Skip to content

Information-Theoretic Foundations

Information-theoretic foundations turn a formal quantum model into an operational question. Given a problem or claim, the goal is to identify the carrier and encoding, state or labeled ensemble, allowed transformation, measurement, classical output, figure of merit, and resource restrictions before choosing a calculation. This guide supplies that routing discipline and two small numerical audits. It does not replace the canonical derivations of state space, measurement theory, entropy inequalities, Schmidt decomposition, or no-go theorems.

Required background. Quantum States supplies the distinction between a state and one of its representations, while the Born Rule supplies the probability audit used throughout.

Helpful background. Tensor Products prepares the subsystem language, and Entropy prepares the classical information conventions.

Information Foundations as an Operational Layer

Section titled “Information Foundations as an Operational Layer”

A useful information-processing description follows a chain rather than starting with a favorite formula:

source or preparation
-> state or labeled ensemble
-> channel or instrument
-> measurement
-> classical record
-> declared loss, payoff, rate, or resource claim

A classical source uses symbols xx with probabilities pxp_x. A quantum source may attach a state ρx\rho_x to each label, producing the ensemble {px,ρx}\{p_x,\rho_x\} and the receiver state

ρˉ=∑xpxρx.\bar\rho=\sum_x p_x\rho_x.

The average state is not the whole source specification when correlations with the label xx matter. Two different ensembles can have the same ρˉ\bar\rho and therefore be indistinguishable to a receiver who lacks the label, while supporting different preparation stories or later side information.

If a channel N\mathcal N is followed by an instrument {Iy}\{\mathcal I_y\}, then the probability of record yy conditional on source label xx is

p(y∣x)=Tr⁡ ⁣[Iy ⁣(N(ρx))]=Tr⁡ ⁣[MyN(ρx)],My=Iy†(I).\begin{aligned} p(y\mid x) &=\operatorname{Tr}\!\left[\mathcal I_y\!\left(\mathcal N(\rho_x)\right)\right]\\ &=\operatorname{Tr}\!\left[M_y\mathcal N(\rho_x)\right], \qquad M_y=\mathcal I_y^\dagger(I). \end{aligned}

The effects MyM_y determine outcome probabilities, but they do not by themselves determine the conditional output states. That distinction is why an operational question must name the instrument, not merely a POVM, whenever backaction or future use of the state matters. The full theory belongs to Generalized Measurements and Instruments.

The chapter can be entered from physics, computer science, or information theory. Use the smallest repair that makes the next operational statement unambiguous.

Capability to testSubstantive repair or next step
Normalize a state, distinguish a ray from a coordinate vector, and state a basisQuantum States, Rays and Global Phase, and Finite-Dimensional Hilbert Spaces
Compute probabilities for a declared measurementBorn Rule
Keep subsystem order and local versus global operators straightTensor Products
Compute H(X)H(X), conditional entropy, and classical relative entropy with a declared logarithm base and supportEntropy and Relative Entropy
Distinguish a physical device from an abstract or encoded carrierBits, Qubits, Qudits, and Modes
Work with noisy preparations, discarded subsystems, or measurement branchesDensity Operators for Quantum Information

These are capability checks, not a demand to reread an entire prerequisite tree. A physics reader may need the source-channel-decision vocabulary; a computer-science reader may need state, measurement, and subsystem discipline; an information theorist may need the distinction between a classical channel and a completely positive map. The Quantum Information Roadmap supplies a longer sequence, while Math Needed for Quantum Information supplies targeted mathematical repairs.

Carriers, Encodings, and State-Space Choices

Section titled “Carriers, Encodings, and State-Space Choices”

An information carrier is an abstract state space together with a declared encoding and an allowed set of operations and measurements. The same abstract qubit can be realized by many devices, and one device can support several inequivalent encodings.

TermOperational question
Classical bitWhich two distinguishable symbols are used, and what stochastic errors act on them?
QubitWhich two-dimensional Hilbert space or code subspace carries the state, and which basis defines the labels 00 and 11?
QuditWhat is the dimension dd, which levels form the computational basis, and which operations preserve the chosen subspace?
ModeWhich field or oscillator degree of freedom is used, and is the encoding continuous-variable, finite-energy, or restricted to a code subspace?
Physical qubitWhich controlled device degrees of freedom implement preparation, gates, storage, and readout?
Logical qubitWhich code, syndrome process, decoder, and logical operators define protected information?

