The Heisenberg group, the Schrödinger representation, and Stone–von Neumann as quantization

Intuition Beginner

Quantum mechanics begins with two operators, position $\hat{q}$ and momentum $\hat{p}$, that refuse to commute: $\hat{q}\hat{p}-\hat{p}\hat{q}=i\hslash$. Every textbook writes down a particular pair — multiply by $x$ for position, take a derivative for momentum — checks the relation, and moves on. A natural worry follows that pause. Could a different physicist, writing down a cleverly different pair of operators on a different space of wavefunctions, build a genuinely different quantum mechanics that still satisfies the same relation?

The Stone–von Neumann theorem answers no. For a system with finitely many degrees of freedom, there is essentially one pair of position and momentum operators. Any two honest realisations of the relation $\hat{q}\hat{p}-\hat{p}\hat{q}=i\hslash$ are the same up to a change of variables — a relabeling of states by a fixed dictionary. The familiar wavefunction picture, the momentum-space picture, and the holomorphic Bargmann–Fock picture of the harmonic oscillator are not three theories but one theory in three costumes.

The cleanest way to see the uniqueness is to package the operators into a group. The position shifts, the momentum shifts, and a phase that records their failure to commute together form the Heisenberg group. Quantum mechanics is then a single irreducible representation of this group: one indivisible way for the group to act by symmetries on a space of states. The theorem says that representation is unique once you fix the value of $\hslash$.

This is what "quantization" means made precise. To quantize a classical system with finitely many coordinates is to pick that one irreducible representation. There is no menu of choices. The surprise comes later: for a field, which carries infinitely many degrees of freedom, the uniqueness breaks, infinitely many inequivalent quantum theories appear, and that breakdown is the doorway into algebraic quantum field theory.

Visual Beginner

Picture the classical phase plane of one particle: a horizontal axis for position and a vertical axis for momentum. A point in that plane is a classical state. Now add a third direction, a vertical "phase" axis stacked above the plane. The Heisenberg group lives in this three-storey picture: an element is a position shift, a momentum shift, and an accumulated phase.

The reason the third floor matters is the order of operations. Shift a wavefunction in position and then in momentum, and you do not land where you would if you shifted in the reverse order. The two routes differ by a phase, and that phase is exactly the extra central coordinate. The Heisenberg group records the geometry of "position-then-momentum is not momentum-then-position."

On the right of the picture sits a space of wavefunctions on the line. The group acts: a position shift slides the wavefunction sideways, a momentum shift multiplies it by an oscillating phase, and the central coordinate multiplies the whole state by a constant phase. This action is the Schrödinger representation. The Stone–von Neumann theorem says that any other faithful, indecomposable action of the group with the same central phase is a relabeled copy of this one. The Fourier transform, drawn as the arrow turning the position picture into the momentum picture, is one such relabeling — the dictionary that proves two costumes wear the same body.

Worked example Beginner

Take the simplest case: one degree of freedom, position and momentum on the real line. Write the position-shift operator and the momentum-shift operator in their exponentiated forms. A position shift by amount $a$ acts on a wavefunction $\psi (x)$ by multiplying with an oscillating factor; a momentum shift by amount $b$ slides the wavefunction, sending $\psi (x)$ to $\psi (x-b)$. Call the first $U(a)$ and the second $V(b)$.

Apply them in both orders to a wavefunction and compare. Doing $U(a)$ then $V(b)$ versus $V(b)$ then $U(a)$ gives the same wavefunction up to a constant phase: the two results differ by the factor $e^{iab}$ (with $\hslash =1$). In symbols,

$$U(a)V(b)=e^{iab}V(b)U(a).$$

That single phase factor is the whole content of "position and momentum do not commute," now expressed with bounded, well-behaved operators instead of the awkward unbounded $\hat{q}$ and $\hat{p}$.

Plug in numbers. Take $a=1$ and $b=1$. Then the mismatch phase is $e^i$, a complex number of modulus one and angle one radian — not equal to $1$, so the two orders genuinely disagree. Take $a=2\pi$ and $b=1$: the phase is $e^{2\pi i}=1$, and for that special pair the operators commute. The amount of non-commuting is dialed by the product $ab$ measured in units of $\hslash$.

