4} = (\frac{3}{2} - \sqrt{2}) - (\frac{3}{2} - \sqrt{2}) = 0\). The trace is \(\frac{\sqrt{2}}{2} + \frac{3\sqrt{2}}{2} - 2 = 2\sqrt{2} - 2 > 0\). Since the determinant is 0 and the trace is positive, the eigenvalues are non-negative (one is 0, one is positive), so \(\hat{E}_3\) is positive semi-definite.

Now, let’s check the probabilities:

• If Alice sends \(\ket{\psi_1} = \ket{0}\): - \(p(1) = \bra{0} \hat{E}_1 \ket{0} = c \bra{0} (\ket{1}\bra{1}) \ket{0} = c |\braket{1|0}|^2 = 0\). (Correct: outcome 1 excludes \(\ket{0}\)) - \(p(2) = \bra{0} \hat{E}_2 \ket{0} = c \bra{0} (\ket{-}\bra{-}) \ket{0} = c |\braket{-|0}|^2 = c |1/\sqrt{2}|^2 = c/2\). (Non-zero probability of identifying \(\ket{0}\)) - \(p(3) = \bra{0} \hat{E}_3 \ket{0} = 1 - p(1) - p(2) = 1 - c/2\). (Probability of inconclusive result)

• If Alice sends \(\ket{\psi_2} = \ket{+}\): - \(p(1) = \bra{+} \hat{E}_1 \ket{+} = c \bra{+} (\ket{1}\bra{1}) \ket{+} = c |\braket{1|+}|^2 = c |1/\sqrt{2}|^2 = c/2\). (Non-zero probability of identifying \(\ket{+}\)) - \(p(2) = \bra{+} \hat{E}_2 \ket{+} = c \bra{+} (\ket{-}\bra{-}) \ket{+} = c |\braket{-|+}|^2 = 0\). (Correct: outcome 2 excludes \(\ket{+}\)) - \(p(3) = \bra{+} \hat{E}_3 \ket{+} = 1 - p(1) - p(2) = 1 - c/2\). (Probability of inconclusive result)

If Bob gets outcome 1, he knows Alice sent \(\ket{+}\). If he gets outcome 2, he knows Alice sent \(\ket{0}\). If he gets outcome 3, he learns nothing. This POVM allows unambiguous identification when it succeeds, but success is not guaranteed. The maximum probability of success for USD is \(1 - |\braket{\psi_1|\psi_2}| = 1 - 1/\sqrt{2}\). The probability of getting an inconclusive result (outcome 3) is \(p(3) = 1 - c/2 = 1 - (2-\sqrt{2})/2 = 1 - 1 + \sqrt{2}/2 = 1/\sqrt{2} = |\braket{\psi_1|\psi_2}|\), regardless of which state was sent. The probability of successfully identifying the state is \(p(1)+p(2) = c/2 + c/2 = c = 2-\sqrt{2} = 1 - |\braket{\psi_1|\psi_2}|\).

Fundamental Concepts and Consequences

Building upon the formalism of quantum measurement, we explore key concepts like observables, the fundamental limits imposed by the uncertainty principle, and the role of phase factors in quantum states.

Observable Operators and Spectral Decomposition

In quantum mechanics, physical quantities that can be measured, known as observables (e.g., position, momentum, energy, spin components), are represented by Hermitian operators \(\hat{A}\) (\(\hat{A}^\dagger = \hat{A}\)).

The possible outcomes of measuring the observable \(\hat{A}\) are restricted to its eigenvalues. A fundamental postulate, known as the spectral decomposition theorem, states that any Hermitian operator \(\hat{A}\) acting on a finite-dimensional Hilbert space can be written as: \[\hat{A} = \sum_m \lambda_m \hat{P}_m \label{eq:spectral_decomposition}\] where:

• \(\lambda_m\) are the distinct eigenvalues of \(\hat{A}\), which are always real numbers.

• \(\hat{P}_m\) is the orthogonal projector onto the eigenspace corresponding to the eigenvalue \(\lambda_m\). The eigenspace may be degenerate (multi-dimensional) if multiple linearly independent eigenvectors share the same eigenvalue.

• The projectors form a complete set resolution of the identity: \(\sum_m \hat{P}_m = \mathds{1}\).

• The projectors are mutually orthogonal: \(\hat{P}_m \hat{P}_n = \delta_{mn} \hat{P}_m\).

This decomposition directly links observables to projective measurements (2.2). Measuring the observable \(\hat{A}\) is equivalent to performing a projective measurement defined by the set of projectors \(\{\hat{P}_m\}\) corresponding to its spectral decomposition. The measurement outcome obtained is one of the eigenvalues \(\lambda_m\).

Expectation Value and Variance

If a system is in state \(\ket{\psi}\), the probability of obtaining the outcome \(\lambda_m\) when measuring \(\hat{A}\) is given by the Born rule using the corresponding projector: \[p(\lambda_m) = \bra{\psi} \hat{P}_m^\dagger \hat{P}_m \ket{\psi} = \bra{\psi} \hat{P}_m \ket{\psi} = \left\|\hat{P}_m \ket{\psi}\right\|^2\] The expectation value (average value) of the observable \(\hat{A}\) in the state \(\ket{\psi}\) is the weighted average of the possible outcomes: \[\left\langle\hat{A}\right\rangle_\psi = \sum_m \lambda_m p(\lambda_m) = \sum_m \lambda_m \bra{\psi} \hat{P}_m \ket{\psi} = \bra{\psi} \left( \sum_m \lambda_m \hat{P}_m \right) \ket{\psi} = \bra{\psi} \hat{A} \ket{\psi}\] The variance \((\Delta A)^2_\psi\) quantifies the spread or uncertainty in the measurement outcomes around the expectation value: \[(\Delta A)^2_\psi = \left\langle(\hat{A} - \left\langle\hat{A}\right\rangle_\psi \mathds{1})^2\right\rangle_\psi = \left\langle\hat{A}^2 - 2\hat{A}\left\langle\hat{A}\right\rangle_\psi + \left\langle\hat{A}\right\rangle_\psi^2 \mathds{1}\right\rangle_\psi = \left\langle\hat{A}^2\right\rangle_\psi - (\left\langle\hat{A}\right\rangle_\psi)^2\] The standard deviation \(\Delta A_\psi = \sqrt{(\Delta A)^2_\psi}\) is often used as the measure of uncertainty.

