nLab rigged Hilbert space

Redirected from "rigged Hilbert spaces".

Contents

Idea

In quantum mechanics one routinely speaks of eigenstates of quantum observables with continuous spectrum, notably of eigenstates of position and of momentum, written in bra-ket notation as |x⟩\vert x \rangle and |p⟩\vert p \rangle and “normalized”, following Dirac, in terms of the Dirac delta distribution, ⟨x|x′⟩=δ(x−x′)\langle x \vert x' \rangle = \delta(x - x'). But these self-adjoint unbounded operators have no actual eigenvectors in the Hilbert space of states (see below). The spectral theorem in its Hilbert space form makes rigorous sense of such operators by trading their eigenvectors for spectral measures (projection-valued measures), but in doing so it gives up the transparent picture of expanding any state in a “basis” of eigenstates, which is what makes Dirac’s formalism so effective in practice.

The notion of rigged Hilbert spaces (also: equipped Hilbert spaces, closer to the Russian original, or Gelfand triples) restores this picture rigorously, by enlarging the given Hilbert space ℋ\mathcal{H} to a space of generalized vectors in which the missing eigenvectors do exist — much as distributions, such as the Dirac distribution, enlarge the space of ordinary functions. To this end one chooses a dense subspace Φ⊂ℋ\Phi \subset \mathcal{H} (notation following Gel’fand & Vilenkin 1961 §4.2–3) of particularly well-behaved vectors, equipped with its own topology, finer than the one induced from ℋ\mathcal{H} (typically making Φ\Phi a nuclear Fréchet space, such as a Schwartz space of test functions). Since continuity with respect to this finer topology is a weaker condition, there are more continuous antilinear functionals on Φ\Phi than on ℋ\mathcal{H} — the latter being, by the Riesz representation theorem, just the vectors of ℋ\mathcal{H} themselves. This yields a triple of continuous inclusions

Φ⊂ℋ⊂Φ ×, \Phi \;\subset\; \mathcal{H} \;\subset\; \Phi^\times \mathrlap{\,,}

where Φ ×\Phi^\times denotes the space of continuous antilinear functionals on Φ\Phi, into which ψ∈ℋ\psi \in \mathcal{H} embeds as ⟨−|ψ⟩\langle - \vert \psi \rangle.

If now a self-adjoint operator AA on ℋ\mathcal{H} restricts to a continuous linear operator A:Φ→ΦA \colon \Phi \to \Phi, then it extends by duality to an operator A ×:Φ ×→Φ ×A^\times \colon \Phi^\times \to \Phi^\times on generalized vectors, given by (A ×F)(ϕ)≔F(Aϕ)(A^\times F)(\phi) \coloneqq F(A \phi), and a generalized eigenvector of AA with generalized eigenvalue λ\lambda is an element F∈Φ ×F \in \Phi^\times with A ×F=λFA^\times F = \lambda F. The nuclear spectral theorem (due to Gel’fand & Kostyuchenko and to Maurin, hence also called the Gel’fand-Maurin theorem, see Gel’fand & Vilenkin 1961, Ch. I, Blanchard & Brüning 2015, Ch. 29) says that when Φ\Phi is nuclear, then AA has a complete system of generalized eigenvectors |λ⟩∈Φ ×\vert \lambda \rangle \in \Phi^\times, indexed by (almost all) points λ\lambda of its spectrum (and in general also by a multiplicity label, here suppressed), in that there is a measure μ\mu on the spectrum such that

⟨ϕ|ψ⟩=∫⟨ϕ|λ⟩⟨λ|ψ⟩dμ(λ)for allϕ,ψ∈Φ. {\langle \phi \vert \psi \rangle} \;=\; \int {\langle \phi \vert \lambda \rangle} \, {\langle \lambda \vert \psi \rangle} \; \mathrm{d}\mu(\lambda) \;\;\;\;\;\; \text{for all}\; \phi, \psi \in \Phi \,.

This is the rigorous form of Dirac’s completeness relation “∫|λ⟩⟨λ|dμ(λ)=id\int {\vert \lambda \rangle} {\langle \lambda \vert} \, \mathrm{d}\mu(\lambda) = \mathrm{id}”.

