Assumptions and Scope

The standard postulates do not describe a physical problem until a system, Hamiltonian, state space, measurement model, and regime of validity have been chosen. They specify how to calculate predictions from a model, while the physical system determines which model is appropriate.

The default package used in Core Formalism is ordinary nonrelativistic quantum mechanics with a specified Hilbert space, external time parameter, fixed degrees of freedom, unitary dynamics when the retained system can be treated as closed, and a stated measurement model. Ideal projective measurements are a useful first example; the minimal postulates also include general effects and instruments. Each phrase is an assumption that can be relaxed, but relaxing it requires additional structure.

This page identifies those assumptions and gives diagnostics for deciding when the compact postulates are sufficient. Interpretive questions that remain after a model is specified belong to What the Postulates Do Not Say.

Helpful background. Minimal postulates supplies the general rules; finite versus infinite dimension explains why domains and spectra need separate care.

Physical regimes and model assumptions

Question	Default model	Standard extension when the default fails
Kinematics	nonrelativistic degrees of freedom; external time	relativistic effective theory or quantum field theory
Particle content	fixed one-particle or fixed-$N$ Hilbert space	Fock space, number-sector dynamics, or fields
System boundary	closed system with unitary evolution	channels, master equations, stochastic dynamics, or a larger system–environment model
Measurement	sharp projective measurement with ideal update	POVMs, instruments, detector models, and continuous measurement
Particle labels	distinguishable subsystems unless stated	symmetric or antisymmetric sectors for identical particles
Mathematical setting	finite matrices or well-controlled Hilbert-space operators	domain analysis, spectral measures, rigged spaces, and operator algebras
Model fidelity	stated Hamiltonian taken as the model	controlled effective Hamiltonian, truncation, or approximation with an error budget

An extension is not automatically a correction to the postulates. It may change the system boundary, enrich the state and operation language, or replace the model with a more accurate effective description while retaining the same probability rule.

What the Postulates Assume and What Models Supply

It is useful to separate three layers.

1. Structural postulates specify states, measurements, probabilities, composition, and allowed dynamics.
2. System models specify the Hilbert space, Hamiltonian, observables, couplings, boundary conditions, and preparation procedure.
3. Regime claims specify why that model is accurate for the energies, times, lengths, controls, and measurements of interest.

For example, the Schrödinger equation

$$i\hbar \frac{d}{dt} \lvert\psi(t)\rangle = H\lvert\psi(t)\rangle$$

is structural once $H$ is given. It does not derive the Coulomb potential, identify which molecular coordinates may be frozen, prove that a two-level truncation is valid, or decide whether an environment can be ignored.

A prediction is therefore conditional:

$$\text{postulates}+\text{model} {}+\text{initial data}+\text{approximations} \longrightarrow\text{prediction}.$$

Agreement with experiment tests this package. A discrepancy may expose a calculation error, an incorrect parameter, a failed approximation, a poor measurement model, or physics beyond the assumed regime. It does not identify the failing layer without further analysis.

Nonrelativistic Setting

The default setting treats time as an external parameter and uses nonrelativistic degrees of freedom. For $N$ distinguishable particles, a typical Hamiltonian has the form

$$H = \sum_{i=1}^N \frac{\mathbf p_i^2}{2m_i} +V(\mathbf x_1,\ldots,\mathbf x_N,t),$$

possibly supplemented by spin, external electromagnetic fields, or effective interactions.

The approximation can be motivated by expanding the relativistic energy:

$$\begin{aligned} E(\mathbf p) &= \sqrt{ m^2c^4+\mathbf p^2c^2 }\\ &= mc^2 +\frac{\mathbf p^2}{2m} -\frac{\mathbf p^4}{8m^3c^2} +\cdots. \end{aligned}$$

After subtracting the constant rest energy, the Schrödinger kinetic term is the leading contribution when

$$\frac{\lvert\mathbf p\rvert}{mc} \ll 1.$$

For a bound system, a related diagnostic is

$$\frac{\lvert E_{\mathrm{bind}}\rvert}{mc^2} \ll 1.$$

These inequalities are useful, but they are not a universal theorem that every correction is negligible. Precision spectroscopy may require relativistic, radiative, recoil, finite-size, or hyperfine corrections even when the leading nonrelativistic model is excellent.