Heisenberg Uncertainty Principle

The uncertainty principle is not an independent axiom but a direct mathematical consequence of the operator formalism, specifically the fact that operators representing certain pairs of observables do not commute. It establishes a fundamental limit on the precision with which the values of such non-commuting observables can be simultaneously determined for a given quantum state.

Derivation using Commutators and Anticommutators

Consider two Hermitian operators (observables) \(\hat{A}\) and \(\hat{B}\), and a system in state \(\ket{\psi}\). Define the commutator \([\hat{A}, \hat{B}] = \hat{A}\hat{B} - \hat{B}\hat{A}\) and the anticommutator \(\{\hat{A}, \hat{B}\} = \hat{A}\hat{B} + \hat{B}\hat{A}\).

Let the expectation value of the product \(\hat{A}\hat{B}\) be \(\bra{\psi} \hat{A}\hat{B} \ket{\psi} = X + iY\), where \(X, Y \in \mathbb{R}\). Using the Hermiticity of \(\hat{A}\) and \(\hat{B}\), we find the expectation value of \(\hat{B}\hat{A}\): \[\bra{\psi} \hat{B}\hat{A} \ket{\psi} = \bra{\psi} \hat{B}^\dagger \hat{A}^\dagger \ket{\psi} = \braket{\hat{B}\psi | \hat{A}\psi} = \braket{\hat{A}\psi | \hat{B}\psi}^* = (\bra{\psi} \hat{A}^\dagger \hat{B} \ket{\psi})^* = (\bra{\psi} \hat{A}\hat{B} \ket{\psi})^* = (X+iY)^* = X-iY\] Now, examine the expectation values of the commutator and anticommutator: \[\begin{aligned} \left\langle[\hat{A}, \hat{B}]\right\rangle_\psi &= \bra{\psi} (\hat{A}\hat{B} - \hat{B}\hat{A}) \ket{\psi} = (X+iY) - (X-iY) = 2iY \\ \left\langle\{\hat{A}, \hat{B}\}\right\rangle_\psi &= \bra{\psi} (\hat{A}\hat{B} + \hat{B}\hat{A}) \ket{\psi} = (X+iY) + (X-iY) = 2X\end{aligned}\] The expectation value of the commutator is purely imaginary, while that of the anticommutator is purely real.

Consider the squared moduli: \[\begin{aligned} |\left\langle[\hat{A}, \hat{B}]\right\rangle_\psi|^2 &= |2iY|^2 = 4Y^2 \\ |\left\langle\{\hat{A}, \hat{B}\}\right\rangle_\psi|^2 &= |2X|^2 = 4X^2\end{aligned}\] Summing these gives: \[|\left\langle[\hat{A}, \hat{B}]\right\rangle_\psi|^2 + |\left\langle\{\hat{A}, \hat{B}\}\right\rangle_\psi|^2 = 4(X^2 + Y^2) = 4 |X + iY|^2 = 4 |\bra{\psi} \hat{A}\hat{B} \ket{\psi}|^2 \label{eq:comm_anticomm_sum_mod}\] Next, apply the Cauchy-Schwarz inequality \(|\braket{u|v}|^2 \le \braket{u|u}\braket{v|v}\) with \(\ket{u} = \hat{A}\ket{\psi}\) and \(\ket{v} = \hat{B}\ket{\psi}\): \[|\bra{\psi} \hat{A}^\dagger \hat{B} \ket{\psi}|^2 \le \braket{\psi | \hat{A}^\dagger \hat{A} | \psi} \braket{\psi | \hat{B}^\dagger \hat{B} | \psi}\] Since \(\hat{A}, \hat{B}\) are Hermitian (\(\hat{A}^\dagger=\hat{A}, \hat{B}^\dagger=\hat{B}\)): \[|\bra{\psi} \hat{A}\hat{B} \ket{\psi}|^2 \le \bra{\psi} \hat{A}^2 \ket{\psi} \bra{\psi} \hat{B}^2 \ket{\psi} = \left\langle\hat{A}^2\right\rangle_\psi \left\langle\hat{B}^2\right\rangle_\psi \label{eq:cauchy_schwarz_ab_applied}\] Combining [eq:comm_anticomm_sum_mod] and [eq:cauchy_schwarz_ab_applied]: \[|\left\langle[\hat{A}, \hat{B}]\right\rangle_\psi|^2 + |\left\langle\{\hat{A}, \hat{B}\}\right\rangle_\psi|^2 = 4 |\bra{\psi} \hat{A}\hat{B} \ket{\psi}|^2 \le 4 \left\langle\hat{A}^2\right\rangle_\psi \left\langle\hat{B}^2\right\rangle_\psi\] Since \(|\left\langle\{\hat{A}, \hat{B}\}\right\rangle_\psi|^2 \ge 0\), we can drop this non-negative term to obtain the inequality: \[|\left\langle[\hat{A}, \hat{B}]\right\rangle_\psi|^2 \le 4 \left\langle\hat{A}^2\right\rangle_\psi \left\langle\hat{B}^2\right\rangle_\psi \label{eq:pre_uncertainty_relation}\] This relation holds for any Hermitian operators \(\hat{A}, \hat{B}\). To arrive at the standard uncertainty principle, we apply this to the deviation operators: \[\begin{aligned} \Delta \hat{A} &= \hat{A} - \left\langle\hat{A}\right\rangle_\psi \mathds{1}\\ \Delta \hat{B} &= \hat{B} - \left\langle\hat{B}\right\rangle_\psi \mathds{1}\end{aligned}\] These operators are also Hermitian and represent the fluctuations around the mean values. Their commutator is the same as the original operators: \([\Delta \hat{A}, \Delta \hat{B}] = [\hat{A}, \hat{B}]\). The expectation values of their squares are the variances: \(\left\langle(\Delta \hat{A})^2\right\rangle_\psi = (\Delta A)^2_\psi\) and \(\left\langle(\Delta \hat{B})^2\right\rangle_\psi = (\Delta B)^2_\psi\). Substituting these into [eq:pre_uncertainty_relation]: \[|\left\langle[\hat{A}, \hat{B}]\right\rangle_\psi|^2 \le 4 \left\langle(\Delta \hat{A})^2\right\rangle_\psi \left\langle(\Delta \hat{B})^2\right\rangle_\psi = 4 (\Delta A)^2_\psi (\Delta B)^2_\psi\] Taking the square root yields the general form of the Heisenberg uncertainty principle, known as the Robertson-Schrödinger uncertainty relation: \[\Delta A_\psi \Delta B_\psi \ge \frac{1}{2} |\left\langle[\hat{A}, \hat{B}]\right\rangle_\psi| \label{eq:heisenberg_general_final}\]