The canonical carrier and encoding taxonomy is Bits, Qubits, Qudits, and Modes. General facts about Cd\mathbb C^d, bases, rays, and spectra remain at their Mathematical Toolkit and Core Formalism homes.

For nn qubits, the register dimension is 2n2^n, but this does not mean that one run reveals 2n2^n classical numbers. A measurement returns a classical record drawn from a declared outcome set. State-description size, accessible information, channel capacity, sample complexity, and computational cost are different quantities and must not be exchanged without an argument.

Before using a state vector or matrix, record the computational basis and subsystem order. For example, the column (0,1,0,0)T(0,1,0,0)^T is not operationally identified until one says whether the basis is ∣00⟩,∣01⟩,∣10⟩,∣11⟩|00\rangle,|01\rangle,|10\rangle,|11\rangle or follows another convention.

Qubit Geometry and Calibrated Measurements

Section titled “Qubit Geometry and Calibrated Measurements”

Every one-qubit density operator has the exact representation

ρ=12(I+r⋅σ),∥r∥≤1.\rho=\frac12\left(I+\mathbf r\cdot\boldsymbol\sigma\right), \qquad \lVert\mathbf r\rVert\le 1.

The vector r\mathbf r is operational: its components are the Pauli expectation values. For a projective measurement along unit vector n\mathbf n,

p(+n)=1+n⋅r2.p(+\mathbf n)=\frac{1+\mathbf n\cdot\mathbf r}{2}.

The Bloch Sphere for Quantum Information owns the calibrated connection among states, measurement axes, tomography, gate rotations, and affine noise maps. The picture is exact for one qubit, not a generic ball representation for qudits, registers, or continuous-variable modes.

Numerical routing audit: a noisy qubit. Consider

ρ=(0.80.30.30.2).\rho= \begin{pmatrix} 0.8 & 0.3\\ 0.3 & 0.2 \end{pmatrix}.

The trace is one, and comparison with the Bloch form gives

r=(0.6,0,0.6),∥r∥=0.72=0.848528.\mathbf r=(0.6,0,0.6), \qquad \lVert\mathbf r\rVert=\sqrt{0.72}=0.848528.

Its eigenvalues, purity, and entropy are

λ±=1±0.722=0.924264, 0.075736,Tr⁡(ρ2)=0.86,S(ρ)=h2(0.075736)=0.386973 bits.\begin{aligned} \lambda_\pm &=\frac{1\pm\sqrt{0.72}}{2} =0.924264,\ 0.075736,\\ \operatorname{Tr}(\rho^2)&=0.86,\\ S(\rho)&=h_2(0.075736)=0.386973\ \text{bits}. \end{aligned}

The two immediate Born checks are p(+x)=0.8p(+x)=0.8 and p(0z)=0.8p(0z)=0.8. Positivity, purity, and possible ensemble interpretations belong to Density Operators for Quantum Information; axes and one-qubit readout geometry belong to the Bloch-sphere page; entropy interpretation belongs to Quantum Entropy. A physical channel or backaction question exits to Measurement and Open Systems. The number S(ρ)S(\rho) is not, by itself, the accessible information of an unspecified source ensemble.

Mixed States, Reductions, and Purifications

Section titled “Mixed States, Reductions, and Purifications”

A finite-dimensional density operator satisfies

ρ=ρ†,ρ≥0,Tr⁡ρ=1.\rho=\rho^\dagger, \qquad \rho\ge 0, \qquad \operatorname{Tr}\rho=1.

Those conditions describe the operational state available to the observer. They do not select a unique ensemble decomposition. If

ρ=∑xpx∣ψx⟩⟨ψx∣,\rho=\sum_x p_x|\psi_x\rangle\langle\psi_x|,

the label xx may represent accessible preparation information, an inaccessible record, or merely one mathematical decomposition among many. Keeping an accessible label requires the classical-quantum state

ρXQ=∑xpx∣x⟩⟨x∣X⊗ρx,\rho_{XQ}=\sum_x p_x|x\rangle\langle x|_X\otimes\rho_x,

not only the marginal ρˉQ\bar\rho_Q.

