05.11.01 · symplectic / geometric-quantization

Prequantum line bundle and the integrality condition

shipped3 tiersLean: none

Anchor (Master): Souriau 1970 Structure des Systèmes Dynamiques Ch. V (originator); Kostant 1970 Quantization and Unitary Representations LNM 170 (originator); Weil 1958 Introduction à l'étude des variétés kählériennes (integrality lemma); Woodhouse 1992 Geometric Quantization Ch. 8; Kirillov EMS Vol 4 Part II Ch. 1

Intuition Beginner

A symplectic manifold is the geometric stage on which classical mechanics is written: positions and momenta paired by an antisymmetric form $ω$ that tracks the area swept out by phase trajectories. Quantum mechanics replaces classical observables — smooth functions on phase space — with operators on a Hilbert space, and the relation $[\overset{q}{^}, \overset{p}{^}] = i ℏ$ says the operator algebra remembers $ω$ . The puzzle of geometric quantisation is to read the operators off the symplectic geometry without first writing down a Hilbert space by hand. The first step is to build a bundle, the prequantum line bundle, whose curvature stores $ω$ at every point.

The bundle does not always exist. There is a single global obstruction: the area integrated over any closed surface in phase space, divided by $2 π ℏ$ , has to be a whole number. This is the integrality condition. Heuristically: when you walk a quantum state around a closed loop in phase space, the wavefunction picks up a phase equal to $exp (i \oint ω /ℏ)$ , and that phase has to come back to itself, so the area enclosed has to be an integer multiple of $2 π ℏ$ . The same condition appears in Dirac's argument for magnetic-monopole quantisation, in Bohr-Sommerfeld quantisation rules, and in the discreteness of spin: a sphere of total area $4 π ℏ j$ is prequantisable iff $2 j$ is an integer.

Once the bundle exists, it is essentially unique — the remaining choice is a flat $U (1)$ -bundle, which parametrises a torus of possibilities. The bundle is much too big to be the quantum Hilbert space directly: its space of smooth sections is infinite-dimensional in the wrong way, with no preferred ground state. The fix is to choose a polarisation — a half-dimensional integrable distribution — and keep only the sections that are constant along the polarisation. That is the subject of the next unit. This unit ships the bundle.

Visual Beginner

A phase-space manifold $M$ drawn as a wavy 2-dimensional sheet at the bottom of the figure, with a vertical fibre rising above each point. The collection of fibres is the prequantum line bundle $L$ . The condition that ties the fibres together coherently — so that walking around any closed loop returns you to where you started — is the integrality condition $[ω / (2 π ℏ)] \in H^{2} (M; Z)$ written across the top of the figure.

The two structures in the picture are coupled: the curvature of the connection on $L$ recovers $ω$ on the base, so the bundle is not free data — it is forced by the symplectic geometry, when the integrality gate opens.

Worked example Beginner

Take the sphere $M = S^{2}$ with the round area form scaled to total area $A$ . Set $ℏ = 1$ for concreteness. The single condition to check is whether $A / (2 π)$ is a whole number.

Step 1. The integral cohomology of $S^{2}$ is $H^{2} (S^{2}; Z) = Z$ , generated by the class with total integral 1. So a class in $H^{2} (S^{2}; R)$ is integral iff its total integral over $S^{2}$ is an integer.

Step 2. For the area- $A$ form $ω$ , the cohomology class $[ω / (2 π)]$ has total integral $A / (2 π)$ . This is an integer iff $A = 2 π k$ for some $k \in Z$ .

Step 3. For $k = 1$ , the prequantum line bundle is the hyperplane bundle $O (1)$ on $CP^{1} ≅ S^{2}$ — the line bundle whose space of holomorphic sections is the two-dimensional space of homogeneous linear polynomials on $C^{2}$ . This is the spin-1/2 representation.

Step 4. For $k = 2$ , the bundle is $O (2)$ , with three-dimensional space of holomorphic sections (homogeneous quadratics) — the spin-1 representation. In general, area $2 π k$ gives $O (k)$ with $(k + 1)$ -dimensional section space, the spin- $k /2$ representation.

What this tells us: the discreteness of quantum spin is not put in by hand; it falls out of the integrality of $[ω / (2 π)]$ on $S^{2}$ . The sphere of half-integer spin $j$ has total area $4 π j = 2 π (2 j)$ , integral exactly when $2 j$ is a whole number — recovering Pauli's original spin-1/2 doublet from the $k = 1$ hyperplane bundle.

Check your understanding Beginner

Exercise (easy, multiple choice).

The prequantum line bundle over $(M, ω)$ stores the symplectic form $ω$ in which way?

A. As the total area of $M$ . B. As the curvature of a connection on the bundle. C. As the volume form of $M$ pulled back to the bundle. D. As the boundary of the bundle.

Hint

The bundle has a connection; what is the geometric quantity built from that connection?