In this way, rigged Hilbert spaces provide a rigorous home for Dirac’s bra-ket formalism: Kets |F⟩{\vert F \rangle} are elements of Φ ×\Phi^\times, the corresponding bras ⟨F|\langle F \vert are their complex conjugates, which are elements of the dual space Φ′\Phi' of continuous linear functionals, and a bracket

⟨ϕ|F⟩≔F(ϕ),⟨F|ϕ⟩≔F(ϕ)¯ \langle \phi \vert F \rangle \;\coloneqq\; F(\phi) \,, \;\;\;\;\;\; \langle F \vert \phi \rangle \;\coloneqq\; \overline{F(\phi)}

is well-defined as long as at least one of its two entries is a test vector in Φ\Phi. (Brackets between two generalized vectors, such as ⟨x|x′⟩\langle x \vert x' \rangle, are not defined in general; see below for what they stand for.)

Examples

Position and momentum on the line

The archetypical example is the quantum mechanics of a particle on the line, with the eigenstates of position and of momentum as generalized vectors. Throughout, let

ℋ≔L 2(ℝ),⟨ϕ|ψ⟩≔∫ ℝϕ(x)¯ψ(x)dx \mathcal{H} \;\coloneqq\; L^2(\mathbb{R}) \,, \;\;\;\;\;\; \langle \phi \vert \psi \rangle \;\coloneqq\; \int_{\mathbb{R}} \overline{\phi(x)} \, \psi(x) \, \mathrm{d}x

be the Hilbert space of square-integrable functions on the real line, with respect to Lebesgue measure, with inner product (as usual in physics) antilinear in its first argument.

The problem

The position operator XX and the momentum operator PP (in units with ℏ=1\hbar = 1)

(Xψ)(x)≔x⋅ψ(x),(Pψ)(x)≔−iddxψ(x) (X \psi)(x) \;\coloneqq\; x \cdot \psi(x) \,, \;\;\;\;\;\;\;\; (P \psi)(x) \;\coloneqq\; - i \frac{\mathrm{d}}{\mathrm{d}x} \psi(x)

are unbounded self-adjoint operators on ℋ\mathcal{H}, defined not on all of ℋ\mathcal{H} but on dense domains (for instance D(X)={ψ∈L 2(ℝ)|x⋅ψ∈L 2(ℝ)}D(X) = \big\{ \psi \in L^2(\mathbb{R}) \,\big\vert\, x \cdot \psi \in L^2(\mathbb{R}) \big\}). Both have spectrum all of ℝ\mathbb{R}, but:

Remark

(no eigenvectors in the Hilbert space)

Neither XX nor PP has a single eigenvector in ℋ\mathcal{H}:

  • If Xψ=ξψX \psi = \xi \psi, then (x−ξ)ψ(x)=0(x - \xi) \psi(x) = 0 for almost all xx, whence ψ=0∈L 2(ℝ)\psi = 0 \in L^2(\mathbb{R}).

  • If Pψ=kψP \psi = k \psi, then ψ′=ikψ\psi' = i k \psi in the sense of distributions, whence ψ(x)=ce ikx\psi(x) = c \, e^{i k x}, which is square-integrable only for c=0c = 0.

The would-be eigenfunctions, namely the Dirac distribution δ(x−ξ)\delta(x - \xi) and the plane wave e ikxe^{i k x}, do make sense, but as distributions instead of as elements of L 2(ℝ)L^2(\mathbb{R}). The rigged Hilbert space makes this precise.

The rigging

As the space of test vectors take the Schwartz space 𝒮(ℝ)\mathcal{S}(\mathbb{R}) of smooth functions all of whose derivatives are rapidly decreasing, with its Fréchet space-topology given by the seminorms

p m,n(ϕ)≔sup x∈ℝ|x mϕ (n)(x)|,m,n∈ℕ, p_{m,n}(\phi) \;\coloneqq\; \sup_{x \in \mathbb{R}} \left\vert x^m \, \phi^{(n)}(x) \right\vert \,, \;\;\;\;\;\; m, n \in \mathbb{N} \,,

and write

ϕ^(k)≔12π∫ ℝe −ikxϕ(x)dx \widehat{\phi}(k) \;\coloneqq\; \frac{1}{\sqrt{2\pi}} \int_{\mathbb{R}} e^{- i k x} \, \phi(x) \, \mathrm{d}x

for the Fourier transform.