A subsystem can also be mixed because it is entangled with something discarded. For a joint state ρAB\rho_{AB}, the reduced state ρA=Tr⁡BρAB\rho_A=\operatorname{Tr}_B\rho_{AB} predicts every local measurement on AA, but it does not retain the correlations with BB. Conversely, a purification represents a mixed state as the marginal of a larger pure state; the purifying system and basis are not unique.

Pure versus Mixed States owns the general criteria. Density Operators for Quantum Information owns the protocol workflow involving source ensembles, classical labels, reduced states, purification, channels, instruments, trace distance, and fidelity. Composite Systems owns the full partial-trace and purification constructions.

An instrument branch Iy(ρ)\mathcal I_y(\rho) can be subnormalized: its trace is the probability of outcome yy. Only after division by that probability, when nonzero, is the conditional output a normalized state. Confusing these two objects is a common source of incorrect probabilities and nonlinear update rules.

Entropy, Correlation, and Distinguishability

Section titled “Entropy, Correlation, and Distinguishability”

Use base-22 logarithms here, so information is reported in bits. The principal quantum entropy is

S(ρ)=−Tr⁡(ρlog⁡2ρ).S(\rho)=-\operatorname{Tr}(\rho\log_2\rho).

It depends on the spectrum of ρ\rho, not on the outcome distribution of an arbitrarily chosen measurement. Several related quantities answer different questions:

QuantityQuestion and canonical scope
H(X)H(X)Uncertainty of a classical label; see Mathematical Toolkit Entropy.
S(ρ)S(\rho)Spectral mixedness of a quantum state; see Quantum Entropy.
I(A:B)=S(A)+S(B)−S(AB)I(A{:}B)=S(A)+S(B)-S(AB)Total correlation between declared subsystems; it is not generally an entanglement measure.
D(ρ∥σ)D(\rho\|\sigma)Directed distinguishability with a support condition; its monotonicity under channels is data processing.
χ=S(ρˉ)−∑xpxS(ρx)\chi=S(\bar\rho)-\sum_xp_xS(\rho_x)The Holevo quantity of a labeled ensemble; it bounds accessible classical information but need not be attained.

Quantum conditional entropy S(A∣B)=S(AB)−S(B)S(A|B)=S(AB)-S(B) can be negative. That is not a failed probability calculation; it is a quantum correlation statement whose operational meaning depends on the task. Likewise, I(A:B)>0I(A{:}B)>0 detects total correlation, including classical correlation in separable mixed states.

Quantum Entropy owns the QI definitions, Rényi variants, conditional and mutual information, quantum relative entropy, data processing, strong subadditivity, Holevo quantities, and finite-dimensional cautions. Mutual Information owns structural bipartite examples, while Mathematical Toolkit Relative Entropy owns classical Kullback–Leibler divergence, support conditions, and statistical interpretation.

Schmidt Spectra and Entanglement Diagnostics

Section titled “Schmidt Spectra and Entanglement Diagnostics”

For a finite-dimensional pure bipartite state, the Schmidt decomposition has the form

∣ψ⟩AB=∑iλi ∣iA⟩∣iB⟩,λi≥0,∑iλi=1.|\psi\rangle_{AB} =\sum_i\sqrt{\lambda_i}\,|i_A\rangle|i_B\rangle, \qquad \lambda_i\ge0, \qquad \sum_i\lambda_i=1.

The nonzero λi\lambda_i are the eigenvalues of both reduced states. Schmidt Decomposition owns the theorem, singular-value proof, reduced spectra, rank, and basis freedom. This chapter uses the spectrum to select an operational quantity.

For a pure bipartite state, the entropy of entanglement is

E(∣ψ⟩AB)=S(ρA)=S(ρB).E(|\psi\rangle_{AB})=S(\rho_A)=S(\rho_B).

That equality is special to pure bipartite states. Mixed-state entanglement requires a declared measure, and multipartite states require a declared partition and notion of resource.

Numerical routing audit: a partially entangled pair. Let

∣ψ⟩=0.9∣00⟩+0.1∣11⟩.|\psi\rangle=\sqrt{0.9}|00\rangle+\sqrt{0.1}|11\rangle.