Answer

B. The prequantum line bundle carries a connection whose curvature reproduces $ω$ (up to a fixed factor involving $ℏ$ ). Curvature measures how parallel transport along a closed loop fails to return a vector to itself; the prequantum condition is that this failure is exactly the symplectic area enclosed. Feedback-correct: curvature stores the symplectic form. Feedback-wrong: $ω$ is not the boundary of the bundle, nor a pulled-back volume form.

Formal definition Intermediate+

Let $(M, ω)$ be a symplectic manifold and fix $ℏ > 0$ . A prequantum line bundle is a triple $(L, \nabla, h)$ where:

$L \to M$ is a smooth complex line bundle;
$h$ is a Hermitian metric on $L$ ;
$\nabla$ is a connection on $L$ compatible with $h$ — that is, $d h (s, t) = h (\nabla s, t) + h (s, \nabla t)$ for all smooth sections $s, t$ ;
the curvature of $\nabla$ satisfies $F^{\nabla} = - \frac{i}{ℏ} ω .$

Equivalently, working with the imaginary-valued connection $1$ -form $A$ in a local trivialisation, $\nabla = d + A$ , the curvature $2$ -form is $F^{\nabla} = d A = - iω /ℏ$ . The first Chern class of $L$ is then $c_{1} (L) = [ω / (2 π ℏ)] \in H^{2} (M; Z)$ , by the Chern-Weil representation of $c_{1}$ as the curvature of any compatible connection divided by $2 π i$ .

Notation conventions

Throughout this unit we adopt Woodhouse-Kostant conventions ^{[Woodhouse 1992; Kostant 1970]}:

$(M, ω)$ is a symplectic manifold of dimension $2 n$ with closed non-degenerate $2$ -form $ω$ .
$ℏ > 0$ is a fixed real parameter; setting $ℏ = 1$ recovers the conventions of Souriau 1970.
$L \to M$ is a smooth complex line bundle with Hermitian metric $h$ and Hermitian connection $\nabla$ .
$X_{f}$ is the Hamiltonian vector field of $f \in C^{\infty} (M)$ : $ι_{X_{f}} ω = df$ .
${f, g} = ω (X_{f}, X_{g})$ is the Poisson bracket.
The de Rham comparison $H^{*} (M; Z) \otimes R ≅ H_{dR}^{*} (M)$ is used implicitly; a real class is integral if it lies in the image of $H^{*} (M; Z) \to H_{dR}^{*} (M)$ .

Integrality condition

The de Rham class $[ω / (2 π ℏ)] \in H_{dR}^{2} (M; R)$ is integral if it lies in the image of the natural map $H^{2} (M; Z) \to H_{dR}^{2} (M; R)$ from singular cohomology to de Rham cohomology. By the de Rham theorem, this happens iff $\int_{Σ} ω / (2 π ℏ) \in Z$ for every closed oriented $2$ -cycle $Σ \subset M$ — equivalently, for every integral $2$ -cycle generator.

Counterexamples to common slips

Integrality is a condition on cohomology, not on $ω$ pointwise. A symplectic form can be smoothly modified to a non-cohomologous form $ω^{'} = ω + d α$ without changing prequantisability: only $[ω]$ matters. Two cohomologous symplectic forms have the same prequantum bundles.
$ℏ$ is part of the data. Scaling $ℏ \mapsto λ ℏ$ rescales $[ω / (2 π ℏ)]$ by $1/ λ$ ; a manifold prequantisable at one value of $ℏ$ may fail to be at another. This is the geometric face of "the classical limit": as $ℏ \to 0$ from a sequence of prequantisable values, the Chern numbers diverge.
Existence is a topological obstruction, not a smoothness obstruction. When the integrality holds, the bundle exists smoothly with a smooth connection; no analytic refinement is required. The existence proof is via Čech cocycles, not via PDE methods.
The bundle is unique only up to the flat torus $H^{1} (M; U (1))$ . Two prequantum line bundles $(L_{1}, \nabla_{1})$ and $(L_{2}, \nabla_{2})$ with the same curvature differ by a flat $U (1)$ -bundle, parametrised by $H^{1} (M; U (1))$ . For simply-connected $M$ — e.g., $S^{2}, CP^{n}$ — this group vanishes and the prequantum bundle is unique.

Key theorem with proof Intermediate+

Theorem (Souriau-Kostant-Weil prequantisation). Let $(M, ω)$ be a symplectic manifold and $ℏ > 0$ . A Hermitian prequantum line bundle $(L, \nabla, h) \to M$ with $F^{\nabla} = - iω /ℏ$ exists if and only if the de Rham class $[ω / (2 π ℏ)] \in H_{dR}^{2} (M; R)$ is integral. When the integrality condition holds, the set of isomorphism classes of prequantum line bundles is a torsor over $H^{1} (M; U (1))$ ^{[Souriau 1970; Kostant 1970; Weil 1958]}.