General Statement and Interpretation

The uncertainty principle provides a quantitative lower bound on the product of the uncertainties (standard deviations) of two observables, determined by the expectation value of their commutator.

For any two Hermitian operators \(\hat{A}\) and \(\hat{B}\), and any quantum state \(\ket{\psi}\): \[\Delta A_\psi \Delta B_\psi \ge \frac{1}{2} |\left\langle[\hat{A}, \hat{B}]\right\rangle_\psi|\] where \(\Delta A_\psi = \sqrt{\left\langle\hat{A}^2\right\rangle_\psi - \left\langle\hat{A}\right\rangle_\psi^2}\) is the standard deviation of observable \(\hat{A}\) in state \(\ket{\psi}\), and similarly for \(\hat{B}\).

Interpretation:

• If two observables \(\hat{A}\) and \(\hat{B}\) commute (\([\hat{A}, \hat{B}] = 0\)), the lower bound is zero. This means it is possible, in principle, to find states (common eigenstates) where both observables have definite values simultaneously (\(\Delta A = 0\) and \(\Delta B = 0\)).

• If two observables do not commute (\([\hat{A}, \hat{B}] \ne 0\)), the lower bound is positive. This implies that it is impossible to prepare a quantum state \(\ket{\psi}\) in which both observables have arbitrarily precise values simultaneously. A state with a very small uncertainty in \(\hat{A}\) (\(\Delta A_\psi \to 0\)) must necessarily have a large uncertainty in \(\hat{B}\) (\(\Delta B_\psi \to \infty\)), and vice versa, such that their product satisfies the inequality.

• This limitation is intrinsic to the nature of quantum systems described by non-commuting operators, not a consequence of imperfect measurement devices or experimental techniques.

The position operator \(\hat{x}\) and the corresponding momentum operator \(\hat{p}_x\) satisfy the canonical commutation relation \([\hat{x}, \hat{p}_x] = i\hbar \mathds{1}\). Therefore, \(\left\langle[\hat{x}, \hat{p}_x]\right\rangle_\psi = \left\langle i\hbar \mathds{1}\right\rangle_\psi = i\hbar\). The uncertainty relation becomes: \[\Delta x_\psi \Delta p_{x,\psi} \ge \frac{1}{2} |i\hbar| = \frac{\hbar}{2}\] This famous result implies that one cannot simultaneously know both the position and momentum of a particle with arbitrary precision.

Phase Factors

Phase factors are complex numbers of modulus 1, typically written as \(e^{i\phi}\) where \(\phi \in \mathbb{R}\), that multiply quantum state vectors or components of state vectors. Their physical significance depends crucially on whether they are applied globally or relatively.

Global Phase (Physically Inconsequential)

A global phase factor multiplies the entire state vector: \(\ket{\psi'} = e^{i\phi} \ket{\psi}\). All physically measurable quantities are derived from probabilities or expectation values, which involve terms like \(|\braket{m|\psi}|^2\) or \(\bra{\psi} \hat{A} \ket{\psi}\). Let’s examine how these change for \(\ket{\psi'}\):

• Probabilities: The probability of measuring outcome \(m\) (corresponding to state \(\ket{m}\)) is: \(p'(m) = |\braket{m|\psi'}|^2 = |\braket{m|e^{i\phi}\psi}|^2 = |e^{i\phi}\braket{m|\psi}|^2 = |e^{i\phi}|^2 |\braket{m|\psi}|^2 = 1 \cdot |\braket{m|\psi}|^2 = p(m)\). The probabilities are unchanged.

• Expectation Values: The expectation value of an observable \(\hat{A}\) is: \(\left\langle\hat{A}\right\rangle_{\psi'} = \bra{\psi'} \hat{A} \ket{\psi'} = (e^{-i\phi}\bra{\psi}) \hat{A} (e^{i\phi}\ket{\psi}) = e^{-i\phi} e^{i\phi} \bra{\psi} \hat{A} \ket{\psi} = 1 \cdot \bra{\psi} \hat{A} \ket{\psi} = \left\langle\hat{A}\right\rangle_\psi\). The expectation values are unchanged.

• Inner Products: The inner product between two states \(\ket{\psi}\) and \(\ket{\chi}\) transforms as: \(\braket{\psi'|\chi'} = \braket{e^{i\phi_\psi}\psi | e^{i\phi_\chi}\chi} = e^{-i\phi_\psi} e^{i\phi_\chi} \braket{\psi|\chi}\). The modulus squared \(|\braket{\psi'|\chi'}|^2 = |\braket{\psi|\chi}|^2\) remains invariant, meaning transition probabilities are unaffected.

Since all observable predictions remain identical, the states \(\ket{\psi}\) and \(e^{i\phi}\ket{\psi}\) are considered physically equivalent. They represent the same physical state. In the geometric picture of quantum mechanics, states are often viewed as rays (one-dimensional subspaces) in Hilbert space, and all vectors within a ray differ only by a complex scalar factor. Normalization fixes the magnitude to 1, leaving only the global phase ambiguity.