What remains inside nonrelativistic quantum mechanics

The nonrelativistic framework can include:

• spin through finite-dimensional internal spaces;
• Pauli Hamiltonians and controlled relativistic corrections;
• many interacting particles;
• time-dependent external controls;
• nonrelativistic Fock space and second quantization;
• open-system models and generalized measurements;
• effective lattice, band, molecular, and few-level descriptions.

None of these automatically turns the model into relativistic quantum field theory. The decisive question is which degrees of freedom and symmetries are structural.

Warning signs that the default is insufficient

The fixed-particle nonrelativistic model needs reconsideration when:

• velocities or momentum transfers are not small compared with relativistic scales;
• particle creation and annihilation are kinematically available;
• antiparticles are required;
• relativistic locality and causality constrain observables;
• electromagnetic radiation must be quantized rather than treated as an external field;
• ultraviolet behavior or vacuum fluctuations are part of the question.

Relativistic one-particle equations remain valuable bridge models, but they do not by themselves provide the full many-particle relativistic framework.

Fixed Particle Number, Unless Stated Otherwise

For one distinguishable particle, the state space may be a one-particle Hilbert space $\mathcal h$. For $N$ distinguishable particles with fixed particle number,

$$\mathcal H_N = \mathcal h_1 \otimes\cdots\otimes \mathcal h_N.$$

The number $N$ is then part of the model definition, not a random observable inside that Hilbert space.

For identical particles built from one-particle space $\mathcal h$, the physical fixed-$N$ space is a symmetric or antisymmetric sector rather than the full slot-labeled tensor product:

$$\mathcal H_N^{(\pm)} = S_\pm \mathcal h^{\otimes N}.$$

The exchange-symmetry qualification is developed below.

Fock-space extension

When particle number can vary in nonrelativistic many-body physics, the natural state space is Fock space:

$$\mathcal F_\pm(\mathcal h) = \bigoplus_{N=0}^{\infty} \mathcal H_N^{(\pm)}.$$

The plus sign denotes bosonic sectors and the minus sign fermionic sectors. The vacuum is the $N=0$ sector. Creation and annihilation operators connect neighboring sectors.

If

$$[H,\widehat N]=0,$$

each number sector evolves independently and a fixed-$N$ treatment may be chosen consistently. For unbounded operators, the precise condition is that the Hamiltonian evolution preserve every number-sector projection; a formal zero commutator alone requires domain justification. If the commutator is nonzero, the Hamiltonian couples number sectors. Even then, a nonrelativistic Fock-space model can be appropriate, for example in effective quasiparticle or many-body descriptions.

Particle-number change is therefore not by itself a sufficient definition of relativistic QFT. The stronger field-theoretic need appears when relativity, local quantum fields, antiparticles, and particle creation are structural rather than effective. Fock Space explains the nonrelativistic construction, and Relationship to Quantum Field Theory explains the transition to fields.

Superselection and coherent number sectors

Writing a direct sum does not guarantee that every relative phase between number sectors is operationally accessible. Charge or particle-number superselection constraints can restrict the allowed preparations and observables. Whether a coherent superposition of sectors is meaningful depends on the conserved quantity, available reference frames, and physical model. This is a scope question, not a consequence of the direct-sum notation alone.

Closed vs Open Systems

A closed system is one whose retained state is modeled as evolving unitarily:

$$\rho(t) = U(t,t_0) \rho(t_0) U(t,t_0)^\dagger.$$

Unitary conjugation preserves trace, positivity, eigenvalues, entropy, and purity. It is reversible when $U$ is known.

Closedness is relative to a declared system boundary. A laboratory system may interact weakly with air, control electronics, radiation, and substrates, yet be modeled as closed over a short time and within a chosen accuracy. The same system may require an open description over longer times or at higher precision.

Reduced dynamics from a larger closed system

Let $S$ be the system of interest and $E$ an environment. The composite state may evolve unitarily:

$$\rho_{SE}(t) = U_{SE}(t) \rho_{SE}(0) U_{SE}(t)^\dagger.$$

The reduced state is

$$\rho_S(t) = \operatorname{Tr}_E \rho_{SE}(t).$$

If the initial state is prepared as a fixed product

$$\rho_{SE}(0) = \rho_S(0)\otimes\eta_E,$$

then the reduced evolution defines a quantum channel:

$$\mathcal E_t(\rho_S) = \operatorname{Tr}_E \left[ U_{SE}(t) \left( \rho_S\otimes\eta_E \right) U_{SE}(t)^\dagger \right].$$

The channel is completely positive and trace preserving. It need not be unitary, reversible, or entropy preserving on $S$ because correlations with $E$ are discarded.