The lesson the theorem draws from this: the family of operators $U(a)$ and $V(b)$, together with the phases they generate, is rigid. Once you insist on the phase relation above with a fixed $\hslash$, you have pinned down the representation up to a relabeling. The wavefunction realisation here and the momentum-space realisation a Fourier transform away are the same single quantum mechanics.

Check your understanding Beginner

Formal definition Intermediate+

The Heisenberg Lie algebra and the Heisenberg group

Fix $n$ degrees of freedom. The Heisenberg Lie algebra ${\mathfrak{h}}_{2n+1}$ is the real $(2n+1)$-dimensional Lie algebra with basis $\{q_1,\dots ,q_n,p_1,\dots ,p_n,c\}$ and brackets

$$[q_j,p_k]={\delta }_{jk}c,[q_j,q_k]=[p_j,p_k]=0,[c,\cdot ]=0,$$

so $c$ is central. This is the Lie algebra of the canonical commutation relations: under quantization the central generator $c$ becomes $i\hslash$ times the identity and the brackets reproduce $[{\hat{q}}_j,{\hat{p}}_k]=i\hslash {\delta }_{jk}1$.

The Heisenberg group $H_{2n+1}$ is the simply connected Lie group with this Lie algebra. Concretely it is ${\mathbb{R}}^{2n}\times \mathbb{R}$ with the group law

$$(a,b,t)\cdot (a^{\prime },b^{\prime },t^{\prime })=(a+a^{\prime },b+b^{\prime },t+t^{\prime }+\frac{1}{2}(a\cdot b^{\prime }-b\cdot a^{\prime }\left)\right),$$

where $a,b\in {\mathbb{R}}^n$ are the position-shift and momentum-shift parts and $t\in \mathbb{R}$ is the central coordinate. The cocycle $\frac{1}{2}(a\cdot b^{\prime }-b\cdot a^{\prime })$ is half the symplectic pairing of $(a,b)$ with $(a^{\prime },b^{\prime })$. Thus $H_{2n+1}$ is a central extension of the abelian phase space ${\mathbb{R}}^{2n}$ by the central subgroup $\mathbb{R}$,

$$1\to \mathbb{R}\to H_{2n+1}\to {\mathbb{R}}^{2n}\to 1,$$

with extension class the standard symplectic form. The non-commutativity of $H_{2n+1}$ is concentrated entirely in the central direction; the quotient by the centre is the commutative phase space.

The Schrödinger representation

A unitary representation $\rho$ of $H_{2n+1}$ on a Hilbert space $\mathcal{K}$ has a fixed central character: the central subgroup $\{(0,0,t)\}$ acts by scalars $\rho (0,0,t)=e^{i\hslash t}1$ for a real number $\hslash$, the value that labels the representation. The Schrödinger representation at central character $\hslash =1$ acts on $\mathcal{K}=L^2({\mathbb{R}}^n)$ by

$$({\rho }_{\mathrm{S}}(a,b,t)\psi )(x)=e^{it}{e}^{i(a\cdot x-\frac{1}{2}a\cdot b)}\psi (x-b),$$

so that position shifts act by modulation, momentum shifts act by translation, and the centre acts by an overall phase. Writing $W(a,b):={\rho }_{\mathrm{S}}(a,b,0)$ for the Weyl operators, one obtains the exponentiated canonical commutation relations

$$W(a,b)W(a^{\prime },b^{\prime })=e^{-\frac{i}{2}(a\cdot b^{\prime }-b\cdot a^{\prime })}W(a+a^{\prime },b+b^{\prime }),W(a,0)W(0,b)=e^{ia\cdot b}W(0,b)W(a,0),$$

the Weyl/CCR relations developed in 12.14.01. Formally $W(a,b)=e^{i(a\cdot \hat{q}+b\cdot \hat{p})}$ with ${\hat{q}}_j$ multiplication by $x_j$ and ${\hat{p}}_j=-i{\partial }_{x_j}$; the self-adjoint generators exist by Stone's theorem 12.02.02 because each one-parameter subgroup acts strongly continuously.

