Common Confusions

Many mistakes in rigorous quantum mechanics come from transferring a true finite-dimensional statement to a setting where one of its hidden assumptions fails. The useful response is to identify the missing hypothesis, rather than to declare the whole physical method invalid. The checks below distinguish the most frequent errors and direct the reader to the precise treatment.

Operator formulae do not determine operator domains

Confusion: real expectation values prove self-adjointness. For a densely defined operator, real diagonal expectation values on its domain establish symmetry under the usual polarization argument. Self-adjointness also requires equality with the adjoint domain. An integration-by-parts calculation must therefore determine which boundary conditions the adjoint permits. See Symmetric Operators and Self-Adjoint Operators.

Confusion: any dense set is a sufficient domain for a closed operator. Hilbert-norm density lets one approximate state vectors. An operator core must approximate both the vector and its image, in graph norm. A dense subspace can omit boundary data that distinguish self-adjoint extensions. The Dense Domains page gives a concrete counterexample.

Confusion: a commutator identity can be multiplied or traced without conditions. Products require their intermediate vectors to lie in the next domain. Trace cyclicity requires suitable ideal assumptions, such as one trace-class factor and one bounded factor. The finite-matrix identity $\operatorname{Tr}[A,B]=0$ is not a license to trace arbitrary unbounded commutators. Common Operator Pitfalls and Trace-Class Operators explain the respective repairs.

Spectrum and eigenvectors answer different questions

Confusion: a spectral value must have a normalizable eigenstate. Spectrum concerns failure of a bounded everywhere-defined resolvent. Eigenvalues describe failure of injectivity. Continuous spectral values can instead have approximate eigenvectors without any nonzero vector in the corresponding kernel. The spectrum classification distinguishes these mechanisms.

Confusion: continuous spectrum always means a probability density. A spectral measure can have absolutely continuous or singular continuous components. The absence of point masses alone does not imply the existence of a Lebesgue density. Even for an absolutely continuous distribution, a density at a point is not the probability of that singleton. Review Measure Theory Minimum.

Confusion: a delta-normalized ket is a physical vector with awkward normalization. A generalized ket acts in a test-space dual. It need not belong to the Hilbert space at all. Its distributional eigen-equation and a normalizable eigenvector equation have different meanings. Distribution Theory Minimum makes the pairing explicit.

Normalization does not supply regularity or finite moments

Confusion: every state has finite energy and entropy. Unit trace or unit norm gives total probability one. Finite moments require weighted integrability, and finite von Neumann entropy requires summability of $-p_n\log p_n$. These are additional constraints. A state can have a finite first moment but fail to belong to the operator domain, which tests a second moment. See Density Operators and Entropy and Trace.

Confusion: mixed states represent a unique classical lottery. Different ensembles can produce the same density operator. A reduced state can also be mixed because it is part of a larger entangled pure state. The operator fixes measurement statistics for the selected system, not a unique preparation history. Pure and Mixed States distinguishes purity from decomposition.

Convergence must name its tests

Confusion: a successful basis cutoff controls all states. Strong convergence controls each fixed input separately; operator-norm convergence controls all unit inputs uniformly. Neither guarantees convergence of unbounded moments without further estimates. A sequence of normalized states can converge weakly as vectors to zero without losing its norm at any finite step. The topology primer identifies the precise quantifiers.

Confusion: an infinite-dimensional identity follows because every finite matrix truncation satisfies it. A truncation may change commutators, domains, or spectral type, and different quantities can converge in different topologies. The limit must be justified for the specific statement being made. This is particularly important for the CCR and thermodynamic limits.

Exercise: a bounded truncation hides an unbounded limit

On $\ell^2(\mathbb N)$, let $N_Me_n=\min(n,M)e_n$. Every $N_M$ is a bounded self-adjoint operator. Does this show that the limit expression $Ne_n=ne_n$ is bounded or defined on every vector?

Solution

No: $\|N_M\|=M$, so there is no common bound. For $\psi=\sum c_ne_n$, convergence of $N_M\psi$ in norm requires $\sum n^2|c_n|^2<\infty$. If this sum diverges, monotone convergence gives $\|N_M\psi\|^2\to\infty$. The limiting self-adjoint operator has the proper domain given by that weighted summability condition. Agreement on every finite linear combination of basis vectors does not remove this domain restriction.

References

• Hall, Brian C., Quantum Theory for Mathematicians, Graduate Texts in Mathematics 267, Springer, 2013, doi:10.1007/978-1-4614-7116-5 — operator and state-space distinctions.
• Holevo, Alexander S., Quantum Systems, Channels, Information: A Mathematical Introduction, 2nd ed., De Gruyter, 2019, doi:10.1515/9783110642490 — density operators, entropy, and infinite systems.
• Teschl, Gerald, Mathematical Methods in Quantum Mechanics: With Applications to Schrödinger Operators, 2nd ed., Graduate Studies in Mathematics 157, American Mathematical Society, 2014, author’s edition — domains, spectral types, and convergence.