Initial system–environment correlations require extra care. A single state-independent channel on every hypothetical input state may not exist without specifying an assignment procedure or restricted compatibility domain. The formula above quietly assumes a fixed environment state independent of the input $\rho_S$.

Open does not mean Markovian

An open-system model may have memory. A Markovian master equation is an additional approximation, usually involving assumptions about coupling strength, environmental correlation time, initial correlations, and secular or coarse-graining steps. The existence of an environment alone does not justify a Lindblad equation.

Use Open Quantum Systems for reduced dynamics, Quantum Channels and Noise for the map language, and the Approximation Checklist before accepting a master-equation reduction.

Idealized Measurements

The first measurement model is a sharp projective measurement. Projectors $\{P_a\}$ satisfy

$$P_aP_b = \delta_{ab}P_a, \qquad \sum_aP_a = I.$$

The probability of outcome $a$ is

$$p(a) = \operatorname{Tr}(\rho P_a).$$

For the ideal Lüders instrument, the unnormalized conditional branch is

$$\widetilde\rho_a = P_a\rho P_a,$$

and, when $p(a)>0$, the normalized conditional state is

$$\rho_a = \frac{ P_a\rho P_a }{ \operatorname{Tr}(\rho P_a) }.$$

This model is repeatable for a sharp outcome and preserves coherence inside a degenerate eigenspace. It is not the unique physical implementation of every measurement with the same projectors.

Statistics and disturbance are separate data

A general measurement has effects $\{E_a\}$ satisfying

$$E_a\geq0, \qquad \sum_aE_a=I,$$

with probabilities

$$p(a) = \operatorname{Tr}(\rho E_a).$$

The effects do not determine the output state. A quantum instrument assigns a linear, completely positive, trace-nonincreasing map $\mathcal I_a$ to each outcome; the sum over all outcomes must be trace preserving:

$$\widetilde\rho_a = \mathcal I_a(\rho), \qquad p(a) = \operatorname{Tr}\widetilde\rho_a.$$

Different instruments can have the same POVM and therefore the same outcome probabilities while disturbing the state differently.

Real devices may be inefficient, biased, noisy, weak, destructive, coarse-grained, continuously monitored, or coupled to uncontrolled modes. Detector response, calibration uncertainty, dark counts, finite resolution, and classical readout errors belong to the measurement model. The projective postulate does not certify that an apparatus realizes an ideal projector.

Generalized Measurements and Instruments develops this extension. The compact Core-level starting points are Projective Measurement and POVMs: First Encounter.

Conditional update is not free evolution

The map $\rho\mapsto\rho_a$ is nonlinear because of normalization and is conditioned on classical information. The unnormalized map $\rho\mapsto\widetilde\rho_a$ is linear and trace nonincreasing. Neither should be confused with the unconditional unitary evolution of a closed system.

An explicit apparatus model can place system, pointer, and environment inside a larger unitary description. Reading or ignoring the resulting record then produces conditional or unconditional reduced states. This connects the postulate-level rule to physical modeling without making every detector literally instantaneous.

Identical Particles and Symmetrization Caveat

The elementary composition rule

$$\mathcal H_{AB} = \mathcal H_A\otimes\mathcal H_B$$

is directly interpreted as two labeled subsystems when $A$ and $B$ are distinguishable. For identical particles, tensor slots are mathematical labels, not observable individual names.

For $N$ ordinary identical particles, a permutation $\pi$ acts through a unitary $P_\pi$. Physical bosonic and fermionic states satisfy

$$P_\pi\lvert\Psi\rangle = \begin{cases} \lvert\Psi\rangle, &\text{bosons},\\ \operatorname{sgn}(\pi) \lvert\Psi\rangle, &\text{fermions}. \end{cases}$$

Physical observables preserve the exchange sector:

$$[A,P_\pi] = 0.$$

Spin and spatial variables must be exchanged together. An antisymmetric fermionic state can have a symmetric spatial part only when another part, such as spin, supplies the compensating antisymmetry.