Then

ρA=diag⁡(0.9,0.1),E(∣ψ⟩)=h2(0.1)=0.468996 ebits.\rho_A=\operatorname{diag}(0.9,0.1), \qquad E(|\psi\rangle)=h_2(0.1)=0.468996\ \text{ebits}.

The Schmidt rank is two, but the state is not maximally entangled and does not contain one ebit. Route the theorem and reduced spectrum to Composite Systems, selection among entropy, concurrence, negativity, formation, or distillation quantities to Entanglement Measures, and conversion under local operations to the substantive LOCC Preview.

Resources, Free Operations, and Conversion Tasks

Section titled “Resources, Free Operations, and Conversion Tasks”

A resource theory begins by declaring more than a desirable state. At minimum it specifies:

  1. the object type, such as states, channels, measurements, or processes;
  2. free objects;
  3. free operations;
  4. a target task or conversion;
  5. exact or approximate error criteria;
  6. a one-shot, finite-copy, or asymptotic regime;
  7. catalysts, classical communication, postselection, and other side resources.

Only after that contract is fixed can a monotone RR be required to satisfy

R(Λ(ρ))≤R(ρ)for every declared free operation Λ.R(\Lambda(\rho))\le R(\rho) \qquad \text{for every declared free operation }\Lambda.

Different choices of free operations produce different orderings and conversion laws. There is no basis-independent scalar called simply “quantumness” that simultaneously measures entanglement, coherence, magic, asymmetry, and thermodynamic value.

Resource settingCurrent substantive entry point
Bipartite entanglement under local processingEntanglement Measures and LOCC Preview
Magic relative to stabilizer operationsMagic State Distillation
Athermality under thermal operationsThermal Operations Preview
Entanglement as a protocol resourceEntanglement in Quantum Information

Resource Theories owns this comparison rule, while the specialist pages retain their detailed conversion theories. A claim about resource advantage is incomplete until it names the free-operation class, target, accuracy, success probability, copy regime, and all consumed side resources.

No-go results rule out precisely specified tasks under precisely specified assumptions. Their slogans are not interchangeable.

  • Universal exact deterministic cloning of arbitrary unknown pure states is impossible, but known orthogonal states can be copied in their distinguishing basis.
  • No-broadcasting extends the issue to mixed-state families: exact broadcasting is possible for commuting families.
  • No-signaling constrains remote outcome statistics; it is not the same assumption as Bell locality and is not the proof of no-cloning.
  • Bell and Kochen–Specker results concern particular hidden-variable or value-assignment assumptions, not the impossibility of every realist model.

No-Cloning and No-Signaling owns the QI operational proofs and communication consequences. Information and Foundations Results supplies compact assumption-aware theorem cards.

Information–disturbance claims require equal precision. State the input ensemble, instrument, information criterion, disturbance criterion, and whether the claim is average-case, worst-case, or state-dependent. A POVM fixes p(y∣x)p(y|x) but not the postmeasurement channel, so the same POVM can be realized by instruments with different disturbance. There is no universal positive scalar disturbance attached to every informative measurement on every possible input.

Measurement Backaction and Quantum Instruments own the measurement dynamics. Why Generalized Measurements Are Needed owns discrimination-driven motivation. BB84 owns the protocol-specific connection between Eve’s intervention and observed statistics, while Why Quantum Error Correction Is Possible owns the distinction between learnable syndrome information and protected logical information.

Use this owner firewall before expanding a derivation. The local QI page owns the operational use named below; the linked upstream page retains the general theorem or formalism.