Relative Phase (Physically Significant)

A relative phase appears between different components in a superposition state. Consider a general qubit state: \[\ket{\psi} = \alpha \ket{0} + \beta \ket{1}\] Writing the complex amplitudes in polar form, \(\alpha = |\alpha| e^{i\phi_\alpha}\) and \(\beta = |\beta| e^{i\phi_\beta}\): \[\ket{\psi} = |\alpha| e^{i\phi_\alpha} \ket{0} + |\beta| e^{i\phi_\beta} \ket{1}\] We can factor out an overall global phase, for instance \(e^{i\phi_\alpha}\): \[\ket{\psi} = e^{i\phi_\alpha} \left( |\alpha| \ket{0} + |\beta| e^{i(\phi_\beta - \phi_\alpha)} \ket{1} \right)\] The factor \(e^{i\phi_\alpha}\) is a global phase and is inconsequential. However, the phase difference \(\theta = \phi_\beta - \phi_\alpha\) is the relative phase between the \(\ket{0}\) and \(\ket{1}\) components. This relative phase is physically significant because it determines the specific superposition and affects interference phenomena.

Consider the Hadamard basis states: \[\begin{aligned} \ket{+} &= \frac{1}{\sqrt{2}}(\ket{0} + \ket{1}) = \frac{1}{\sqrt{2}}e^{i0}\ket{0} + \frac{1}{\sqrt{2}}e^{i0}\ket{1} \\ \ket{-} &= \frac{1}{\sqrt{2}}(\ket{0} - \ket{1}) = \frac{1}{\sqrt{2}}e^{i0}\ket{0} + \frac{1}{\sqrt{2}}e^{i\pi}\ket{1} \end{aligned}\] In \(\ket{+}\), the relative phase between \(\ket{0}\) and \(\ket{1}\) is \(\theta = 0 - 0 = 0\). In \(\ket{-}\), the relative phase between \(\ket{0}\) and \(\ket{1}\) is \(\theta = \pi - 0 = \pi\). These states are physically distinct: they are orthogonal (\(\braket{+|-}=0\)) and yield different outcomes when measured in bases other than the computational basis (e.g., measuring in the \(\{\ket{+}, \ket{-}\}\) basis itself). For instance, if we prepare \(\ket{+}\) and measure in the \(\{\ket{+}, \ket{-}\}\) basis, we always get outcome ‘+’. If we prepare \(\ket{-}\), we always get outcome ‘-’. The relative phase determines the interference properties when combining basis states.

Therefore, while global phases can be ignored, relative phases are crucial parameters defining a quantum state and cannot be disregarded. They encode important physical information.

Composite Systems and Entanglement

Quantum mechanics describes systems composed of multiple parts (e.g., multiple qubits) using the mathematical framework of tensor products. This framework naturally leads to the uniquely quantum phenomenon of entanglement.

Multiple Qubits and Tensor Product Spaces

When combining multiple quantum systems, the Hilbert space of the composite system is the tensor product of the Hilbert spaces of the individual subsystems. If system 1 is described by Hilbert space \(\mathcal{H}_1\) (with dimension \(d_1\)) and system 2 by \(\mathcal{H}_2\) (with dimension \(d_2\)), the composite system (system 1 + system 2) is described by the tensor product space: \[\mathcal{H} = \mathcal{H}_1 \otimes \mathcal{H}_2\] The dimension of the composite space is the product of the individual dimensions: \(\dim(\mathcal{H}) = d_1 \times d_2\).

For a system of \(n\) qubits, where each qubit lives in a 2-dimensional space \(\mathbb{C}^2\), the composite Hilbert space is: \[\mathcal{H}_n = (\mathbb{C}^2)^{\otimes n} = \underbrace{\mathbb{C}^2 \otimes \mathbb{C}^2 \otimes \dots \otimes \mathbb{C}^2}_{n \text{ times}}\] This space has dimension \(2^n\).

If \(\{\ket{i}_1\}\) is an orthonormal basis for \(\mathcal{H}_1\) (e.g., \(\{\ket{0}_1, \ket{1}_1\}\) for a qubit) and \(\{\ket{j}_2\}\) is an orthonormal basis for \(\mathcal{H}_2\) (e.g., \(\{\ket{0}_2, \ket{1}_2\}\)), then the set of all tensor product states \(\{\ket{i}_1 \otimes \ket{j}_2\}\) forms an orthonormal basis for the composite space \(\mathcal{H} = \mathcal{H}_1 \otimes \mathcal{H}_2\). For \(n\) qubits, the computational basis consists of the \(2^n\) states of the form \(\ket{x_1} \otimes \ket{x_2} \otimes \dots \otimes \ket{x_n}\), where each \(x_k \in \{0, 1\}\).

Notation Conventions

Several shorthand notations are commonly used for tensor product states to simplify expressions:

• Explicit Tensor Product: \(\ket{\psi}_1 \otimes \ket{\phi}_2\)

• Implicit Tensor Product (Juxtaposition): \(\ket{\psi}_1 \ket{\phi}_2\) or simply \(\ket{\psi} \ket{\phi}\) if the subsystems are clear from context.

• Combined Ket: \(\ket{\psi \phi}\) (less common for general states, frequent for basis states).

• Indexed Ket (Computational Basis): \(\ket{x_1 x_2 \dots x_n} \equiv \ket{x_1}_1 \otimes \ket{x_2}_2 \otimes \dots \otimes \ket{x_n}_n\). For example, \(\ket{01}\) represents \(\ket{0}_1 \otimes \ket{1}_2\), and \(\ket{101}\) represents \(\ket{1}_1 \otimes \ket{0}_2 \otimes \ket{1}_3\). The position within the ket label usually indicates the subsystem (e.g., first position for qubit 1, second for qubit 2, etc.).

Operators on Composite Systems

Operators acting on the composite space \(\mathcal{H} = \mathcal{H}_1 \otimes \mathcal{H}_2\) can be constructed from operators acting on the individual subsystems.

• Local Operators: An operator \(\hat{A}\) that acts only on system 1 is formally represented on the composite space as \(\hat{A} \otimes \mathds{1}_2\), where \(\mathds{1}_2\) is the identity operator on \(\mathcal{H}_2\). Similarly, an operator \(\hat{B}\) acting only on system 2 is \(\mathds{1}_1 \otimes \hat{B}\). Often, the identity operator is implicitly assumed and omitted for brevity. For example, if we write \(\hat{\sigma}_x\ket{01}\), and \(\hat{\sigma}_x\) is stated to act on the first qubit, it means \((\hat{\sigma}_x\otimes \mathds{1}) (\ket{0}_1 \otimes \ket{1}_2) = (\hat{\sigma}_x\ket{0}_1) \otimes (\mathds{1}\ket{1}_2) = \ket{1}_1 \otimes \ket{1}_2 = \ket{11}\).