The distinction is stronger than ignorance about labels. There is no hidden observable that says which electron is permanently “particle 1.” Particle entanglement language must therefore be handled carefully; modes, spatial regions, or operationally addressable degrees of freedom may provide the physically meaningful subsystem decomposition.

The standard boson–fermion symmetrization rule is the default for ordinary identical particles in three spatial dimensions. Lower-dimensional systems can support anyonic exchange structure, and more general field-theoretic statistics require additional assumptions. These cases are extensions, not exceptions that can be represented by silently dropping exchange symmetry.

Continue with Indistinguishability and the Symmetrization Postulate.

Relativistic and Field-Theoretic Limits

Nonrelativistic quantum mechanics can include relativistic corrections and can use relativistic wave equations as bridge models. It does not, by itself, provide the complete framework demanded by relativistic locality and variable particle content.

Several transitions should be kept distinct:

Situation	Appropriate response
small relativistic correction to a bound state	add a controlled effective operator
spin-$1/2$ particle in an external field at moderate energy	Pauli or effective Dirac-based Hamiltonian
relativistic one-particle kinematics used pedagogically	state the positive-frequency and interaction limitations
pair creation, annihilation, or antiparticles are dynamical	use quantum fields
local relativistic observables and causal propagation are structural	use quantum field theory
dynamical electromagnetic radiation and radiative corrections matter	quantize the relevant field modes or use QED

The boundary is not determined by whether an equation contains $c$. A Hamiltonian with a relativistic correction can remain an effective nonrelativistic model, while a low-energy many-body theory may use field operators without being a relativistic fundamental theory.

Relativistic quantum field theory changes the organizing objects. Fields, local algebras, vacuum structure, particle sectors, renormalization, and causal spacetime relations become central. Relationship to Quantum Field Theory explains these changes.

Effective Models and Approximations

Many of the most successful quantum models are deliberately incomplete. A two-level atom, Born–Oppenheimer Hamiltonian, tight-binding band, spin chain, pseudopotential, rotating-wave model, and low-energy scattering theory all discard degrees of freedom or scales.

The postulates compute predictions inside the chosen model. They do not prove that the model represents the target experiment.

Projection and omitted-sector feedback

Let $P$ project onto retained states and $Q=I-P$ onto discarded states. The naive truncation is

$$H_{\mathrm{trunc}} = P H P.$$

It omits virtual excursions into the $Q$ sector. In a finite block model, eliminate the $Q$ component of $H\psi=E\psi$ when $E-QHQ$ is invertible on $Q\mathcal H$. The remaining equation uses the exact energy-dependent operator

$$H_{\mathrm{eff}}(E) = P H P + P H Q \frac{1}{ E-Q H Q } Q H P.$$

The inverse acts only on the discarded subspace. Near an eigenvalue of $QHQ$ it can become large, so omitting that subspace can fail even for a small coupling. In infinite dimensions, domains and resolvent existence need additional care. Perturbative reductions expand this expression under stated gap and coupling assumptions.

Three validity questions

Every effective model should answer:

1. Kinematic validity: Are the retained states and degrees of freedom sufficient over the energy and momentum range being probed?
2. Dynamical validity: Are discarded couplings small over the relevant evolution time, including resonant and cumulative effects?
3. Observational validity: Do the modeled observables correspond to what the apparatus resolves, and are omitted variables genuinely inaccessible?

A small instantaneous correction can accumulate over long times. A weak coupling can dominate near resonance. A low-energy truncation can fail under strong driving even when the undriven state remains low energy.

Control parameters and error estimates

An approximation should be organized by dimensionless quantities, for example

$$\epsilon_{\mathrm{rel}} = \frac{p}{mc}, \qquad \epsilon_{\mathrm{leak}} = \frac{\hbar|g|}{|\Delta_{\mathrm{det}}|},$$

where $g$ is the actual drive-induced coupling rate to an omitted level and $\Delta_{\mathrm{det}}$ is its energy detuning from the relevant drive-assisted transition. A large undriven energy separation is insufficient if the drive is resonant with that transition. The exact control parameter depends on the model. A credible validity statement names the parameter, the observable, the time window, and the expected order of the neglected correction.