If the next question asks about …Use this canonical owner and boundary
A classical source, stochastic channel, code and decoder, error or secrecy criterion, access model, or fair quantum comparatorClassical Information Review owns the operational baseline ledger; Mathematical Toolkit retains the Shannon and Kullback–Leibler derivations, while specialist QI pages retain capacities, security, complexity, and advantage evidence.
Carrier dimension, physical versus logical encoding, leakage, or modesBits, Qubits, Qudits, and Modes owns the QI taxonomy; Mathematical Toolkit owns finite-dimensional Hilbert-space theory.
One-qubit states, measurement axes, tomography, gates, or affine noise geometryBloch Sphere for Quantum Information owns the operational geometry; spin and linear-algebra pages own the geometric derivations.
Ensembles, classical labels, reduced states, purification, instruments, fidelity, or trace distance in a protocolDensity Operators for Quantum Information owns the workflow; Core and Composite Systems own the general formalism.
Von Neumann or Rényi entropy, conditional information, mutual information, relative entropy, data processing, or Holevo quantitiesQuantum Entropy owns the QI use; Math and Composite Systems own classical and structural foundations.
Schmidt theorem, spectrum, rank, or basis freedomSchmidt Decomposition owns the result and proof.
Which entanglement quantity answers a specified taskEntanglement Measures owns task-aware selection; measure-specific derivations remain in Composite Systems.
Free objects, free operations, monotones, or conversion ratesResource Theories owns the general conversion ledger; then use the precise specialist owner, such as LOCC Preview, Magic State Distillation, or Thermal Operations Preview.
POVMs, instruments, channels, or backactionGeneralized Measurements and Instruments and Measurement/Open Systems own the formalism.
No-cloning, no-broadcasting, no-signaling, Bell, or Kochen–Specker assumptionsUse No-Cloning and No-Signaling or the Reference theorem gateway rather than creating a second proof.
A compact formula with conventionsUse Quantum Information Formulas and then follow its canonical explanation link.

The firewall prevents a local convenience calculation from silently becoming a second canonical home. Cross-links should translate the object and task between pages, not repeat a proof with slightly different notation.

Choose a route by the task you need to perform.

Represent and control one carrier. Start with Bits, Qubits, Qudits, and Modes, continue to the Bloch Sphere for calibrated one-qubit work, and exit to Gates, Circuits, and Computation Models when the state and measurement conventions are fixed.

Handle uncertainty, discarded systems, or noise. Use Density Operators for Quantum Information first. Continue to Quantum Entropy when the question is spectral uncertainty, correlation, distinguishability, or a data-processing bound.

Diagnose entanglement. Learn the theorem at Schmidt Decomposition, then use Entanglement Measures to match the state class and operational task to a quantity. Use Entanglement in Quantum Information for protocol context.

Audit a measurement or security claim. Move from Generalized Measurements and Instruments to No-Cloning and No-Signaling or BB84, depending on whether the claim is a formal prohibition or a protocol security statement.

At the end of the sequence, a reader should be able to write the source–state–process–measurement–record chain, name all conventions, calculate a small validity or information check, and identify one canonical owner for the next nontrivial step.

Common Routing and Interpretation Mistakes

Section titled “Common Routing and Interpretation Mistakes”

Treating Hilbert-space dimension as accessible information. Dimension counts independent state-space directions. Accessible information depends on an ensemble, a measurement, and an operational optimization; one measurement shot does not reveal a generic amplitude list.

Assigning a density operator a unique ignorance ensemble. A density operator admits many decompositions. Preserve an accessible preparation label explicitly as a classical-quantum register when its correlations matter.

Using a POVM as a state-update rule. POVM effects determine probabilities. An instrument is required to determine conditional output states and disturbance.

Extending the Bloch ball to arbitrary systems. The three-component Bloch ball is exact for one qubit. Qudits, multiqubit registers, and modes require different state descriptions.

Calling mutual information an entanglement measure. Mutual information counts total correlation and can be nonzero for separable states. Entanglement claims require a declared measure, partition, and task.

Equating Schmidt rank with amount of entanglement. Rank distinguishes product from entangled pure states and counts nonzero Schmidt terms. It does not determine their weights or imply maximal entanglement.

Defining a resource theory only by free states. Free operations, object type, target, error, regime, catalysts, and side resources are also part of the theory. Changing them can change both monotones and possible conversions.

Rejecting negative quantum conditional entropy as an error. Classical conditional entropy is nonnegative, but quantum conditional entropy can be negative. Interpret it through the declared quantum task rather than a classical probability slogan.

Deriving no-cloning from no-signaling. No-cloning follows directly from linearity or inner-product preservation. No-signaling is a distinct constraint on remote observable statistics.

Claiming a universal information–disturbance number. Disturbance depends on the ensemble, instrument, metric, and averaging convention. Even a fixed POVM can have multiple instrument realizations.

Applying finite-dimensional formulas without a domain check. Infinite-dimensional systems require trace-class, support, domain, and often energy-constraint assumptions. A finite truncation is a model choice, not an automatic theorem transfer.