Proposition

(Schwartz space as test space for position and momentum)

  1. The inclusion 𝒮(ℝ)↪L 2(ℝ)\mathcal{S}(\mathbb{R}) \hookrightarrow L^2(\mathbb{R}) is continuous with dense image.

  2. XX and PP restrict to continuous linear operators 𝒮(ℝ)→𝒮(ℝ)\mathcal{S}(\mathbb{R}) \to \mathcal{S}(\mathbb{R}).

  3. The Fourier transform restricts to a linear automorphism of 𝒮(ℝ)\mathcal{S}(\mathbb{R}) which preserves the inner product (by the Plancherel theorem) and turns PP into multiplication by the momentum variable, Pϕ^(k)=kϕ^(k)\widehat{P\phi}(k) = k \, \widehat{\phi}(k) (and XX into iddki \frac{\mathrm{d}}{\mathrm{d}k}).

  4. 𝒮(ℝ)\mathcal{S}(\mathbb{R}) is a nuclear space.

Proof

Using 1+x 2≤(1+|x|) 21 + x^2 \leq (1 + \vert x \vert)^2 we have

‖ϕ‖ L 2 2=∫ ℝ(1+x 2)|ϕ(x)| 21+x 2dx≤π(p 0,0(ϕ)+p 1,0(ϕ)) 2, \Vert \phi \Vert^2_{L^2} \;=\; \int_{\mathbb{R}} \frac{ (1 + x^2) \, \vert \phi(x) \vert^2 }{ 1 + x^2 } \, \mathrm{d}x \;\leq\; \pi \, \big( p_{0,0}(\phi) + p_{1,0}(\phi) \big)^2 \,,

which gives the continuity in the first item, while density holds since already the compactly supported smooth functions are dense in L 2(ℝ)L^2(\mathbb{R}). For the second item, the Leibniz rule gives

p m,n(Xϕ)≤p m+1,n(ϕ)+np m,n−1(ϕ),p m,n(Pϕ)=p m,n+1(ϕ) p_{m,n}(X \phi) \;\leq\; p_{m+1,n}(\phi) + n \, p_{m,n-1}(\phi) \,, \;\;\;\;\;\; p_{m,n}(P \phi) \;=\; p_{m,n+1}(\phi)

(with p m,−1≔0p_{m,-1} \coloneqq 0). The last two items are standard, see at Schwartz space and at nuclear space.

Now let 𝒮 ×(ℝ)\mathcal{S}^\times(\mathbb{R}) be the space of continuous antilinear functionals F:𝒮(ℝ)→ℂF \colon \mathcal{S}(\mathbb{R}) \to \mathbb{C}. Since 𝒮(ℝ)\mathcal{S}(\mathbb{R}) is closed under complex conjugation, F↦F((−)¯)F \mapsto F\big(\overline{(-)}\big) is a linear isomorphism onto the usual space 𝒮′(ℝ)\mathcal{S}'(\mathbb{R}) of tempered distributions, so that 𝒮 ×(ℝ)\mathcal{S}^\times(\mathbb{R}) is just the space of tempered distributions in a guise adapted to the inner product. Concretely, every locally integrable function ff of at most polynomial growth defines an element of 𝒮 ×(ℝ)\mathcal{S}^\times(\mathbb{R}) by

⟨ϕ|f⟩≔∫ ℝϕ(x)¯f(x)dx, \langle \phi \vert f \rangle \;\coloneqq\; \int_{\mathbb{R}} \overline{\phi(x)} \, f(x) \,\mathrm{d}x \,,

and so does every f∈L 2(ℝ)f \in L^2(\mathbb{R}), continuously so by the Cauchy-Schwarz inequality and the first item of Prop. . The resulting map L 2(ℝ)→𝒮 ×(ℝ)L^2(\mathbb{R}) \to \mathcal{S}^\times(\mathbb{R}) is injective, since 𝒮(ℝ)\mathcal{S}(\mathbb{R}) is dense in L 2(ℝ)L^2(\mathbb{R}), and so we have the rigged Hilbert space

𝒮(ℝ)⊂L 2(ℝ)⊂𝒮 ×(ℝ)≃𝒮′(ℝ). \mathcal{S}(\mathbb{R}) \;\subset\; L^2(\mathbb{R}) \;\subset\; \mathcal{S}^\times(\mathbb{R}) \;\simeq\; \mathcal{S}'(\mathbb{R}) \,.