Numerical convergence is another check. Increasing basis size, grid resolution, cutoff, or environment dimension should stabilize the quantities being reported. Convergence of one low-energy eigenvalue does not prove that all dynamics or observables have converged.

Mathematical Idealizations

The compact postulates are easiest in finite dimension. Infinite-dimensional quantum mechanics adds analytic assumptions that are often suppressed in introductory notation.

Domains and self-adjointness

Position, momentum, and most Hamiltonians are unbounded. A differential expression such as

$$p = -i\hbar\frac{d}{dx}$$

does not define an observable until its domain and boundary conditions are specified. Symmetry under integration by parts is not always enough; self-adjointness controls real spectra and unitary evolution.

Different self-adjoint boundary conditions can define different physics. A particle on an interval with periodic boundary conditions is not the same model as a particle with hard-wall boundary conditions, even if the interior differential expression is identical.

Generalized eigenstates and continuous spectra

Position kets, momentum kets, and plane waves are usually generalized eigenvectors rather than normalizable Hilbert-space states. Equations such as

$$\langle x\mid x'\rangle = \delta(x-x')$$

belong to a distributional calculus. Physical wave packets remain square-integrable. Probability statements for continuous observables are measures on intervals, not probabilities assigned to an exact point by $\lvert\psi(x)\rvert^2$ alone.

Separability and superselection

Most ordinary models use separable Hilbert spaces, allowing countable orthonormal bases. Infinite systems and algebraic formulations can require more careful representation theory. Superselection rules can also prevent coherent observables between sectors even when a formal direct sum exists.

Finite vs Infinite-Dimensional Quantum Mechanics provides the orientation, and Hermitian vs Self-Adjoint Operators explains the domain distinction.

Worked Example: One Interaction, Three Descriptions

Let a normalized qubit $S$ interact with an initially uncorrelated environment $E$ in a fixed normalized state $|e\rangle$. The normalized states $|e_0\rangle$ and $|e_1\rangle$ need not be orthogonal. Suppose the joint unitary maps

$$\begin{aligned} \lvert0\rangle\lvert e\rangle &\longmapsto \lvert0\rangle\lvert e_0\rangle,\\ \lvert1\rangle\lvert e\rangle &\longmapsto \lvert1\rangle\lvert e_1\rangle. \end{aligned}$$

For the initial qubit state

$$\lvert\psi\rangle = \alpha\lvert0\rangle +\beta\lvert1\rangle,$$

the joint output is

$$\lvert\Psi_{SE}\rangle = \alpha \lvert0\rangle\lvert e_0\rangle + \beta \lvert1\rangle\lvert e_1\rangle.$$

The composite description is closed and pure. Define

$$\gamma = \langle e_1\mid e_0\rangle.$$

After ignoring the environment,

$$\rho_S = \begin{pmatrix} \lvert\alpha\rvert^2 & \alpha\beta^*\gamma \\ \alpha^*\beta\gamma^* & \lvert\beta\rvert^2 \end{pmatrix}.$$

Three modeling choices now give three descriptions:

1. Retain $E$. The joint state evolves unitarily and no information is discarded.
2. Ignore $E$. The qubit undergoes a dephasing channel. When $\lvert\gamma\rvert<1$, its coherence is reduced.
3. Measure $E$. An observed environmental record defines conditional qubit branches through an instrument.

If the environment states agree up to phase, $|\gamma|=1$ and the qubit remains pure, possibly with a changed relative phase: the environment acquired no distinguishing information. In particular, equal environmental kets give $\gamma=1$. If the two environment states are orthogonal, then $\gamma=0$ and the qubit is fully dephased in the computational basis.

The physical interaction can be the same in all three cases. What changes is what is retained, ignored, or read out. This is why “closed,” “open,” and “measured” are statements about a system boundary and information flow, not three unrelated dynamical laws.

Exercises

1. Expand the relativistic energy $E=\sqrt{m^2c^4+p^2c^2}$ through order $p^4$ and identify the relative size of the first correction to the nonrelativistic kinetic energy.