Proof. The proof has two halves: necessity from Chern-Weil, and sufficiency by Čech construction. The moduli statement follows from the exponential exact sequence.

Necessity. Suppose $(L, \nabla, h)$ is a prequantum line bundle. By the Chern-Weil representation of $c_{1}$ , the first Chern class is computed from the curvature as $$ c_1(L) = \left[\frac{F^\nabla}{2\pi i}\right] = \left[\frac{-i\omega/\hbar}{2\pi i}\right] = -\left[\frac{\omega}{2\pi\hbar}\right] \in H^2(M; \mathbb{R}). $$ The image of $c_{1} (L)$ under the natural map $H^{2} (M; Z) \to H^{2} (M; R)$ recovers the de Rham class of the curvature divided by $2 π i$ up to sign. (The sign convention $F^{\nabla} = - iω /ℏ$ matches Woodhouse-Kostant; reversing it gives $c_{1} (L) = [ω / (2 π ℏ)]$ directly, used by Souriau.) Either way, $[ω / (2 π ℏ)]$ lies in the image of $H^{2} (M; Z) \to H_{dR}^{2} (M)$ , so it is integral.

Sufficiency. Suppose $[ω / (2 π ℏ)] = c$ for some integral class $c \in H^{2} (M; Z)$ . Choose a good cover ${U_{α}}$ of $M$ — one in which every finite intersection $U_{α_{1}} \cap \dots \cap U_{α_{k}}$ is either empty or contractible. On each $U_{α}$ , the form $ω /ℏ$ is exact: pick $A_{α} \in Ω^{1} (U_{α})$ with $d A_{α} = ω /ℏ$ . On double overlaps $U_{α β} = U_{α} \cap U_{β}$ , the difference $A_{α} - A_{β}$ is closed, hence exact: $A_{α} - A_{β} = d ϕ_{α β}$ for some $ϕ_{α β} \in C^{\infty} (U_{α β}; R)$ . On triple overlaps $U_{α β γ}$ , the function $ϕ_{α β} + ϕ_{β γ} - ϕ_{α γ}$ is closed under $d$ , hence locally constant on a contractible piece — that is, a constant $c_{α β γ} \in R$ . The Čech cochain $(c_{α β γ})$ represents the de Rham class $[ω / (2 π ℏ)]$ up to the factor $2 π$ , which the integrality hypothesis says lifts to an integer-valued Čech cocycle.

Define transition functions $g_{α β} = exp (2 π i ϕ_{α β}) : U_{α β} \to U (1)$ . The triple-overlap defect is $$ g_{\alpha\beta}, g_{\beta\gamma}, g_{\gamma\alpha} = \exp\bigl(2\pi i (\phi_{\alpha\beta} + \phi_{\beta\gamma} - \phi_{\alpha\gamma})\bigr) = \exp(2\pi i c_{\alpha\beta\gamma}) = 1, $$ the last equality holding because $c_{α β γ} \in Z$ by integrality. So $(g_{α β})$ is a genuine $U (1)$ -valued Čech cocycle, defining a Hermitian line bundle $L \to M$ . The local connection forms $- 2 π i A_{α}$ patch on overlaps via $A_{α} = A_{β} + d ϕ_{α β}$ , which is exactly the transformation rule for a connection $1$ -form under the gauge change $g_{α β} = exp (2 π i ϕ_{α β})$ . Setting $\nabla ∣_{U_{α}} = d - 2 π i A_{α}$ — equivalently $d - iω A_{α} \cdot 2 π / (ℏ \cdot 2 π) = d + (rescaled connection)$ — gives a globally defined Hermitian connection on $L$ whose curvature is $F^{\nabla} = - 2 π i \cdot ω /ℏ \cdot 1/ (2 π) = - iω /ℏ$ , as required.

Moduli. Suppose $(L, \nabla)$ and $(L^{'}, \nabla^{'})$ are two prequantum line bundles. Their tensor product $L \otimes (L^{'})^{- 1}$ carries the tensor-product connection, whose curvature is $F^{\nabla} - F^{\nabla^{'}} = 0$ . So $L \otimes (L^{'})^{- 1}$ is a flat Hermitian line bundle, classified by $H^{1} (M; U (1))$ via the holonomy representation $π_{1} (M) \to U (1)$ . Conversely, tensoring any prequantum bundle with a flat $U (1)$ -bundle yields another prequantum bundle. The set of isomorphism classes of prequantum line bundles is therefore a torsor over $H^{1} (M; U (1))$ (a principal homogeneous space — choose any one prequantum bundle, and the rest are obtained by tensoring with flat bundles). $□$

Bridge. The integrality theorem turns the existence of a prequantum line bundle into a single global topological condition, separating the symplectic-geometric input — which manifolds admit prequantum bundles — from the bundle-theoretic output. The structure that organises this is the exponential sheaf sequence $0 \to Z \to O_{R} \to U (1) \to 0$ , whose long exact cohomology sequence inserts the Chern-class map $c_{1} : H^{1} (M; U (1)) \to H^{2} (M; Z)$ between the moduli group of flat $U (1)$ -bundles and the integral classes that the prequantum bundle realises. The integrality condition is the geometric face of this exact-sequence position. The construction generalises in three directions explored downstream: replacing $U (1)$ by a general structure group gives the Kostant-Weil cocycle classification of $G$ -prequantisations (e.g. metaplectic over symplectic); replacing the line bundle by a half-density bundle gives the metaplectic correction needed to make geometric quantisation independent of polarisation; and replacing the symplectic base by a Poisson manifold opens the symplectic-groupoid prequantisation of Weinstein and Xu.