Notation conventions

• $(a,b)\in {\mathbb{R}}^{2n}$ are phase-space coordinates; $\sigma ((a,b),(a^{\prime },b^{\prime }))=a\cdot b^{\prime }-b\cdot a^{\prime }$ is the standard symplectic form.
• $\hslash$ denotes the central character; the Schrödinger representation above is normalised at $\hslash =1$, and the general $\hslash \ne 0$ case is obtained by the rescaling $x\mapsto {\hslash }^{1/2}x$.
• A representation is regular when each one-parameter subgroup $s\mapsto \rho (sa,sb,0)$ is strongly continuous, equivalently when self-adjoint generators $\hat{q},\hat{p}$ exist.
• $\mathbb{H}=H_3$ is the one-degree-of-freedom Heisenberg group; $H_{2n+1}$ is the $n$-degree-of-freedom version.

The Stone–von Neumann theorem

Theorem (Stone–von Neumann; Stone 1930, von Neumann 1931 ^{[von Neumann 1931]}). Fix $\hslash \ne 0$. Every irreducible regular unitary representation of $H_{2n+1}$ with central character $\rho (0,0,t)=e^{i\hslash t}1$ is unitarily equivalent to the Schrödinger representation on $L^2({\mathbb{R}}^n)$ at that $\hslash$. More generally, every regular unitary representation with that central character is a direct sum (a multiple) of copies of the Schrödinger representation.

Equivalently, on the Weyl side, every regular irreducible representation of the Weyl relations with a fixed nonzero $\hslash$ is unitarily equivalent to the Schrödinger one. The Schrödinger, momentum-space, and Bargmann–Fock representations are three coordinatisations of this single equivalence class. The intertwiner from the Schrödinger to the momentum picture is the Fourier transform $\mathcal{F}$; the intertwiner from the Schrödinger to the Bargmann–Fock picture is the Bargmann transform $B$ ^{[Bargmann 1961]}, which maps the oscillator ground state of 12.04.02 to the constant holomorphic function.

Quantization as choosing the unique irreducible

Reading the theorem forward, quantization of a finite system is the selection of this unique irreducible. The classical phase space ${\mathbb{R}}^{2n}$ has the Heisenberg group sitting above it as a central extension; a quantum theory is a unitary representation of that extension at fixed $\hslash \ne 0$; Stone–von Neumann says there is exactly one irreducible such representation. Through Mackey's classification of the unitary dual 07.07.07, the irreducible unitary representations of $H_{2n+1}$ split into two families: the representations with $\hslash \ne 0$, all of which are the Schrödinger representation at that $\hslash$, sitting over the single non-degenerate coadjoint orbit; and the representations with $\hslash =0$, which factor through the abelian quotient ${\mathbb{R}}^{2n}$ and are the one-dimensional characters $(a,b)\mapsto e^{i(\alpha \cdot a+\beta \cdot b)}$ — the classical observables. Quantization is the passage from the $\hslash =0$ characters to the single $\hslash \ne 0$ infinite-dimensional representation.

Counterexamples to common slips

• The theorem requires regularity (strong continuity). Dropping it admits pathological non-regular representations — for instance representations where the position operator has no self-adjoint generator — that are not unitarily equivalent to the Schrödinger one.

• The theorem requires a fixed nonzero central character. Representations at different values of $\hslash$ are inequivalent, and the $\hslash =0$ representations are the classical characters, not quantum mechanics. The central character is the invariant that the theorem holds $\hslash$ fixed to compare.

• The theorem requires finitely many degrees of freedom. For a quantum field the phase space is infinite-dimensional, the Weyl relations admit uncountably many inequivalent irreducible regular representations 12.14.01, and uniqueness fails — the structural fact behind Haag's theorem ^{[Haag 1955]}.

• "Unitarily equivalent" is stronger than "abstractly isomorphic as representations of the relations." The content is a concrete unitary intertwiner, exhibited explicitly by the Fourier and Bargmann transforms; the theorem guarantees that intertwiner exists and is unique up to a phase by Schur's lemma.

Key theorem with proof Intermediate+

Theorem (Stone–von Neumann uniqueness, one degree of freedom). Fix $\hslash =1$. Let $\rho$ be an irreducible regular unitary representation of the Heisenberg group $H_3$ on a separable Hilbert space $\mathcal{K}$ with central character $\rho (0,0,t)=e^{it}1$, and let $W(a,b)=\rho (a,b,0)$ be its Weyl operators. Then $\rho$ is unitarily equivalent to the Schrödinger representation on $L^2(\mathbb{R})$.