Solution

Write

$$E = mc^2 \sqrt{ 1+\frac{p^2}{m^2c^2} }.$$

Using

$$\sqrt{1+x} = 1+\frac{x}{2}-\frac{x^2}{8}+O(x^3),$$

gives

$$E = mc^2+\frac{p^2}{2m} - \frac{p^4}{8m^3c^2} + O\!\left( \frac{p^6}{m^5c^4} \right).$$

Relative to $p^2/(2m)$, the magnitude of the first correction is

$$\frac{ p^4/(8m^3c^2) }{ p^2/(2m) } = \frac14 \left( \frac{p}{mc} \right)^2.$$

Thus $p/(mc)$ is the natural small parameter, while the precision target determines whether the correction may be ignored.

2. Explain the difference among a fixed-$N$ Hilbert space, a Fock space with a number-conserving Hamiltonian, and a Fock space with number-changing interactions.

Solution

A fixed-$N$ Hilbert space contains only one particle-number sector. The value $N$ is built into the model.

Fock space is a direct sum of all number sectors. If $[H,\widehat N]=0$, each sector is invariant, so a state prepared in one sector remains there even though the larger space is available.

If $[H,\widehat N]\neq0$, the Hamiltonian connects sectors and particle number can change. This may describe a nonrelativistic effective process or a field-theoretic process depending on the degrees of freedom, symmetries, and regime. Number change alone does not settle that classification.

3. Derive the reduced qubit state in the worked system–environment example.

Solution

Expand the joint density operator:

$$\rho_{SE} ={} \lvert\alpha\rvert^2 \lvert0\rangle\langle0\rvert \otimes \lvert e_0\rangle\langle e_0\rvert + \alpha\beta^* \lvert0\rangle\langle1\rvert \otimes \lvert e_0\rangle\langle e_1\rvert + \alpha^*\beta \lvert1\rangle\langle0\rvert \otimes \lvert e_1\rangle\langle e_0\rvert + \lvert\beta\rvert^2 \lvert1\rangle\langle1\rvert \otimes \lvert e_1\rangle\langle e_1\rvert.$$

Taking the environmental trace uses

$$\operatorname{Tr}_E \left( \lvert e_i\rangle\langle e_j\rvert \right) = \langle e_j\mid e_i\rangle.$$

With $\gamma=\langle e_1\mid e_0\rangle$, this gives

$$\rho_S = \begin{pmatrix} \lvert\alpha\rvert^2 & \alpha\beta^*\gamma \\ \alpha^*\beta\gamma^* & \lvert\beta\rvert^2 \end{pmatrix}.$$

4. Suppose the environment begins in $\lvert e\rangle$, evolves jointly by $U$, and is traced out. In an orthonormal environment basis $\{\lvert k\rangle\}$, define $K_k=\langle k\mid U\mid e\rangle$. Show that the reduced map has a Kraus representation and is trace preserving.

Solution

Insert the environment basis into the partial trace:

$$\begin{aligned} \mathcal E(\rho) &= \sum_k \langle k\mid U \left( \rho\otimes \lvert e\rangle\langle e\rvert \right) U^\dagger \mid k\rangle\\ &= \sum_k K_k\rho K_k^\dagger. \end{aligned}$$

Unitarity gives

$$\begin{aligned} \sum_kK_k^\dagger K_k &= \sum_k \langle e\mid U^\dagger\mid k\rangle \langle k\mid U\mid e\rangle\\ &= \langle e\mid U^\dagger U\mid e\rangle\\ &= I_S. \end{aligned}$$

Therefore

$$\operatorname{Tr}\mathcal E(\rho) = \operatorname{Tr}\rho.$$

5. A noisy two-outcome detector intended to measure $Z$ has effects

$$E_+ = (1-\epsilon) \lvert0\rangle\langle0\rvert + \epsilon \lvert1\rangle\langle1\rvert,$$

and $E_-=I-E_+$, with $0\leq\epsilon\leq1/2$. Verify that this is a POVM and find the probability of outcome $+$ for $\rho=(I+\mathbf r\cdot\boldsymbol\sigma)/2$.

Solution

The eigenvalues of $E_+$ are $1-\epsilon$ and $\epsilon$, both in $[0,1]$. Therefore $E_+\geq0$ and $E_-=I-E_+\geq0$. Their sum is $I$, so they form a POVM.