Exercises Intermediate+

Exercise 5 (medium, short-answer).

State the prequantisation rule $f \mapsto \hat{f} = - i ℏ \nabla_{X_{f}} + f$ and verify the commutator identity $[\hat{f}, \overset{g}{^}] = - i ℏ {f, g}$ on smooth sections of $L$ .

Hint

Use $[\nabla_{X}, \nabla_{Y}] = \nabla_{[X, Y]} + F^{\nabla} (X, Y)$ , the bracket identity $X_{{f, g}} = [X_{f}, X_{g}]$ , and $F^{\nabla} (X_{f}, X_{g}) = - iω (X_{f}, X_{g}) /ℏ = - i {f, g} /ℏ$ .

Answer

Write out the commutator on a section $s$ : $[\hat{f}, \overset{g}{^}] s = (- i ℏ)^{2} [\nabla_{X_{f}}, \nabla_{X_{g}}] s + (- i ℏ) [\nabla_{X_{f}}, g] s + (- i ℏ) [f, \nabla_{X_{g}}] s + [f, g] s .$ The last term vanishes ( $f, g$ scalars). The middle two combine: $[\nabla_{X_{f}}, g] s = X_{f} (g) s = {f, g} s$ by the definition of the Lie derivative and Hamiltonian flow, and similarly $[f, \nabla_{X_{g}}] s = - {g, f} s = {f, g} s$ . So the middle contributes $(- i ℏ) \cdot 2 {f, g} s$ . The first term uses the curvature identity: $[\nabla_{X_{f}}, \nabla_{X_{g}}] = \nabla_{[X_{f}, X_{g}]} + F^{\nabla} (X_{f}, X_{g}) = \nabla_{X_{{f, g}}} - i {f, g} /ℏ,$ multiplied by $(- i ℏ)^{2} = - ℏ^{2}$ , giving $- ℏ^{2} \nabla_{X_{{f, g}}} + i ℏ {f, g}$ . Combining: $[\hat{f}, \overset{g}{^}] s = - ℏ^{2} \nabla_{X_{{f, g}}} s + i ℏ {f, g} s + (- 2 i ℏ) {f, g} s = - ℏ^{2} \nabla_{X_{{f, g}}} s - i ℏ {f, g} s = - i ℏ (- i ℏ \nabla_{X_{{f, g}}} + {f, g}) s = - i ℏ {f, g} s$ . Rubric: full credit for the curvature identity and the cancellation pattern.

Exercise 6 (medium, symbolic).

On $S^{2} ≅ CP^{1}$ with the Fubini-Study form $ω_{F S}$ scaled to have total area $2 π k$ , identify the prequantum line bundle as a tensor power of a standard bundle on $CP^{1}$ . What is its space of holomorphic sections, and what dimension does this space have?

Hint

The hyperplane bundle $O (1)$ on $CP^{1}$ has $c_{1} = 1$ . Holomorphic sections of $O (k)$ are degree- $k$ homogeneous polynomials.

Answer

$L_{k} = O (1)^{\otimes k} = O (k)$ , the $k$ -th tensor power of the hyperplane bundle. Its space of holomorphic sections $H^{0} (CP^{1}, O (k))$ is the space of homogeneous polynomials of degree $k$ in two variables $z_{0}, z_{1}$ , with basis ${z_{0}^{k}, z_{0}^{k - 1} z_{1}, \dots, z_{1}^{k}}$ — dimension $k + 1$ . As a representation of $SU (2)$ (acting on $C^{2}$ and hence on polynomials in $z_{0}, z_{1}$ ), this is the irreducible spin- $k /2$ representation. The construction is the simplest worked instance of the Borel-Weil theorem. Rubric: full credit for identifying $O (k)$ and computing the section-space dimension.

Exercise 8 (hard, symbolic).

The Dirac quantisation of magnetic charge follows the same integrality pattern. A magnetic monopole of charge $g$ at the origin of $R^{3} ∖ {0}$ generates a $2$ -form $F = g ω_{S^{2}} / (4 π)$ on $S^{2}$ . State the Dirac quantisation condition on $g$ in terms of an integrality requirement on a $U (1)$ -bundle, and connect to the prequantisation framework of this unit.