• Combined Operators: Operators acting non-trivially on multiple subsystems can be formed by tensor products (e.g., \(\hat{A} \otimes \hat{B}\)) or sums of such products (representing interactions). The action of \(\hat{A} \otimes \hat{B}\) on a product state is \((\hat{A} \otimes \hat{B}) (\ket{\psi}_1 \otimes \ket{\phi}_2) = (\hat{A}\ket{\psi}_1) \otimes (\hat{B}\ket{\phi}_2)\). Its action on general states is defined by linearity.

System Evolution and Entanglement Generation

The way a composite system evolves depends on whether its subsystems interact.

Independent Evolution (Non-Interacting Systems)

If the subsystems do not interact with each other, the total Hamiltonian is simply the sum of the individual Hamiltonians: \[\hat{H} = \hat{H}_1 \otimes \mathds{1}_2 + \mathds{1}_1 \otimes \hat{H}_2\] Since the terms \(\hat{H}_1 \otimes \mathds{1}_2\) and \(\mathds{1}_1 \otimes \hat{H}_2\) commute, the time evolution operator \(\hat{U}(t) = e^{-i\hat{H}t/\hbar}\) factors into a tensor product: \[\begin{aligned} \hat{U}(t) &= e^{-i(\hat{H}_1 \otimes \mathds{1}_2 + \mathds{1}_1 \otimes \hat{H}_2)t/\hbar} \\ &= e^{-i(\hat{H}_1 \otimes \mathds{1}_2)t/\hbar} e^{-i(\mathds{1}_1 \otimes \hat{H}_2)t/\hbar} \quad (\text{because the terms commute}) \\ &= (e^{-i\hat{H}_1 t/\hbar} \otimes \mathds{1}_2) (\mathds{1}_1 \otimes e^{-i\hat{H}_2 t/\hbar}) \\ &= \hat{U}_1(t) \otimes \hat{U}_2(t)\end{aligned}\] where \(\hat{U}_1(t) = e^{-i\hat{H}_1 t/\hbar}\) and \(\hat{U}_2(t) = e^{-i\hat{H}_2 t/\hbar}\) are the evolution operators for the individual subsystems. If the initial state is a product state \(\ket{\psi(0)} = \ket{\psi_1(0)} \otimes \ket{\psi_2(0)}\), its state at time \(t\) is: \[\ket{\psi(t)} = (\hat{U}_1(t) \otimes \hat{U}_2(t)) (\ket{\psi_1(0)} \otimes \ket{\psi_2(0)}) = (\hat{U}_1(t)\ket{\psi_1(0)}) \otimes (\hat{U}_2(t)\ket{\psi_2(0)}) = \ket{\psi_1(t)} \otimes \ket{\psi_2(t)}\] Crucially, if the subsystems evolve independently, an initial product state remains a product state at all times. No entanglement is generated.

Interacting Systems and Entanglement Generation

If the subsystems interact, the total Hamiltonian includes an interaction term \(\hat{H}_{\text{int}}\) that acts non-trivially on both subsystems (i.e., it cannot be written as \(\hat{A}\otimes\mathds{1}+ \mathds{1}\otimes\hat{B}\)): \[\hat{H} = \hat{H}_1 \otimes \mathds{1}_2 + \mathds{1}_1 \otimes \hat{H}_2 + \hat{H}_{\text{int}}\] In this case, the terms in the Hamiltonian generally do not commute, and the evolution operator \(\hat{U}(t) = e^{-i\hat{H}t/\hbar}\) does not factor into a simple tensor product \(\hat{U}_1(t) \otimes \hat{U}_2(t)\). Applying such an evolution operator (representing an interacting gate) to an initial product state can result in a final state that is no longer a product state – it becomes entangled.

The Controlled-NOT (CNOT) gate is a fundamental two-qubit gate representing an interaction. Its action on the computational basis states \(\ket{xy}\) (where \(x\) is the control qubit and \(y\) is the target qubit) is defined as \(\text{CNOT}\ket{xy} = \ket{x(y \oplus x)}\), where \(\oplus\) denotes addition modulo 2. Consider applying CNOT to the initial product state \(\ket{+0} = \ket{+}_1 \otimes \ket{0}_2 = \frac{1}{\sqrt{2}}(\ket{0}_1 + \ket{1}_1) \otimes \ket{0}_2 = \frac{1}{\sqrt{2}}(\ket{00} + \ket{10})\). \[\begin{aligned} \text{CNOT} \ket{+0} &= \text{CNOT} \left[ \frac{1}{\sqrt{2}}(\ket{00} + \ket{10}) \right] \\ &= \frac{1}{\sqrt{2}} (\text{CNOT}\ket{00} + \text{CNOT}\ket{10}) \quad (\text{by linearity}) \\ &= \frac{1}{\sqrt{2}} (\ket{0(0 \oplus 0)} + \ket{1(0 \oplus 1)}) \\ &= \frac{1}{\sqrt{2}} (\ket{00} + \ket{11}) \end{aligned}\] The resulting state, \(\frac{1}{\sqrt{2}}(\ket{00} + \ket{11})\), is the famous Bell state \(\ket{\Phi^+}\). This state cannot be written as a simple tensor product \(\ket{\psi}_1 \otimes \ket{\phi}_2\), demonstrating that the CNOT interaction has generated entanglement from an initially separable state.

Entanglement

Entanglement is a key feature distinguishing quantum mechanics from classical physics, representing correlations between quantum systems that are stronger than any classical correlation.