A report says, “This transmon is a qubit that stores a logical qubit.” Separate the physical device, selected code subspace, abstract qubit state, and error-corrected logical encoding. Identify the canonical owner for each layer.

Solution

A transmon is a physical anharmonic circuit with more than two energy levels. Choosing two controlled levels, commonly denoted ∣0⟩|0\rangle and ∣1⟩|1\rangle, defines a two-dimensional computational or code subspace. An abstract qubit state is a normalized ray in that subspace,

∣ψ⟩=α∣0⟩+β∣1⟩,∣α∣2+∣β∣2=1,|\psi\rangle=\alpha|0\rangle+\beta|1\rangle, \qquad |\alpha|^2+|\beta|^2=1,

independent of whether the carrier is a transmon, ion, spin, or photon. Population outside the selected subspace is leakage.

A logical qubit requires an error-correcting encoding, logical operators, syndrome extraction, and a decoder. It is generally distributed across several physical degrees of freedom; calling one unencoded transmon a logical qubit collapses two architecture layers.

Bits, Qubits, Qudits, and Modes owns the carrier and encoding taxonomy. Hardware pages own the circuit, control, coherence, and readout physics. Quantum Error Correction owns the code, syndrome, decoder, and logical-error contract.

For

ρ=(0.80.30.30.2),\rho= \begin{pmatrix} 0.8 & 0.3\\ 0.3 & 0.2 \end{pmatrix},

check physical validity, find the Bloch vector, purity, and the probabilities p(+x)p(+x) and p(0z)p(0z). Explain which retained page owns each part of the analysis.

Solution

The matrix is Hermitian and has trace one. In the computational basis,

ρ=12(1+rzrx−iryrx+iry1−rz),\rho=\frac12 \begin{pmatrix} 1+r_z & r_x-ir_y\\ r_x+ir_y & 1-r_z \end{pmatrix},

so r=(0.6,0,0.6)\mathbf r=(0.6,0,0.6) and ∥r∥=0.72<1\lVert\mathbf r\rVert=\sqrt{0.72}<1. The eigenvalues are

λ±=1±0.722=0.924264, 0.075736,\lambda_\pm=\frac{1\pm\sqrt{0.72}}2 =0.924264,\ 0.075736,

which are nonnegative, so ρ\rho is positive semidefinite and therefore a valid density operator.

The purity is

Tr⁡(ρ2)=1+∥r∥22=0.86.\operatorname{Tr}(\rho^2) =\frac{1+\lVert\mathbf r\rVert^2}{2} =0.86.

For the xx-axis and computational zz-axis measurements,

p(+x)=1+rx2=0.8,p(0z)=1+rz2=0.8.p(+x)=\frac{1+r_x}{2}=0.8, \qquad p(0z)=\frac{1+r_z}{2}=0.8.

Density Operators for Quantum Information owns positivity, normalization, purity, and ensemble interpretation. Bloch Sphere for Quantum Information owns the vector and measurement-axis geometry. Quantum Entropy owns the spectral entropy if that is the next question.

For ∣Φ+⟩=(∣00⟩+∣11⟩)/2|\Phi^+\rangle=(|00\rangle+|11\rangle)/\sqrt2, compute the reduced state of subsystem AA. Explain why the result does not select a unique ignorance ensemble.

Solution

The joint density operator is

ρAB=12(∣00⟩⟨00∣+∣00⟩⟨11∣+∣11⟩⟨00∣+∣11⟩⟨11∣).\rho_{AB}=\frac12\left( |00\rangle\langle00| +|00\rangle\langle11| +|11\rangle\langle00| +|11\rangle\langle11| \right).

Taking the partial trace over BB removes the cross terms because ⟨1∣0⟩=⟨0∣1⟩=0\langle1|0\rangle=\langle0|1\rangle=0:

ρA=Tr⁡BρAB=12(∣0⟩⟨0∣+∣1⟩⟨1∣)=I2.\rho_A=\operatorname{Tr}_B\rho_{AB} =\frac12\left(|0\rangle\langle0|+|1\rangle\langle1|\right) =\frac I2.