For F∈𝒮 ×(ℝ)F \in \mathcal{S}^\times(\mathbb{R}) and ϕ∈𝒮(ℝ)\phi \in \mathcal{S}(\mathbb{R}) we write, as in the Idea section, ⟨ϕ|F⟩≔F(ϕ)\langle \phi \vert F \rangle \coloneqq F(\phi) and ⟨F|ϕ⟩≔F(ϕ)¯\langle F \vert \phi \rangle \coloneqq \overline{F(\phi)}, which for F∈L 2(ℝ)F \in L^2(\mathbb{R}) reproduces the inner product.

Finally, for A∈{X,P}A \in \{X, P\} define A ×:𝒮 ×(ℝ)→𝒮 ×(ℝ)A^\times \colon \mathcal{S}^\times(\mathbb{R}) \to \mathcal{S}^\times(\mathbb{R}) by

⟨ϕ|A ×F⟩≔⟨Aϕ|F⟩. \langle \phi \vert A^\times F \rangle \;\coloneqq\; \langle A \phi \vert F \rangle \,.

By self-adjointness of AA, this restricts on the domain D(A)⊂L 2(ℝ)⊂𝒮 ×(ℝ)D(A) \subset L^2(\mathbb{R}) \subset \mathcal{S}^\times(\mathbb{R}) to the original operator AA, while under the identification 𝒮 ×(ℝ)≃𝒮′(ℝ)\mathcal{S}^\times(\mathbb{R}) \simeq \mathcal{S}'(\mathbb{R}) it is the usual action of x⋅(−)x \cdot (-) and of −iddx-i \frac{\mathrm{d}}{\mathrm{d}x} on tempered distributions.

Generalized eigenvectors

For ξ,k∈ℝ\xi, k \in \mathbb{R} consider the generalized vectors δ ξ,e k∈𝒮 ×(ℝ)\delta_\xi, e_k \in \mathcal{S}^\times(\mathbb{R}) given by

⟨ϕ|δ ξ⟩≔ϕ(ξ)¯,e k(x)≔12πe ikx, \langle \phi \vert \delta_\xi \rangle \;\coloneqq\; \overline{\phi(\xi)} \,, \;\;\;\;\;\;\;\; e_k(x) \;\coloneqq\; \frac{1}{\sqrt{2\pi}} e^{i k x} \,,

hence with

⟨δ ξ|ϕ⟩=ϕ(ξ),⟨e k|ϕ⟩=ϕ^(k). \langle \delta_\xi \vert \phi \rangle \;=\; \phi(\xi) \,, \;\;\;\;\;\;\;\; \langle e_k \vert \phi \rangle \;=\; \widehat{\phi}(k) \,.

(These are continuous on 𝒮(ℝ)\mathcal{S}(\mathbb{R}), since |⟨ϕ|δ ξ⟩|≤p 0,0(ϕ)\vert \langle \phi \vert \delta_\xi \rangle \vert \leq p_{0,0}(\phi) and |⟨ϕ|e k⟩|≤π/2(p 0,0(ϕ)+p 2,0(ϕ))\vert \langle \phi \vert e_k \rangle \vert \leq \sqrt{\pi/2} \, \big( p_{0,0}(\phi) + p_{2,0}(\phi) \big).)

Proposition

(generalized position and momentum eigenvectors)

For all ξ,k∈ℝ\xi, k \in \mathbb{R} we have

X ×δ ξ=ξδ ξ,P ×e k=ke k. X^\times \delta_\xi \;=\; \xi \, \delta_\xi \,, \;\;\;\;\;\;\;\; P^\times e_k \;=\; k \, e_k \,.
Proof

For all ϕ∈𝒮(ℝ)\phi \in \mathcal{S}(\mathbb{R}) we have, using that ξ\xi is real,

⟨ϕ|X ×δ ξ⟩=⟨Xϕ|δ ξ⟩=ξϕ(ξ)¯=ξ⟨ϕ|δ ξ⟩, \langle \phi \vert X^\times \delta_\xi \rangle \;=\; \langle X\phi \vert \delta_\xi \rangle \;=\; \overline{\xi \, \phi(\xi)} \;=\; \xi \, \langle \phi \vert \delta_\xi \rangle \,,

and, by integration by parts (the boundary terms vanishing since ϕ\phi is rapidly decreasing while e ke_k is bounded),