Definition: Separable vs. Entangled States

A pure state \(\ket{\Psi}\) in a composite Hilbert space \(\mathcal{H} = \mathcal{H}_1 \otimes \mathcal{H}_2\) is called separable (or a product state) if it can be written in the form: \[\ket{\Psi} = \ket{\psi}_1 \otimes \ket{\phi}_2\] for some state \(\ket{\psi}_1 \in \mathcal{H}_1\) and some state \(\ket{\phi}_2 \in \mathcal{H}_2\).

A pure state \(\ket{\Psi}\) in a composite Hilbert space \(\mathcal{H} = \mathcal{H}_1 \otimes \mathcal{H}_2\) is called entangled if it is not separable.

Entangled states describe situations where the quantum state of the composite system is well-defined, but the states of the individual subsystems are not independently defined. Measuring one subsystem provides information about the other, regardless of the distance separating them.

Example: The Bell States

The four Bell states are canonical examples of maximally entangled states for a two-qubit system. They form an orthonormal basis for the \(\mathbb{C}^2 \otimes \\C^2 = \mathbb{C}^4\) Hilbert space: \[\begin{aligned} \ket{\Phi^+} &= \frac{1}{\sqrt{2}}(\ket{00} + \ket{11}) \label{eq:bell_phi_plus_def}\\ \ket{\Phi^-} &= \frac{1}{\sqrt{2}}(\ket{00} - \ket{11}) \label{eq:bell_phi_minus_def}\\ \ket{\Psi^+} &= \frac{1}{\sqrt{2}}(\ket{01} + \ket{10}) \label{eq:bell_psi_plus_def}\\ \ket{\Psi^-} &= \frac{1}{\sqrt{2}}(\ket{01} - \ket{10}) \label{eq:bell_psi_minus_def}\end{aligned}\] It can be shown that none of these four states can be factored into a tensor product \(\ket{\psi}_1 \otimes \ket{\phi}_2\). They represent the strongest possible correlation between two qubits.

Properties: Strong Correlations and Non-Locality

Entangled states exhibit correlations that are fundamentally quantum mechanical. Consider Alice holding the first qubit and Bob holding the second qubit of the Bell state \(\ket{\Phi^+} = \frac{1}{\sqrt{2}}(\ket{00} + \ket{11})\). If Alice measures her qubit in the computational basis \(\{\ket{0}, \ket{1}\}\):

• If Alice obtains outcome 0 (with probability \(|\frac{1}{\sqrt{2}}|^2 = 1/2\)), the state of the composite system instantaneously collapses to \(\ket{00}\). If Bob subsequently measures his qubit, he is guaranteed to find it in state \(\ket{0}\).

• If Alice obtains outcome 1 (with probability \(|\frac{1}{\sqrt{2}}|^2 = 1/2\)), the state collapses to \(\ket{11}\). Bob is guaranteed to find his qubit in state \(\ket{1}\).

Alice’s measurement outcome appears random to her (50

Despite the instantaneous nature of state collapse and the perfect correlations, entanglement cannot be used to transmit information faster than the speed of light. Alice cannot control her measurement outcome; it is fundamentally probabilistic. Therefore, she cannot use her choice of measurement or her outcome to send a predetermined message to Bob. Bob’s measurement outcomes, considered alone without knowledge of Alice’s results, also appear completely random (50

Entanglement vs. Hidden Variables (Bell’s Theorem)

The seemingly paradoxical nature of entanglement led Einstein, Podolsky, and Rosen (EPR) in 1935 to question the completeness of quantum mechanics. They argued that the strong correlations implied the existence of pre-determined properties (local "hidden variables") that were simply unknown to the observer before measurement, analogous to classical correlations (e.g., knowing one glove is left implies the other, distant one is right, a property determined when they were paired).

However, in 1964, John S. Bell proved a remarkable theorem. He showed that any theory based on the assumptions of locality (no instantaneous influence at a distance) and realism (properties exist independent of measurement – the core idea of hidden variables) must satisfy certain statistical constraints known as Bell’s inequalities. Quantum mechanics, on the other hand, predicts violations of these inequalities for measurements on entangled states performed in appropriately chosen bases.

Numerous experiments, starting with those by Alain Aspect in the early 1980s and continuing with increasing precision, have consistently confirmed the predictions of quantum mechanics and demonstrated violations of Bell’s inequalities. This provides strong evidence against local hidden variable theories and supports the fundamentally non-local and non-realist nature of quantum entanglement. The properties revealed by measuring entangled particles are not predetermined but are genuinely indefinite until the measurement occurs.

Remark. Bell’s theorem demonstrates that no physical theory of local hidden variables can ever reproduce all of the predictions of quantum mechanics. It shows that the correlations predicted by quantum mechanics for entangled states are incompatible with the joint assumptions of locality and realism. Experimental violations of Bell inequalities confirm this incompatibility, suggesting that at least one of these classical assumptions must be abandoned.

Entanglement and the Measurement Problem

Entanglement is deeply intertwined with the measurement problem in quantum mechanics – the question of how and why the deterministic, unitary evolution described by the Schrödinger equation gives way to probabilistic outcomes and state collapse upon measurement. When measuring one part of an entangled system, the state of the entire system, including distant parts, appears to change instantly.

Different interpretations of quantum mechanics offer varying perspectives:

• Copenhagen Interpretation: Often invokes a somewhat artificial distinction between the quantum system and the classical measurement apparatus, accepting state collapse as a fundamental process occurring upon interaction with the classical realm.

• Many-Worlds Interpretation (MWI): Proposed by Hugh Everett III, MWI avoids state collapse entirely. It posits that the measurement interaction causes the observer and the system to become entangled, and the entire combined system (including the observer and environment) evolves unitarily into a superposition of branches, each representing a possible outcome. From the perspective of an observer within a specific branch, it appears as if a collapse occurred, but the universal wave function never collapses. While conceptually radical (implying a constantly splitting multiverse), MWI is mathematically parsimonious as it relies solely on unitary evolution without additional collapse postulates.

Understanding entanglement is crucial for understanding these foundational issues, although for most practical applications in quantum computation and information, the standard formalism including the collapse postulate is sufficient.

Applications of Quantum Principles

Entanglement, beyond its foundational significance, serves as a crucial resource for quantum information processing tasks, enabling capabilities beyond classical limits. One such example is superdense coding.