The global state remains pure because ρAB2=ρAB\rho_{AB}^2=\rho_{AB}, while the local state has purity Tr⁡(ρA2)=1/2\operatorname{Tr}(\rho_A^2)=1/2. The same I/2I/2 can be decomposed as an equal mixture of zz-basis states, xx-basis states, or infinitely many other ensembles. Without an accessible classical label, none is the unique physical ignorance story.

Composite Systems owns partial trace and entanglement structure. Density Operators for Quantum Information owns the protocol-level distinction among labeled preparations, discarded subsystems, and operationally equivalent local states.

Let a classical bit have distribution p=(0.9,0.1)p=(0.9,0.1) and encode its values in orthogonal states, giving ρ=diag⁡(0.9,0.1)\rho=\operatorname{diag}(0.9,0.1). Compute H(X)H(X) and S(ρ)S(\rho), and state the scope of their equality.

Solution

With base-22 logarithms,

H(X)=−0.9log⁡2(0.9)−0.1log⁡2(0.1)=h2(0.1)=0.468996 bits.\begin{aligned} H(X) &=-0.9\log_2(0.9)-0.1\log_2(0.1)\\ &=h_2(0.1)=0.468996\ \text{bits}. \end{aligned}

The eigenvalues of the diagonal density operator are also 0.90.9 and 0.10.1, so

S(ρ)=h2(0.1)=0.468996 bits.S(\rho)=h_2(0.1)=0.468996\ \text{bits}.

The equality holds because the classical probabilities are the eigenvalues of an orthogonal encoding. For a nonorthogonal ensemble, the label entropy H(X)H(X) and the entropy of the average state S(ρˉ)S(\bar\rho) need not agree; moreover, accessible information depends on an optimized measurement and is bounded by the Holevo quantity. Mathematical Toolkit Entropy owns the classical calculation, while Quantum Entropy owns the spectral and operational quantum distinctions.

Compute I(A:B)I(A{:}B) for the pure product state ∣00⟩|00\rangle and for ∣Φ+⟩|\Phi^+\rangle. Compare each result with the entropy of entanglement.

Solution

For ∣00⟩|00\rangle, the global and reduced states are pure, so

S(A)=S(B)=S(AB)=0,I(A:B)=0.S(A)=S(B)=S(AB)=0, \qquad I(A{:}B)=0.

Its entropy of entanglement is also zero.

For ∣Φ+⟩|\Phi^+\rangle, the global state is pure but both reduced states are maximally mixed:

S(AB)=0,S(A)=S(B)=1 bit.S(AB)=0, \qquad S(A)=S(B)=1\ \text{bit}.

Therefore

I(A:B)=S(A)+S(B)−S(AB)=2 bits.I(A{:}B)=S(A)+S(B)-S(AB)=2\ \text{bits}.

The entropy of entanglement is S(A)=1S(A)=1 ebit, not two ebits. Mutual information measures total correlation and counts both subsystem entropy contributions in this identity; it is not numerically the same object as the pure-state entanglement resource.

Quantum Entropy owns the identity, Mutual Information owns structural interpretation, and Entanglement Measures owns measure selection.

For ∣ψ⟩=0.9∣00⟩+0.1∣11⟩|\psi\rangle=\sqrt{0.9}|00\rangle+\sqrt{0.1}|11\rangle, find the Schmidt rank and entropy of entanglement. Explain why rank, entropy, concurrence, negativity, and conversion rate are different answers.

Solution

The state is already in Schmidt form, with nonzero Schmidt probabilities 0.90.9 and 0.10.1. Hence its Schmidt rank is two and

E(∣ψ⟩)=h2(0.1)=0.468996 ebits.E(|\psi\rangle)=h_2(0.1)=0.468996\ \text{ebits}.

Rank records only how many Schmidt coefficients are nonzero. Entropy also uses their weights. For this two-qubit pure state, concurrence is

C=20.9×0.1=0.6,C=2\sqrt{0.9\times0.1}=0.6,

while negativity is 0.9×0.1=0.3\sqrt{0.9\times0.1}=0.3. These numbers use different normalizations and answer different mathematical or operational questions. In the asymptotic pure bipartite LOCC setting, the entropy gives the reversible conversion rate in ebits per copy; one-shot, approximate, mixed-state, or restricted-operation tasks require other quantities and error conventions.

Schmidt Decomposition owns the theorem and spectrum. Entanglement Measures owns the decision about which quantity matches the state class and task.