⟨ϕ|P ×e k⟩ =⟨Pϕ|e k⟩=∫ ℝiϕ′(x)¯e k(x)dx =−i∫ ℝϕ(x)¯e k′(x)dx=k⟨ϕ|e k⟩. \begin{aligned} \langle \phi \vert P^\times e_k \rangle & =\; \langle P \phi \vert e_k \rangle \;=\; \int_{\mathbb{R}} i \, \overline{\phi'(x)} \, e_k(x) \,\mathrm{d}x \\ & =\; - i \int_{\mathbb{R}} \overline{\phi(x)} \, e_k'(x) \,\mathrm{d}x \;=\; k \, \langle \phi \vert e_k \rangle \,. \end{aligned}

Completeness and Dirac’s notation

Proposition

(completeness relations)

For all ϕ,ψ∈𝒮(ℝ)\phi, \psi \in \mathcal{S}(\mathbb{R}) we have

⟨ϕ|ψ⟩=∫ ℝ⟨ϕ|δ ξ⟩⟨δ ξ|ψ⟩dξ=∫ ℝ⟨ϕ|e k⟩⟨e k|ψ⟩dk; \langle \phi \vert \psi \rangle \;=\; \int_{\mathbb{R}} \langle \phi \vert \delta_\xi \rangle \, \langle \delta_\xi \vert \psi \rangle \,\mathrm{d}\xi \;=\; \int_{\mathbb{R}} \langle \phi \vert e_k \rangle \, \langle e_k \vert \psi \rangle \,\mathrm{d}k \,;

equivalently, every test vector ψ\psi has the generalized eigenvector expansions

ψ=∫ ℝψ(ξ)δ ξdξ=∫ ℝψ^(k)e kdk, \psi \;=\; \int_{\mathbb{R}} \psi(\xi) \, \delta_\xi \,\mathrm{d}\xi \;=\; \int_{\mathbb{R}} \widehat{\psi}(k) \, e_k \,\mathrm{d}k \,,

where the integrals converge weakly in 𝒮 ×(ℝ)\mathcal{S}^\times(\mathbb{R}), i.e. after pairing with any ϕ∈𝒮(ℝ)\phi \in \mathcal{S}(\mathbb{R}).

Proof

The integrands are ϕ(ξ)¯ψ(ξ)\overline{\phi(\xi)} \, \psi(\xi) and ϕ^(k)¯ψ^(k)\overline{\widehat{\phi}(k)} \, \widehat{\psi}(k), respectively. Hence the first equality is the definition of the inner product on L 2(ℝ)L^2(\mathbb{R}), and the second is the Plancherel theorem. In the weak form, the second expansion of ψ\psi is thereby Fourier inversion.

Remark

(Dirac’s notation)

In Dirac’s bra-ket notation, δ x\delta_x is the position eigenket |x⟩\vert x \rangle and e pe_p is the momentum eigenket |p⟩\vert p \rangle, and the above says precisely that:

  • ⟨x|ψ⟩=ψ(x)\langle x \vert \psi \rangle = \psi(x) and ⟨p|ψ⟩=ψ^(p)\langle p \vert \psi \rangle = \widehat{\psi}(p) are the wavefunctions of ψ\psi in the position and in the momentum representation, respectively;

  • ∫|x⟩⟨x|dx=id=∫|p⟩⟨p|dp\int \vert x \rangle \langle x \vert \, \mathrm{d}x = \mathrm{id} = \int \vert p \rangle \langle p \vert \, \mathrm{d}p are the completeness relations of Prop. ;

  • ⟨x|p⟩=12πe ipx\langle x \vert p \rangle = \frac{1}{\sqrt{2\pi}} e^{i p x} — here, exceptionally, a bracket between two generalized vectors makes sense, because e pe_p is a smooth function, whose value at xx is defined.

On the other hand, formulas such as ⟨x|x′⟩=δ(x−x′)\langle x \vert x' \rangle = \delta(x - x') and ⟨p|p′⟩=δ(p−p′)\langle p \vert p' \rangle = \delta(p - p') pair two generalized vectors and have no literal meaning in the rigged Hilbert space; they are mnemonics for the completeness relations, obtained by formally inserting ψ=|x′⟩\psi = \vert x' \rangle (resp. ψ=|p′⟩\psi = \vert p' \rangle) into the expansions of Prop. .