Proof. The strategy isolates a distinguished vector inside $\mathcal{K}$ — the abstract analogue of the oscillator ground state — and shows that the Weyl operators reconstruct all of $\mathcal{K}$ from it in a way fixed by the relations.

Step 1 (the Gaussian projection). Define the operator

$$P:=\frac{1}{2\pi }{\int }_{{\mathbb{R}}^2}{e}^{-\frac{1}{4}(a^2+b^2)}W(a,b)dadb,$$

a strongly convergent Bochner integral because $\Vert W(a,b)\Vert =1$ and the Gaussian weight is integrable. Using the Weyl relation $W(a,b)W(a^{\prime },b^{\prime })=e^{-\frac{i}{2}(a{b}^{\prime }-b{a}^{\prime })}W(a+a^{\prime },b+b^{\prime })$ and the Gaussian integral, a direct computation gives $W(a^{\prime },b^{\prime })PW(a^{\prime },b^{\prime }{)}^{\ast }=$ a Gaussian-weighted operator that, after completing the square, satisfies $P^2=P$ and $P^{\ast }=P$. Hence $P$ is an orthogonal projection. Its range is non-zero because the integrand at $(a,b)=0$ contributes the identity.

Step 2 (rank one). Compute $W(a,b)P$ and $PW(a,b)$ from the same Gaussian-integral identity. One finds, for all $(a,b)$,

$$PW(a,b)P=e^{-\frac{1}{4}(a^2+b^2)}P.$$

If ${\Omega }_1,{\Omega }_2$ are unit vectors in the range of $P$, then $⟨{\Omega }_1,W(a,b){\Omega }_2⟩=e^{-\frac{1}{4}(a^2+b^2)}⟨{\Omega }_1,{\Omega }_2⟩$. Were the range of $P$ at least two-dimensional, one could choose ${\Omega }_1\perp {\Omega }_2$, forcing $⟨{\Omega }_1,W(a,b){\Omega }_2⟩=0$ for all $(a,b)$; but the Weyl operators act irreducibly, so $\{W(a,b){\Omega }_2\}$ spans a dense subspace and no non-zero vector is orthogonal to all of it. Therefore $P$ has rank one, with a unit vector $\Omega$ spanning its range.

Step 3 (the matrix coefficient is the Schrödinger one). The single number $⟨\Omega ,W(a,b)\Omega ⟩=e^{-\frac{1}{4}(a^2+b^2)}$ is exactly the matrix coefficient of the Gaussian ground state ${\Omega }_{\mathrm{S}}(x)={\pi }^{-1/4}{e}^{-x^2/2}$ in the Schrödinger representation: a direct Gaussian integral on $L^2(\mathbb{R})$ gives $⟨{\Omega }_{\mathrm{S}},W_{\mathrm{S}}(a,b){\Omega }_{\mathrm{S}}⟩=e^{-\frac{1}{4}(a^2+b^2)}$. The cyclic vector and its matrix coefficient determine the representation: the map

$$U:\mathcal{K}\to L^2(\mathbb{R}),U(W(a,b)\Omega ):=W_{\mathrm{S}}(a,b){\Omega }_{\mathrm{S}}$$

is well-defined and isometric on the dense span because

$$⟨W(a,b)\Omega ,W(a^{\prime },b^{\prime })\Omega ⟩=⟨\Omega ,W(a,b{)}^{\ast }W(a^{\prime },b^{\prime })\Omega ⟩$$

depends only on $(a,b),(a^{\prime },b^{\prime })$ through the Weyl relations and the single matrix coefficient — and the identical computation on the Schrödinger side produces the identical number. The isometry extends to a unitary by density and irreducibility, and it intertwines $W$ with $W_{\mathrm{S}}$ by construction. Therefore $\rho$ is unitarily equivalent to the Schrödinger representation. $\square$

The same argument runs verbatim for $H_{2n+1}$ with the $n$-dimensional Gaussian, and the Bargmann transform ^{[Bargmann 1961]} is the explicit form of $U$ when the target is realised as the Bargmann–Fock holomorphic model rather than $L^2({\mathbb{R}}^n)$.