Superdense Coding

Superdense coding is a quantum communication protocol that remarkably allows one party (Alice) to transmit two classical bits of information to another party (Bob) by sending only a single qubit through a quantum channel, provided they share a pair of entangled qubits beforehand.

Protocol Setup: Shared Entangled Pair

1. Entanglement Distribution: An entangled pair of qubits is created, typically in the Bell state \(\ket{\Phi^+} = \frac{1}{\sqrt{2}}(\ket{00} + \ket{11})\).

2. Qubit Distribution: One qubit of the pair (let’s call it qubit A) is given to Alice, and the other qubit (qubit B) is given to Bob. They may then move apart, maintaining their respective qubits in isolation.

This shared entanglement is the key resource enabling the protocol.

Encoding Classical Bits (Alice’s Operations)

Alice wishes to send one of the four possible two-bit classical messages (00, 01, 10, or 11) to Bob. To do this, she performs a specific local unitary operation (one of the Pauli gates or the identity) on her qubit (A). The choice of operation depends deterministically on the two classical bits she wants to send:

• To send "00": Alice applies the Identity gate (\(\hat{I} = \hat{\sigma}_0\)) to her qubit A. The state of the shared pair remains unchanged: \((\hat{I} \otimes \hat{I}) \ket{\Phi^+} = \ket{\Phi^+} = \frac{1}{\sqrt{2}}(\ket{00} + \ket{11})\).

• To send "01": Alice applies the Pauli-X gate (\(\hat{X} = \hat{\sigma}_x\)) to her qubit A. The resulting state is: \((\hat{\sigma}_x\otimes \hat{I}) \ket{\Phi^+} = (\hat{\sigma}_x\otimes \hat{I}) \frac{1}{\sqrt{2}}(\ket{00} + \ket{11}) = \frac{1}{\sqrt{2}}(\hat{\sigma}_x\ket{0}\otimes\ket{0} + \hat{\sigma}_x\ket{1}\otimes\ket{1}) = \frac{1}{\sqrt{2}}(\ket{10} + \ket{01}) = \ket{\Psi^+}\).

• To send "10": Alice applies the Pauli-Z gate (\(\hat{Z} = \hat{\sigma}_z\)) to her qubit A. The resulting state is: \((\hat{\sigma}_z\otimes \hat{I}) \ket{\Phi^+} = (\hat{\sigma}_z\otimes \hat{I}) \frac{1}{\sqrt{2}}(\ket{00} + \ket{11}) = \frac{1}{\sqrt{2}}(\hat{\sigma}_z\ket{0}\otimes\ket{0} + \hat{\sigma}_z\ket{1}\otimes\ket{1}) = \frac{1}{\sqrt{2}}(\ket{00} - \ket{11}) = \ket{\Phi^-}\).

• To send "11": Alice applies the Pauli-Y gate (\(\hat{Y} = \hat{\sigma}_y= i\hat{\sigma}_x\hat{\sigma}_z\)) to her qubit A. (Note: Applying \(\hat{Y}\) is equivalent to applying \(\hat{X}\hat{Z}\) up to a phase \(i\). Some sources use \(\hat{X}\hat{Z}\) directly for 11). \((\hat{\sigma}_y\otimes \hat{I}) \ket{\Phi^+} = (\hat{\sigma}_y\otimes \hat{I}) \frac{1}{\sqrt{2}}(\ket{00} + \ket{11}) = \frac{1}{\sqrt{2}}(\hat{\sigma}_y\ket{0}\otimes\ket{0} + \hat{\sigma}_y\ket{1}\otimes\ket{1}) = \frac{1}{\sqrt{2}}(i\ket{1}\otimes\ket{0} + (-i)\ket{0}\otimes\ket{1}) = \frac{i}{\sqrt{2}}(\ket{10} - \ket{01}) = -i \ket{\Psi^-}\). The state is \(\ket{\Psi^-}\) up to the global phase \(-i\).

Notice that Alice’s local operation transforms the initial shared Bell state \(\ket{\Phi^+}\) into one of the four orthogonal Bell states, uniquely determined by the two classical bits she intended to send.

After performing her local operation, Alice sends her qubit (qubit A) to Bob via a quantum channel. Bob now possesses both qubits, A and B, which are collectively in one of the four Bell states.

Decoding Procedure (Bob’s Operations and Measurement)

To decode Alice’s message, Bob needs to distinguish which of the four Bell states he holds. This can be achieved by performing a measurement in the Bell basis. A standard circuit to perform a Bell basis measurement involves a CNOT gate followed by a Hadamard gate, and then measurement in the computational basis:

1. Receive Qubit: Bob receives qubit A from Alice. He now holds both qubits A and B.

2. Apply CNOT Gate: Bob applies a CNOT gate, using qubit A (received from Alice) as the control and his original qubit B as the target.

3. Apply Hadamard Gate: Bob applies a Hadamard gate (\(\hat{H}\)) to qubit A.

4. Measure: Bob measures both qubits (A and B) in the computational basis \(\{\ket{0}, \ket{1}\}\). The two-bit outcome directly reveals the classical message Alice sent.

Let’s trace the decoding process for each of the four possible Bell states Bob might receive:

• Input \(\ket{\Phi^+} = \frac{1}{\sqrt{2}}(\ket{00} + \ket{11})\) (Alice sent 00):

  1. CNOT(A,B): \(\frac{1}{\sqrt{2}}(\text{CNOT}\ket{00} + \text{CNOT}\ket{11}) = \frac{1}{\sqrt{2}}(\ket{00} + \ket{10}) = \frac{1}{\sqrt{2}}(\ket{0}+\ket{1})\otimes\ket{0} = \ket{+}\otimes\ket{0}\).

  2. H(A): \((\hat{H} \otimes \hat{I}) (\ket{+}\otimes\ket{0}) = (\hat{H}\ket{+})\otimes\ket{0} = \ket{0}\otimes\ket{0} = \ket{00}\).