Hint

The total flux $\int_{S^{2}} F = g$ . A $U (1)$ -bundle over $S^{2}$ with curvature $F$ exists iff $[F / (2 π ℏ)] \in H^{2} (S^{2}; Z) = Z$ .

Answer

The magnetic field of a monopole of charge $g$ at the origin is $B = (g /4 π r^{2}) \overset{r}{^}$ on $R^{3} ∖ {0}$ . On a sphere of any radius, the flux is $\int_{S^{2}} B \cdot d A = g$ . The $U (1)$ -bundle realising this connection — equivalently, the charged wavefunction of a particle of charge $e$ — exists globally on $R^{3} ∖ {0}$ iff $[F / (2 π ℏ/ e)] \in H^{2} (S^{2}; Z)$ , since $H^{2} (R^{3} ∖ {0}) = H^{2} (S^{2}) = Z$ . This gives $e g / (2 π ℏ) \in Z$ , i.e., $e g / (2 π ℏ)$ is a non-negative integer (Dirac 1931). The connection to prequantisation: replacing the symplectic base $(M, ω) = (S^{2}, F)$ and reading off the integrality condition exactly recovers the Dirac monopole rule, with $ω$ playing the role of the magnetic field. The prequantum line bundle is the monopole wavefunction bundle. Rubric: full credit for the integrality argument and the identification of $F$ as a prequantum-style curvature.

Exercise 9 (hard, short-answer).

Suppose $(M, ω)$ is prequantisable and $π : \tilde{M} \to M$ is a finite cover of degree $n$ . Show that $(\tilde{M}, π^{*} ω)$ is prequantisable, and describe the relation between the prequantum bundles on $M$ and $\tilde{M}$ . What happens to $H^{1} (\tilde{M}; U (1))$ relative to $H^{1} (M; U (1))$ ?

Hint

Pullback of integral cohomology is integral. The covering map induces a homomorphism $π_{1} (\tilde{M}) ↪ π_{1} (M)$ of finite index.

Answer

If $[ω / (2 π ℏ)] \in H^{2} (M; Z)$ , then $π^{*} [ω / (2 π ℏ)] = [π^{*} ω / (2 π ℏ)] \in H^{2} (\tilde{M}; Z)$ since pullback preserves integrality of cohomology classes. Hence $(\tilde{M}, π^{*} ω)$ is prequantisable. The prequantum bundle on $\tilde{M}$ is the pullback $π^{*} L$ of the prequantum bundle on $M$ . The moduli space: $H^{1} (\tilde{M}; U (1)) = Hom (π_{1} (\tilde{M}), U (1))$ , where $π_{1} (\tilde{M})$ embeds as a finite-index subgroup of $π_{1} (M)$ — so the moduli space on $\tilde{M}$ is larger than that on $M$ (after pullback to $\tilde{M}$ , additional flat $U (1)$ -bundles appear, parametrised by the cokernel of $π^{*} : H^{1} (M; U (1)) \to H^{1} (\tilde{M}; U (1))$ ). Concretely, going to a finite cover does not break prequantisability and can introduce new gauge degrees of freedom. Rubric: full credit for the pullback argument and the moduli observation.

Exercise 10 (hard, symbolic).

Compute the prequantum representation $\hat{f}$ on $R^{2 n}$ for $f (q, p) = p_{1}$ (the first momentum coordinate). What operator does $\overset{p}{^}_{1}$ become? Compare with the canonical Schrödinger representation $p_{1} \mapsto - i ℏ \partial / \partial q_{1}$ .

Hint

Use the product prequantum bundle with connection $\nabla = d - i θ /ℏ$ where $θ = \sum_{i} p_{i} d q_{i}$ . The Hamiltonian vector field of $p_{1}$ is $X_{p_{1}} = - \partial / \partial q_{1}$ .

Answer

$X_{p_{1}}$ is defined by $ι_{X_{p_{1}}} ω_{0} = d p_{1}$ , so with $ω_{0} = \sum_{i} d q_{i} \land d p_{i}$ , we get $X_{p_{1}} = - \partial / \partial q_{1}$ (sign depends on convention). Hence $\overset{p}{^}_{1} = - i ℏ \nabla_{X_{p_{1}}} + p_{1} = - i ℏ (- \partial / \partial q_{1} - i θ (X_{p_{1}}) /ℏ) + p_{1} = i ℏ \partial / \partial q_{1} - θ (- \partial / \partial q_{1}) + p_{1}$ . Since $θ = \sum_{i} p_{i} d q_{i}$ and $θ (- \partial / \partial q_{1}) = - p_{1}$ , this gives $\overset{p}{^}_{1} = i ℏ \partial / \partial q_{1} + p_{1} + p_{1} = i ℏ \partial / \partial q_{1} + 2 p_{1}$ . The prequantum operator differs from the Schrödinger operator $- i ℏ \partial / \partial q_{1}$ by both a sign and an additive $2 p_{1}$ term — the prequantum representation acts on the full section space $Γ (L) = C^{\infty} (R^{2 n})$ , not on the polarised half. Polarisation (next unit) cuts the section space down to functions of $q$ alone, and on this subspace the prequantum operator restricts to the standard Schrödinger operator (up to gauge). Rubric: full credit for working through the connection terms and noting the polarisation step needed to reach Schrödinger.