Bridge. This uniqueness result is the foundational reason canonical quantization is well-posed for a finite system, and it builds toward the whole apparatus of algebraic quantum field theory. The proof's distinguished Gaussian vector is exactly the harmonic-oscillator ground state of 12.04.02, so this is exactly the statement that the oscillator's Bargmann–Fock Hilbert space and the wavefunction Hilbert space are one representation in two costumes, with the Bargmann transform the explicit intertwiner. The theorem generalises the elementary observation that the Fourier transform turns the position picture into the momentum picture: both transforms are instances of the unique intertwiner that Schur's lemma guarantees once irreducibility and a fixed central character are fixed. Putting these together, the result appears again in 12.14.01, where the failure of this very uniqueness in infinitely many degrees of freedom — uncountably many inequivalent regular irreducible representations of the Weyl relations — is precisely what forces the algebraic stance of treating the Weyl algebra and its states, rather than any single Hilbert space, as primary. The bridge is therefore a single theorem read in two directions: forward it certifies that quantizing a particle is unambiguous; backward, its breakdown is the birth certificate of algebraic QFT.

Exercises Intermediate+

A graded set covering the Heisenberg group law, the Schrödinger representation, the Weyl relations, the Stone–von Neumann theorem, the Fourier and Bargmann intertwiners, and the Mackey reading.

Advanced results Master

The Bargmann–Fock realisation and the explicit intertwiner

The Stone–von Neumann equivalence acquires its most useful concrete form through the Bargmann–Fock model. Replace $L^2({\mathbb{R}}^n)$ by the Segal–Bargmann space $\mathcal{F}=\{F:{\mathbb{C}}^n\to \mathbb{C}\text{ holomorphic}:\int |F(z){|}^2{e}^{-|z{|}^2}dz<\infty \}$, on which the creation and annihilation operators of the oscillator 12.04.02 act as multiplication by $z_j$ and as $\partial /\partial z_j$. The Bargmann transform $B:L^2({\mathbb{R}}^n)\to \mathcal{F}$ ^{[Bargmann 1961]} is the unitary intertwiner predicted by the theorem: it sends the Gaussian oscillator ground state to the constant function $1$ and conjugates the Schrödinger Weyl operators into the Bargmann–Fock ones. The three pictures — Schrödinger on $L^2({\mathbb{R}}^n)$, momentum on $L^2({\mathbb{R}}^n)$ via the Fourier transform, Bargmann–Fock on $\mathcal{F}$ via $B$ — are thus a single equivalence class of the unique irreducible, each adapted to a different computation (wave mechanics, scattering, coherent states).

The metaplectic representation

The symplectic group $Sp(2n,\mathbb{R})$ acts on the phase space ${\mathbb{R}}^{2n}$ preserving $\sigma$, hence acts on the Heisenberg group by automorphisms fixing the centre. For each $g\in Sp(2n,\mathbb{R})$, composing the Schrödinger representation with that automorphism yields another irreducible representation at the same $\hslash$; by Stone–von Neumann it is unitarily equivalent to the original, so there is a unitary $\mu (g)$, unique up to a phase, with $\mu (g)W(v)\mu (g{)}^{-1}=W(gv)$. The assignment $g\mapsto \mu (g)$ is a projective unitary representation of $Sp(2n,\mathbb{R})$, the metaplectic representation; the phase ambiguity lifts to an honest representation of the double cover $Mp(2n,\mathbb{R})$. The Fourier transform is $\mu$ of the $90^{\circ }$ symplectic rotation, and the squeezing operators of quantum optics are $\mu$ of the symplectic boosts. The metaplectic representation is therefore the precise sense in which "linear canonical transformations are implemented by unitaries" — a direct corollary of the uniqueness theorem, and the entry point to geometric quantization of the symplectic group; the detailed half-density and Maslov-class story is developed in the geometric-quantization chapter.

The failure in infinite dimensions

For a field, the phase space $V$ is infinite-dimensional and the Heisenberg-group framing persists formally, but the rank-one Gaussian projection of the Key-theorem proof no longer pins down a single representation. Distinct compatible complex structures on $V$ give Fock representations whose vacua are not related by any unitary on a fixed Hilbert space; the Weyl relations admit uncountably many unitarily inequivalent regular irreducible representations 12.14.01. This is not a defect to be repaired but the physical content of inequivalent vacua: the Unruh and Hawking effects, spontaneous symmetry breaking, and superselection sectors all live in this failure. The algebraic response is to keep the Weyl algebra and treat the choice of representation as the choice of a quasi-free state, which is exactly the program of 12.14.01.