Remark

(dependence on the choice of test space)

In this example, the generalized eigenvalues of XX and PP on 𝒮 ×(ℝ)\mathcal{S}^\times(\mathbb{R}) are exactly the points of their spectrum ℝ\mathbb{R}: For λ∈ℂ∖ℝ\lambda \in \mathbb{C} \setminus \mathbb{R}, multiplication by the function 1/(x−λ¯)1/(x - \overline{\lambda}) (all of whose derivatives are bounded) is a continuous inverse to X−λ¯X - \overline{\lambda} on 𝒮(ℝ)\mathcal{S}(\mathbb{R}), so that X ×−λ=(X−λ¯) ×X^\times - \lambda = (X - \overline{\lambda})^\times is invertible on 𝒮 ×(ℝ)\mathcal{S}^\times(\mathbb{R}) and hence has trivial kernel; similarly for PP, via Fourier transform.

But the generalized eigenvalues do depend on the choice of test space:

  • A smaller test space has a larger antidual, which may contain spurious generalized eigenvectors. For instance, with the space C c ∞(ℝ)C^\infty_c(\mathbb{R}) of compactly supported smooth functions (also nuclear and also preserved by XX and PP) as test space, the antidual (equivalently: all distributions) contains e ikxe^{i k x} for every complex number kk, and these are generalized eigenvectors of PP with non-real eigenvalue kk, which do not contribute to the completeness relation.

  • A larger test space with coarser topology may lose the generalized eigenvectors. For instance, on the domain D(X)D(X) with its graph norm, point evaluations ϕ↦ϕ(ξ)\phi \mapsto \phi(\xi) are not even defined (elements of D(X)D(X) being only L 2L^2-classes), and so its antidual contains no δ ξ\delta_\xi.

The Schwartz space sits well between these extremes, simultaneously for XX and for PP, which is no accident:

Remark

(intrinsic characterization via the harmonic oscillator)

The Schwartz space is intrinsically determined by XX and PP: It is the space of smooth vectors of the Schrödinger representation of the Heisenberg group on L 2(ℝ)L^2(\mathbb{R}), and coincides with the intersection ⋂ nD(N n)\bigcap_{n} D(N^n) of the domains of all powers of the harmonic oscillator Hamiltonian N≔12(X 2+P 2)N \coloneqq \tfrac{1}{2}(X^2 + P^2). Expanding in the orthonormal basis of Hermite functions h nh_n, which are honest eigenvectors Nh n=(n+12)h nN h_n = (n + \tfrac{1}{2}) h_n, identifies the above rigged Hilbert space with

s⊂ℓ 2⊂s ×, s \;\subset\; \ell^2 \;\subset\; s^\times \,,

where ss is the space of rapidly decreasing sequences and s ×s^\times that of sequences of at most polynomial growth. In this form the nuclearity of 𝒮(ℝ)≃s\mathcal{S}(\mathbb{R}) \simeq s is manifest: ss is the intersection of the weighted Hilbert spaces ℓ m 2≔{(c n) n∈ℕ|∑ n(n+1) 2m|c n| 2<∞}\ell^2_m \coloneqq \big\{ (c_n)_{n \in \mathbb{N}} \,\big\vert\, \sum_n (n+1)^{2m} \vert c_n \vert^2 \lt \infty \big\}, and the inclusions ℓ m+1 2↪ℓ m 2\ell^2_{m+1} \hookrightarrow \ell^2_m are Hilbert-Schmidt.

References

The notion of rigged Hilbert spaces, together with the nuclear spectral theorem for them, is due to:

  • I. M. Gel'fand, N. Ya. Vilenkin, §I.4 of: Generalized Functions, Vol. 4: Applications of Harmonic Analysis, Fizmatgiz, Moscow (1961); English translation: Academic Press (1964); reprint: AMS Chelsea (2016) [doi:10.1090/chel/380]

Early treatments in the context of Dirac’s bra-ket formalism:

Textbook account, including the nuclear spectral theorem:

  • Philippe Blanchard, Erwin Brüning, Ch. 29 of: Mathematical Methods in Physics – Distributions, Hilbert Space Operators, Variational Methods, and Applications in Quantum Physics, Progress in Mathematical Physics 69, Birkhäuser (2015) [doi:10.1007/978-3-319-14045-2]

Pedagogical introduction aimed at physicists, via an exactly solvable example:

Claim of a unification of various inequivalent approaches:

  • M. Gadella, F. Gómez, A Unified Mathematical Formalism for the Dirac Formulation of Quantum Mechanics, Foundations of Physics 32 (2002) 815–869 [doi:10.1023/A:1016069311589]

See also:

Last revised on October 26, 2026 at 20:11:40. See the history of this page for a list of all contributions to it.