Audit the statement “a TT state contains quantumness.” List the information required to turn it into a meaningful stabilizer-resource claim.

Solution

First declare the object: for example, a single-qubit state intended for non-Clifford gate injection. Next choose the free states, commonly the stabilizer-state polytope, and the free operations, such as a specified class of stabilizer operations with Clifford gates, Pauli measurements, classical control, and explicitly declared postselection rules.

Then declare the target task: preparing a higher-fidelity magic state, injecting a TT gate, or supplying a fault-tolerant factory. Choose a monotone appropriate to that free-operation class, such as robustness-type magic or mana where applicable. State whether conversion is exact or approximate, deterministic or probabilistic, one-shot or asymptotic, and whether catalysts are returned unchanged.

Finally count consumed stabilizer ancillas, Clifford operations, measurements, classical randomness, discarded failure branches, and any correlated-input assumptions. Only with that tuple can “magic” be a resource claim. A task- and basis-independent scalar “quantumness” is not defined by the sentence.

Magic State Distillation owns the concrete protocol and factory implications. The general lesson is that free objects alone do not define a resource theory.

8. BB84 cloning, disturbance, and no-signaling audit

Section titled “8. BB84 cloning, disturbance, and no-signaling audit”

Evaluate the claim: “BB84 is secure because every informative measurement causes a fixed disturbance, and cloning is impossible only because it would signal.” Identify the correct owners of each corrected statement.

Solution

BB84 uses four nonorthogonal signal states in two conjugate bases. Eve must be modeled by an instrument acting on that ensemble, together with any retained quantum system and classical record. The observable disturbance is inferred from declared parameter-estimation statistics. It is not a universal constant attached to every informative measurement.

For the intercept–resend example in which Eve chooses a BB84 basis uniformly, she chooses the wrong basis half the time. Conditional on that event and on Alice and Bob retaining a same-basis round, Bob receives the wrong bit with probability one half. The sifted-key error rate is therefore 12×12=14\tfrac12\times\tfrac12=\tfrac14. Other attacks produce different information–disturbance relations, and a complete security proof also requires authentication, finite statistics, error correction, privacy amplification, and device assumptions.

No-cloning does not require no-signaling. If a universal unitary cloner satisfied

U∣ψ⟩∣0⟩=∣ψ⟩∣ψ⟩U|\psi\rangle|0\rangle=|\psi\rangle|\psi\rangle

for every input, inner-product preservation for ∣ψ⟩|\psi\rangle and ∣ϕ⟩|\phi\rangle would require

⟨ϕ∣ψ⟩=⟨ϕ∣ψ⟩2,\langle\phi|\psi\rangle =\langle\phi|\psi\rangle^2,

which fails for distinct nonorthogonal states. No-signaling is instead a constraint that one party’s local choice cannot change another party’s unconditional local statistics.

BB84 and its QKD parent own the protocol and security conditions. Measurement Backaction and Quantum Instruments own disturbance modeling. No-Cloning and No-Signaling owns the operational proofs, and the Reference theorem cards own compact assumption checks.

  • E. Chitambar and G. Gour, “Quantum Resource Theories,” Reviews of Modern Physics 91, 025001 (2019), doi:10.1103/RevModPhys.91.025001.
  • T. M. Cover and J. A. Thomas, Elements of Information Theory, 2nd ed., Wiley (2006).
  • R. Horodecki, P. Horodecki, M. Horodecki, and K. Horodecki, “Quantum Entanglement,” Reviews of Modern Physics 81, 865–942 (2009), doi:10.1103/RevModPhys.81.865.
  • M. A. Nielsen and I. L. Chuang, Quantum Computation and Quantum Information, 10th anniversary ed., Cambridge University Press (2010), doi:10.1017/CBO9780511976667.
  • C. E. Shannon, “A Mathematical Theory of Communication,” Bell System Technical Journal 27, 379–423 and 623–656 (1948), doi:10.1002/j.1538-7305.1948.tb01338.x.
  • J. Watrous, The Theory of Quantum Information, Cambridge University Press (2018), doi:10.1017/9781316848142.
  • M. M. Wilde, Quantum Information Theory, 2nd ed., Cambridge University Press (2017), doi:10.1017/9781316809976.