Advanced results Master

Existence via the Bockstein and the exponential sequence. The integrality condition has a clean cohomological packaging via the exponential sheaf sequence of abelian groups over $M$ : $$ 0 \to \mathbb{Z} \to \mathbb{R} \to U(1) \to 0, $$ where the second map is $x \mapsto exp (2 π i x)$ . Taking sheaf cohomology yields the long exact sequence $$ \cdots \to H^1(M; \mathbb{R}) \to H^1(M; U(1)) \to H^2(M; \mathbb{Z}) \to H^2(M; \mathbb{R}) \to \cdots, $$ in which the connecting map $H^{1} (M; U (1)) \to H^{2} (M; Z)$ is the first Chern class of a flat $U (1)$ -bundle, and the map $H^{2} (M; Z) \to H^{2} (M; R) ≅ H_{dR}^{2} (M)$ realises the integral lift of a real class. The integrality of $[ω / (2 π ℏ)]$ is exactly the condition that it lies in the image of the last map. The kernel — the difference between two prequantum bundles with the same curvature — is the image of $H^{1} (M; U (1))$ , recovering the moduli statement of the main theorem.

Connection to the Bohr-Sommerfeld quantisation rule. Old-quantum-theory Bohr-Sommerfeld rules quantise the classical action integrals $\oint_{γ} p d q$ to integer multiples of Planck's constant. On a Lagrangian torus $T \subset M$ with $ω ∣_{T} = 0$ , the cohomology class $[θ ∣_{T} / (2 π ℏ)] \in H^{1} (T; R)$ — where $θ$ is a local symplectic potential — must lift to $H^{1} (T; Z)$ for the prequantum line bundle to admit a horizontal section over $T$ . This is the same integrality, restricted to the Lagrangian fibre: the prequantum bundle has covariantly-constant sections over $T$ iff the holonomy around every loop in $T$ vanishes, iff every period $\oint_{γ} θ$ is in $2 π ℏ Z$ . Bohr-Sommerfeld is the integrality condition fibre-by-fibre on a Lagrangian fibration ^{[Souriau 1970; Kostant 1970]}.

Prequantum-bundle moduli on Kähler manifolds. When $M$ is a compact Kähler manifold and $ω$ is the Kähler form, prequantisability becomes a Hodge-theoretic condition: $[ω / (2 π ℏ)] \in H^{2} (M; Z) \cap H^{1, 1} (M)$ , the integral $(1, 1)$ -classes. By the Lefschetz $(1, 1)$ -theorem, every integral $(1, 1)$ -class is the first Chern class of a holomorphic line bundle, and the prequantum bundle inherits a holomorphic structure. The space $H^{0} (M, L)$ of holomorphic sections becomes the natural candidate for the quantum Hilbert space — this is the Kähler polarisation, and the section-space dimension is computed by Hirzebruch-Riemann-Roch. The Kodaira embedding theorem then guarantees that for $ω$ positive, sufficiently large tensor powers $L^{\otimes k}$ embed $M$ into projective space, giving Kostant's projective-embedding-via-prequantisation construction.

Prequantum line bundles over coadjoint orbits. For $G$ a compact connected Lie group and $O_{λ} \subset g^{*}$ a coadjoint orbit through $λ$ with the Kirillov-Kostant-Souriau form, $(O_{λ}, ω_{K K S})$ is prequantisable iff $λ$ is an integral element of $g^{*}$ (i.e., $λ$ exponentiates to a character of the maximal torus). The prequantum bundle is $G \times_{T} C_{λ}$ , the associated line bundle to the principal $T$ -bundle $G \to G / T$ and the one-dimensional $T$ -representation of weight $λ$ . The holomorphic sections form the irreducible representation of highest weight $λ$ by the Borel-Weil theorem. This is the Kirillov orbit method in its starting form: representations of $G$ are constructed by quantising integral coadjoint orbits. The spin-coadjoint case [05.11.02] is the worked instance for $G = SU (2)$ — every integer spin $j$ corresponds to an integral coadjoint orbit, and the corresponding prequantum bundle has holomorphic sections forming the $(2 j + 1)$ -dimensional spin representation.

Heuristic of holonomy and the Aharonov-Bohm phase. The Aharonov-Bohm phase $exp (i \oint A /ℏ)$ acquired by a charged particle traversing a closed loop in a magnetic-vector-potential field is the holonomy of the prequantum connection along the loop. For the phase to be well-defined globally (independent of the choice of local primitive $A$ for the field strength $F = d A$ ), the holonomy around contractible loops must vanish modulo $2 π$ — which is the integrality of $\int_{Σ} F / (2 π ℏ)$ over the disc bounded by the loop. The same global-consistency requirement that forced Dirac's monopole quantisation forces the prequantum integrality. This connection makes geometric quantisation more than a pretty formalism: it identifies a specific topological condition that quantum mechanics imposes on classical phase space.