  3. Measurement: Bob measures 00.

• Input \(\ket{\Psi^+} = \frac{1}{\sqrt{2}}(\ket{01} + \ket{10})\) (Alice sent 01):

  1. CNOT(A,B): \(\frac{1}{\sqrt{2}}(\text{CNOT}\ket{01} + \text{CNOT}\ket{10}) = \frac{1}{\sqrt{2}}(\ket{01} + \ket{11}) = \frac{1}{\sqrt{2}}(\ket{0}+\ket{1})\otimes\ket{1} = \ket{+}\otimes\ket{1}\).

  2. H(A): \((\hat{H} \otimes \hat{I}) (\ket{+}\otimes\ket{1}) = (\hat{H}\ket{+})\otimes\ket{1} = \ket{0}\otimes\ket{1} = \ket{01}\).

  3. Measurement: Bob measures 01.

• Input \(\ket{\Phi^-} = \frac{1}{\sqrt{2}}(\ket{00} - \ket{11})\) (Alice sent 10):

  1. CNOT(A,B): \(\frac{1}{\sqrt{2}}(\text{CNOT}\ket{00} - \text{CNOT}\ket{11}) = \frac{1}{\sqrt{2}}(\ket{00} - \ket{10}) = \frac{1}{\sqrt{2}}(\ket{0}-\ket{1})\otimes\ket{0} = \ket{-}\otimes\ket{0}\).

  2. H(A): \((\hat{H} \otimes \hat{I}) (\ket{-}\otimes\ket{0}) = (\hat{H}\ket{-})\otimes\ket{0} = \ket{1}\otimes\ket{0} = \ket{10}\).

  3. Measurement: Bob measures 10.

• Input \(\ket{\Psi^-} = \frac{1}{\sqrt{2}}(\ket{01} - \ket{10})\) (Alice sent 11): (Ignoring the potential global phase \(-i\) from Alice’s \(\hat{\sigma}_y\) operation, as it doesn’t affect measurement)

  1. CNOT(A,B): \(\frac{1}{\sqrt{2}}(\text{CNOT}\ket{01} - \text{CNOT}\ket{10}) = \frac{1}{\sqrt{2}}(\ket{01} - \ket{11}) = \frac{1}{\sqrt{2}}(\ket{0}-\ket{1})\otimes\ket{1} = \ket{-}\otimes\ket{1}\).

  2. H(A): \((\hat{H} \otimes \hat{I}) (\ket{-}\otimes\ket{1}) = (\hat{H}\ket{-})\otimes\ket{1} = \ket{1}\otimes\ket{1} = \ket{11}\).

  3. Measurement: Bob measures 11.

The sequence of CNOT and Hadamard gates effectively transforms the Bell basis states into the computational basis states \(\ket{00}, \ket{01}, \ket{10}, \ket{11}\), allowing Bob to perfectly determine which Bell state he received, and thus which two classical bits Alice intended to send, by performing a simple measurement in the computational basis.

& & & & & &
& & & & & &

Quantum circuit diagram for Superdense Coding. Initially, Alice and Bob share the entangled state (). Alice applies a local gate (_{msg} {, _x, _z, _y}) corresponding to the two classical bits she wants to send. She then sends her qubit (A) to Bob. Bob performs a CNOT gate (A controlling B) followed by a Hadamard gate on A, and then measures both qubits in the computational basis to retrieve the two classical bits.

Information Transfer Efficiency

The remarkable feature of superdense coding is its efficiency. Alice transmits two classical bits of information (encoded in her choice of \(\hat{U}_{msg}\)) to Bob, but she only needs to physically send one qubit (qubit A) through the quantum channel. Classically, sending two bits would require sending two physical signals. This doubling of communication capacity is enabled by the pre-shared entanglement between Alice and Bob. The entanglement acts as a resource that allows Alice’s single qubit to carry more information than classically possible. This highlights how quantum resources can enhance information processing tasks.

Remark. The protocol requires the reliable creation and distribution of entangled pairs and the ability to maintain qubit coherence until Bob performs his measurement. It also requires a quantum channel capable of transmitting qubit A from Alice to Bob without destroying its quantum state. Noise and decoherence in any of these steps can reduce the fidelity of the protocol and introduce errors in the transmitted information.

Conclusion

This lecture covered several essential concepts in quantum mechanics and information. We explored how quantum gates can be physically realized through controlled interactions, using spin systems as an example. The formalism of quantum measurement was detailed, including general measurements (operators \(\hat{M}_m\)), projective measurements (projectors \(\hat{P}_m\)), and POVMs (\(\hat{E}_m\)), emphasizing the probabilistic outcomes, state collapse, and the completeness relation \(\sum \hat{M}_m^\dagger \hat{M}_m = \mathds{1}\) (or \(\sum \hat{E}_m = \mathds{1}\)). We established that orthogonal states can be perfectly distinguished, while non-orthogonal states cannot, leading to the utility of POVMs for partial or unambiguous discrimination.

Fundamental consequences like the Heisenberg Uncertainty Principle (\(\Delta A \Delta B \ge \frac{1}{2} |\left\langle[\hat{A}, \hat{B}]\right\rangle|\)) were derived from operator algebra, highlighting the intrinsic limits on simultaneous knowledge of non-commuting observables. The distinction between inconsequential global phases and physically significant relative phases was clarified.

The description of composite systems using tensor products (\(\mathcal{H}_1 \otimes \mathcal{H}_2\)) led to the concept of entanglement – states that cannot be factored into products of subsystem states (e.g., Bell states). Entanglement enables strong, non-local correlations that violate Bell’s inequalities, ruling out local hidden variable theories.

Finally, the superdense coding protocol showcased a practical application of entanglement, allowing the transmission of two classical bits by sending only one qubit, leveraging the correlations of a shared Bell state.

Key takeaways include the power of unitary evolution for gate implementation, the probabilistic and state-altering nature of quantum measurement, the fundamental limits imposed by the uncertainty principle, and the unique properties and applications of entanglement as a quantum resource.

Follow-up Questions/Topics:

• How are specific POVMs physically implemented? (Neumark’s extension theorem)

• Exploration of Bell’s theorem and experimental tests.

• Further applications of entanglement: Quantum teleportation, entanglement swapping.

• Introduction to density matrices for describing mixed states and decoherence.

• Quantum error correction codes.