Using

$$\rho_{00} = \frac{1+r_z}{2}, \qquad \rho_{11} = \frac{1-r_z}{2},$$

the probability is

$$\begin{aligned} p(+) &= \operatorname{Tr}(\rho E_+)\\ &= (1-\epsilon)\rho_{00} +\epsilon\rho_{11}\\ &= \frac12 \left[ 1+(1-2\epsilon)r_z \right]. \end{aligned}$$

Noise reduces the detector contrast by the factor $1-2\epsilon$. The effects specify the statistics but not the post-measurement state.

6. Let $P$ project onto a degenerate eigenspace. Show that the Lüders nonselective map preserves coherences inside that subspace but removes coherences between different outcome subspaces.

Solution

For a projective decomposition $\{P_a\}$, the nonselective map is

$$\mathcal L(\rho) = \sum_aP_a\rho P_a.$$

Decompose

$$\rho = \sum_{a,b} P_a\rho P_b.$$

The map keeps only terms with $a=b$:

$$\mathcal L(\rho) = \sum_aP_a\rho P_a.$$

Within one degenerate subspace $P_a\mathcal H$, the full block $P_a\rho P_a$ remains, including off-diagonal matrix elements between different vectors in that subspace. Coherences $P_a\rho P_b$ with $a\neq b$ are removed.

7. Two identical fermions occupy one-particle states $\lvert u\rangle$ and $\lvert v\rangle$. Construct the normalized antisymmetric state when the orbitals are orthonormal, and show why it vanishes if $\lvert v\rangle=\lvert u\rangle$.

Solution

For orthonormal orbitals, the normalized antisymmetric state is

$$\lvert\Psi_-\rangle = \frac{1}{\sqrt2} \left( \lvert u\rangle_1\lvert v\rangle_2 - \lvert v\rangle_1\lvert u\rangle_2 \right).$$

The two product terms are orthogonal, so the norm is one. If $\lvert v\rangle=\lvert u\rangle$, then

$$\lvert\Psi_-\rangle = \frac{1}{\sqrt2} \left( \lvert u\rangle_1\lvert u\rangle_2 - \lvert u\rangle_1\lvert u\rangle_2 \right) = 0.$$

No antisymmetric two-fermion state can place both fermions in the same complete one-particle state. This is the state-space origin of Pauli exclusion.

8. A driven atom is modeled as two levels with Rabi frequency $\Omega$. The nearest omitted level is separated from the driven manifold by energy $\Delta_{\mathrm{leak}}$. Explain which conditions must be checked before trusting long-time population predictions. Distinguish the undriven energy gap from the detuning of a drive-assisted transition.

Solution

The checks should include:

• the actual coupling to the omitted level and its detuning from each drive component; the intended Rabi frequency $\Omega$ and the undriven gap alone do not determine leakage;
• whether multiphoton or near-resonant processes couple to omitted levels;
• whether the rotating-wave or other time-dependent approximation is valid;
• whether spontaneous emission, dephasing, or control noise matters over the requested time;
• whether small leakage or phase errors accumulate over many cycles;
• whether the measured observable is confined to the two-level manifold;
• whether increasing the retained basis changes the predicted population.

A small instantaneous leakage amplitude does not guarantee accurate long-time phase or population dynamics. The claimed validity must name the time window and observable.

References

• L. E. Ballentine, Quantum Mechanics: A Modern Development, 2nd ed., World Scientific, 2014.
• H.-P. Breuer and F. Petruccione, The Theory of Open Quantum Systems, Oxford University Press, 2002.
• P. Busch, P. Lahti, J.-P. Pellonpää, and K. Ylinen, Quantum Measurement, Springer, 2016.
• P. A. M. Dirac, The Principles of Quantum Mechanics, 4th ed., Oxford University Press, 1958.
• A. Messiah, Quantum Mechanics, Dover, 1999.
• M. A. Nielsen and I. L. Chuang, Quantum Computation and Quantum Information, Cambridge University Press, 2010.
• M. Reed and B. Simon, Methods of Modern Mathematical Physics, Volume I: Functional Analysis, Academic Press, 1980.
• J. J. Sakurai and J. Napolitano, Modern Quantum Mechanics, 3rd ed., Cambridge University Press, 2020.
• R. Shankar, Principles of Quantum Mechanics, 2nd ed., Springer, 1994.
• J. von Neumann, Mathematical Foundations of Quantum Mechanics, Princeton University Press, 1955.
• S. Weinberg, The Quantum Theory of Fields, Volume I, Cambridge University Press, 1995.