Synthesis. The prequantum line bundle is the first geometric object in a chain that ends with the quantum Hilbert space, and the integrality condition is the gate that decides whether the chain can start at all. The condition organises three distinct strands. Topologically, integrality is the integral-cohomology lift of a de Rham class, controlled by the exponential sheaf sequence. Physically, integrality is the well-definedness of holonomy, recovering Dirac, Bohr-Sommerfeld, and Aharonov-Bohm as instances of the same principle. Representation-theoretically, integrality on a coadjoint orbit picks out the integral weights of a Lie group and pairs them with prequantum bundles whose holomorphic sections are the irreducible representations — the Kirillov orbit method. The construction continues in three downstream directions. Choosing a polarisation cuts $Γ (L)$ down to the half-density / holomorphic / Bohr-Sommerfeld sections, producing the quantum Hilbert space [05.11.02]. Adding the metaplectic correction makes the quantisation independent of polarisation choice. Extending from line bundles to higher principal bundles produces gauge-theoretic prequantisation, used in Chern-Simons theory and Yang-Mills geometric quantisation. Each step is constrained by the original integrality condition: the integral coadjoint orbits, the integral Kähler classes, the integral Bohr-Sommerfeld tori — all are images of the same arithmetic gate.

Full proof set Master

Proposition (uniqueness up to flat torus). Let $(L, \nabla)$ and $(L^{'}, \nabla^{'})$ be two prequantum line bundles over $(M, ω)$ with the same curvature $F^{\nabla} = F^{\nabla^{'}} = - iω /ℏ$ . Then $L \otimes (L^{'})^{- 1}$ is a flat Hermitian line bundle, classified by an element of $H^{1} (M; U (1))$ , and conversely tensoring $(L, \nabla)$ with any flat $U (1)$ -bundle produces another prequantum line bundle.

Proof. The dual bundle $(L^{'})^{- 1}$ has connection $- \nabla^{'}$ (acting on the dual sections), and the tensor-product connection on $L \otimes (L^{'})^{- 1}$ is $\nabla \otimes 1 + 1 \otimes (- \nabla^{'})$ . The curvature is additive on tensor products: $F^{\nabla \otimes (- \nabla^{'})} = F^{\nabla} - F^{\nabla^{'}} = 0$ . Hence $L \otimes (L^{'})^{- 1}$ is flat. Flat Hermitian line bundles over $M$ are classified by their holonomy representation $π_{1} (M) \to U (1)$ , which is parametrised by $Hom (π_{1} (M), U (1)) = H^{1} (M; U (1))$ . Conversely, given any flat $U (1)$ -bundle $F$ with connection $\nabla_{F}$ , the tensor product $L \otimes F$ with connection $\nabla \otimes 1 + 1 \otimes \nabla_{F}$ has curvature $F^{\nabla} + 0 = - iω /ℏ$ , so it is again prequantum. $□$

Proposition (prequantum representation is a Lie-algebra homomorphism). The map $f \mapsto \hat{f} := - i ℏ \nabla_{X_{f}} + f$ from $C^{\infty} (M)$ to operators on $Γ (L)$ satisfies $[\hat{f}, \overset{g}{^}] = - i ℏ {f, g}$ for all $f, g \in C^{\infty} (M)$ .

Proof. Compute the commutator on a section $s$ piece by piece. The four terms of $[\hat{f}, \overset{g}{^}]$ split as $$ [\hat f, \hat g] = (-i\hbar)^2 [\nabla_{X_f}, \nabla_{X_g}] + (-i\hbar)\bigl([\nabla_{X_f}, g] + [f, \nabla_{X_g}]\bigr) + [f, g]. $$ The last term vanishes (scalar multiplications commute). The middle two contribute $[\nabla_{X_{f}}, g] s = X_{f} (g) \cdot s = {f, g} s$ , and $[f, \nabla_{X_{g}}] s = - X_{g} (f) s = - {g, f} s = {f, g} s$ . Summed, the middle gives $(- i ℏ) (2 {f, g}) s$ . For the first term, the curvature identity for connections on line bundles reads $[\nabla_{X}, \nabla_{Y}] = \nabla_{[X, Y]} + F^{\nabla} (X, Y)$ acting on sections. Substituting $X = X_{f}$ , $Y = X_{g}$ and using $[X_{f}, X_{g}] = X_{{f, g}}$ (Hamiltonian flows generate a Poisson-bracket Lie subalgebra of vector fields) plus $F^{\nabla} (X_{f}, X_{g}) = - iω (X_{f}, X_{g}) /ℏ = - i {f, g} /ℏ$ : $$ [\nabla_{X_f}, \nabla_{X_g}] = \nabla_{X_{{f, g}}} - i{f, g}/\hbar. $$ Multiplying by $(- i ℏ)^{2} = - ℏ^{2}$ : $- ℏ^{2} \nabla_{X_{{f, g}}} + i ℏ {f, g}$ . Combine all three contributions: $$ [\hat f, \hat g] s = \bigl(-\hbar^2 \nabla_{X_{{f, g}}} + i\hbar{f, g}\bigr) s + (-i\hbar)(2{f, g}) s = -\hbar^2 \nabla_{X_{{f, g}}},s - i\hbar,{f, g},s. $$ Factor: $- i ℏ (- i ℏ \nabla_{X_{{f, g}}} s + {f, g} s) = - i ℏ {f, g} s$ . $□$