Synthesis. The Heisenberg-group picture is the central insight that unifies these strands: it shows that the foundational reason quantization of a finite system is unambiguous is a single group-representation fact, and this is exactly the statement that the Schrödinger, momentum, and Bargmann–Fock pictures are one irreducible representation in three polarisations, related by the Fourier and Bargmann intertwiners. The metaplectic representation generalises the Fourier transform from a single $90^{\circ }$ rotation to the full symplectic group, so that every linear canonical transformation is dual to a unitary on the unique quantum Hilbert space — quantum optics' squeezing operators and the wave-mechanical Fourier transform being two faces of the same projective $Sp(2n)$ action. Putting these together, the uniqueness theorem read backward builds toward algebraic quantum field theory: its breakdown in infinitely many degrees of freedom is the foundational reason the Weyl algebra and its states, rather than any single Hilbert space, must be taken as primary, a story that appears again in 12.14.01 and in every inequivalent-vacuum phenomenon of curved-spacetime and thermal field theory. The bridge is one theorem with two careers: forward, the certificate that there is one quantum mechanics; backward, the birth certificate of the algebraic stance.

Full proof set Master

Proposition (central character determines the representation up to multiplicity). Let $\rho$ be a regular unitary representation of $H_{2n+1}$ on a separable Hilbert space $\mathcal{K}$ with central character $e^{i\hslash t}$, $\hslash \ne 0$. Then $\rho$ is unitarily equivalent to a direct sum of copies of the Schrödinger representation at $\hslash$, with the number of copies (finite or countably infinite) a complete invariant.

Proof. Normalise $\hslash =1$ without loss, by rescaling phase-space coordinates. Form the Gaussian projection $P=\frac{1}{(2\pi {)}^n}\int e^{-\frac{1}{4}|v{|}^2}W(v)dv$ as in the Key theorem, now on the possibly-reducible $\mathcal{K}$. The same Weyl-relation computation shows $P$ is an orthogonal projection and $PW(v)P=e^{-\frac{1}{4}|v{|}^2}P$ for all $v\in {\mathbb{R}}^{2n}$. Let ${\mathcal{K}}_0:=P\mathcal{K}$ be its range, with orthonormal basis $\{{\Omega }_j{\}}_{j\in J}$, where $J$ is finite or countable because $\mathcal{K}$ is separable.

For each $j$, set ${\mathcal{K}}_j:=\overline{\mathrm{span}}\{W(v){\Omega }_j:v\in {\mathbb{R}}^{2n}\}$, the cyclic subspace generated by ${\Omega }_j$. The matrix-coefficient identity $⟨{\Omega }_j,W(v){\Omega }_j⟩=e^{-\frac{1}{4}|v{|}^2}$ is independent of $j$, so by the construction of the Key-theorem proof each ${\mathcal{K}}_j$ carries a copy of the Schrödinger representation: the map $W(v){\Omega }_j\mapsto W_{\mathrm{S}}(v){\Omega }_{\mathrm{S}}$ extends to a unitary intertwiner ${\mathcal{K}}_j\to L^2({\mathbb{R}}^n)$.

The subspaces ${\mathcal{K}}_j$ are mutually orthogonal and $W$-invariant: for $j\ne k$, $⟨W(v){\Omega }_j,W(v^{\prime }){\Omega }_k⟩=⟨{\Omega }_j,W(v{)}^{\ast }W(v^{\prime }){\Omega }_k⟩$, and inserting $P$ via the rank computation shows this pairing factors through $⟨{\Omega }_j,{\Omega }_k⟩=0$. Their orthogonal direct sum exhausts $\mathcal{K}$: any vector orthogonal to every ${\mathcal{K}}_j$ is in particular orthogonal to $P\mathcal{K}$, hence killed by $P$; but $P$ is built so that $P\ne 0$ on every non-zero $W$-invariant subspace (the Gaussian integral of a non-zero positive operator is non-zero), forcing the orthogonal complement to be $\{0\}$. Therefore $\mathcal{K}={⨁}_{j\in J}{\mathcal{K}}_j\cong {⨁}_{j\in J}{L}^2({\mathbb{R}}^n)$, a direct sum of $|J|$ copies of the Schrödinger representation.

Finally $|J|=\dim {\mathcal{K}}_0$ is a unitary invariant: any intertwiner between two such representations restricts to a unitary between their Gaussian-projection ranges, since $P$ is constructed intrinsically from the representation. Hence the multiplicity $|J|$ is a complete invariant, and the irreducible case $|J|=1$ is the Stone–von Neumann theorem proper. $\square$

Corollary. Two regular unitary representations of $H_{2n+1}$ with the same nonzero central character are unitarily equivalent if and only if they have the same multiplicity. In particular every irreducible such representation is the Schrödinger representation, and quantization — the demand for an irreducible representation at the physical $\hslash$ — has a unique answer for a finite system.

Connections Master

• CCR algebra, Weyl algebra, and quasi-free states 12.14.01. Direct sibling and the infinite-dimensional sequel. This unit's Weyl operators $W(a,b)$ are the finite-dimensional case of the Weyl algebra over a symplectic vector space, and the Stone–von Neumann uniqueness proved here is exactly the statement that fails for the infinite-dimensional Weyl algebra of 12.14.01. Where this unit ends — one irreducible per nonzero $\hslash$ — the sibling begins: uncountably many inequivalent representations, classified by quasi-free states and Bogoliubov transformations. The two units are one theorem read in opposite directions, and together they explain why finite quantum mechanics is rigid while quantum field theory needs the algebraic framework.

• Operators, observables, and hermiticity 12.02.02. Substrate. Stone's theorem on one-parameter strongly continuous unitary groups, developed there, is the analytic engine that turns the regularity hypothesis of the Heisenberg-group representation into self-adjoint generators $\hat{q},\hat{p}$ with $[\hat{q},\hat{p}]=i\hslash 1$. The Weyl-exponentiated form $W(a,b)=e^{i(a\hat{q}+b\hat{p})}$ used throughout this unit is precisely the device by which the unbounded canonical operators of 12.02.02 become the bounded group elements on which Stone–von Neumann is stated.

• Quantum harmonic oscillator 12.04.02. The distinguished vector in the Stone–von Neumann proof — the rank-one range of the Gaussian projection — is the oscillator ground state. The Bargmann–Fock holomorphic model of the oscillator is the third coordinatisation of the unique irreducible, and the Bargmann transform is the explicit intertwiner from the wavefunction picture to the oscillator's number-state picture. The oscillator is therefore not a separate example but the canonical witness that the Schrödinger and Fock pictures are unitarily equivalent.

• Mackey theory of induced representations and systems of imprimitivity 07.07.07. The conceptual home of "quantization as choosing the unique irreducible." Mackey's classification of the unitary dual realises the Stone–von Neumann theorem as the statement that the nonzero-central-charge coadjoint orbit of $H_{2n+1}$ carries a single irreducible, while the zero-charge point-orbits give the classical characters. The imprimitivity theorem is the general machine of which the Heisenberg-group uniqueness is the cleanest special case, and it supplies the orbit-method language in which quantization is the passage from classical characters to the quantum irreducible.

• The metaplectic and geometric-quantization circle (lateral, chapter 05-symplectic/11-geometric-quantization). The projective action of $Sp(2n,\mathbb{R})$ on the unique Schrödinger representation — the metaplectic representation — is a direct corollary of Stone–von Neumann: each linear canonical transformation, being a symplectic automorphism, must be implemented by a unitary unique up to phase. The Fourier transform and the squeezing operators are special metaplectic elements. The half-density, Maslov-class, and polarisation refinements that make the projective phase precise are the substance of geometric quantization, which takes the present unit's uniqueness theorem as the linear model it then curves and globalises.

Historical & philosophical context Master

The exponentiated form of the canonical commutation relations was introduced by Hermann Weyl in 1927 ^{[Weyl 1927]} in Quantenmechanik und Gruppentheorie (Zeitschrift für Physik 46, 1), where he proposed packaging position and momentum into the relation $U(a)V(b)=e^{iab}V(b)U(a)$ for one-parameter unitary groups, sidestepping the domain pathologies of the unbounded operators $\hat{q},\hat{p}$ themselves. Weyl's reformulation is what makes the group $H_{2n+1}$ — and hence the representation-theoretic question — visible: the relation is the multiplication law of a central extension of phase space. The kinematic uniqueness this suggests was proved by John von Neumann in 1931 ^{[von Neumann 1931]} in Die Eindeutigkeit der Schrödingerschen Operatoren (Mathematische Annalen 104, 570), combining with Marshall Stone's 1930 characterisation ^{[Stone 1930]} of one-parameter unitary groups to give the result now called the Stone–von Neumann theorem. The theorem settled a foundational question of the late 1920s — whether Schrödinger's wave mechanics and Heisenberg's matrix mechanics were genuinely the same theory. They are: both are the unique irreducible representation of the Heisenberg group at a fixed $\hslash$, related by the Fourier–Bargmann intertwiners, so there is one quantum mechanics of a finite system, not several competing ones.

The group-theoretic reading was deepened by George Mackey in the 1940s–70s ^{[Mackey 1976]}, whose theory of systems of imprimitivity in The Theory of Unitary Group Representations (Chicago, 1976) exhibited Stone–von Neumann as a special case of the classification of the unitary dual of a group with a normal abelian subgroup — placing quantization inside the orbit method, where the Schrödinger representation sits over the unique nonzero coadjoint orbit. Valentine Bargmann's 1961 holomorphic model ^{[Bargmann 1961]} supplied the third standard coordinatisation and the explicit transform realising the equivalence with the oscillator's Fock space, and Gerald Folland's 1989 Harmonic Analysis in Phase Space ^{[Folland 1989]} became the definitive modern synthesis of the whole circle of ideas. The philosophical pivot came with the recognition, sharpened by Rudolf Haag's 1955 theorem ^{[Haag 1955]} in On quantum field theories (Det. Kgl. Danske Vidensk. Selsk. Mat.-fys. Medd. 29, no. 12), that the uniqueness is special to finitely many degrees of freedom: a field admits uncountably many inequivalent representations of the same Weyl relations, so "the Hilbert space" of a quantum field is not canonically given. Woit's 2017 textbook ^{[Woit 2017]} reorganises introductory quantum theory around exactly this realisation, making the Heisenberg group and its unique representation the spine from which both ordinary quantum mechanics and the algebraic stance of field theory descend.

Bibliography Master

Primary literature, originating papers (1927–1955):

• Weyl, H., Z. Phys. 46, 1 (1927). Quantenmechanik und Gruppentheorie — the exponentiated canonical commutation relations and the Heisenberg-group packaging.
• Stone, M. H., Proc. Natl. Acad. Sci. USA 16, 172 (1930); Ann. Math. 33, 643 (1932). One-parameter unitary groups and their self-adjoint generators.
• von Neumann, J., Math. Ann. 104, 570 (1931). Die Eindeutigkeit der Schrödingerschen Operatoren — the uniqueness theorem.
• Haag, R., Det. Kgl. Danske Vidensk. Selsk. Mat.-fys. Medd. 29, no. 12 (1955). On quantum field theories — failure of uniqueness in infinitely many degrees of freedom.

Primary literature, structural developments (1961–1976):

• Bargmann, V., Comm. Pure Appl. Math. 14, 187 (1961). On a Hilbert space of analytic functions and an associated integral transform — the Bargmann–Fock model and the Bargmann transform.
• Mackey, G. W., The Theory of Unitary Group Representations (University of Chicago Press, 1976). Ch. 3 — systems of imprimitivity and the unitary dual of the Heisenberg group.

Canonical monographs and textbooks:

• Folland, G. B., Harmonic Analysis in Phase Space, Annals of Mathematics Studies 122 (Princeton University Press, 1989). Ch. 1 — Heisenberg group, Schrödinger representation, Stone–von Neumann theorem, Fourier/Bargmann intertwiners, metaplectic representation.
• Hall, B. C., Quantum Theory for Mathematicians, GTM 267 (Springer, 2013). Ch. 14 — Stone–von Neumann theorem with a complete proof via the Bargmann transform.
• Woit, P., Quantum Theory, Groups and Representations: An Introduction (Springer, 2017). Ch. 13–14, 20–23 — Heisenberg group, Schrödinger representation, Stone–von Neumann as quantization, metaplectic representation.

Physics-side reference:

• Tong, D., Quantum Field Theory, DAMTP Cambridge lecture notes (raw/pdfs/qft/qft.pdf). §2 canonical quantisation and the position/momentum operators. [Have.]