Theorem (Borel-Weil for the prequantum bundle on $O_{λ}$ ). Let $G$ be a compact connected Lie group with maximal torus $T$ and dominant integral weight $\lambda \in \mathfrak{t}^ \cap \Lambda^{\mathrm{wt}} $. T h eco a d j o in t or bi t$ \mathcal{O}\lambda = G \cdot \lambda \subset \mathfrak{g}^ $w i t h t h eK i r i l l o v - K os t an t - S o u r ia u sy m pl ec t i c f or mi s p r e q u an t i s ab l e, t h e p r e q u an t u m l in e b u n d l e i s$ L_\lambda = G \times_T \mathbb{C}_\lambda $, an d t h es p a ceo f h o l o m or p hi csec t i o n s i s t h e i r r e d u c ib l e$ G $- r e p r ese n t a t i o n o f hi g h es tw e i g h t$ \lambda$.*

Proof. The coadjoint orbit $O_{λ}$ identifies with $G / T$ (the flag variety) by sending $g λ \mapsto g T$ , since the stabiliser of $λ$ in $G$ is $T$ when $λ$ is regular (and a parabolic when $λ$ is degenerate; the argument adapts). The KKS form on $O_{λ}$ pulled back to $G / T$ is given at the identity coset by $ω_{e} (X, Y) = ⟨ λ, [X, Y]⟩$ for $X, Y \in g / t$ .

The integral class $[ω_{K K S} / (2 π)] \in H^{2} (G / T; Z)$ pulls back from the integral character of $T$ corresponding to $λ$ . Concretely, the line bundle $L_{λ} = G \times_{T} C_{λ}$ associated to the principal $T$ -bundle $G \to G / T$ and the $1$ -dimensional weight- $λ$ $T$ -representation has Chern class $c_{1} (L_{λ}) = [ω_{K K S} / (2 π)]$ . (See Bott-Tu §23 for the topological calculation, or Kostant 1970 for the direct construction.) Hence integrality holds, and $L_{λ}$ is the prequantum bundle.

The Borel-Weil theorem now identifies $H^{0} (G / T, L_{λ})$ with the irreducible $G$ -representation $V_{λ}$ of highest weight $λ$ , via the action of $G$ by left translation on sections ^{[Kirillov EMS Vol 4 Part II Ch. 1]}. The full proof is in $[05.11.02]$ for the case $G = SU (2)$ — the proof here is the general $G$ extension. $□$

Connections Master

Symplectic manifold 05.01.02. This unit takes a symplectic manifold $(M, ω)$ and asks the prequantisability question. Symplectic forms with non-integral periods — abundant on Kähler tori, on general moduli spaces, on coadjoint orbits through non-integral weights — fail the integrality gate and produce no prequantum bundle. The unit therefore selects a proper subclass of symplectic manifolds: the integral ones.
Coadjoint orbit 05.03.01. Every integral coadjoint orbit $O_{λ}$ of a compact Lie group is prequantisable, with the prequantum bundle realising the irreducible representation of highest weight $λ$ via Borel-Weil. The orbit method of Kirillov is built on this identification: representations of compact groups are the same as integral coadjoint orbits prequantised. The detailed worked $SU (2)$ case is in the next unit.
Hermitian metric and Chern connection 03.05.20. The prequantum line bundle is a Hermitian holomorphic line bundle (when the Kähler polarisation is chosen), and its Chern connection is the prequantum connection. The Chern-connection unit supplies the bundle-with-connection machinery; this unit supplies the constraint on which manifolds admit such structures.
Prequantisation of the spin coadjoint orbit 05.11.02. The next unit in this chapter is the prototypical worked example: $G = SU (2)$ , $O_{j} = S_{r}^{2}$ with $r = j$ , integrality $2 j \in Z$ , prequantum bundle $O (2 j)$ on $CP^{1}$ , holomorphic sections forming the spin- $j$ representation. The general theorem ships here; the working example ships there.
Hamiltonian vector field 05.02.01. The prequantum representation $\hat{f} = - i ℏ \nabla_{X_{f}} + f$ refers to the Hamiltonian vector field $X_{f}$ of $f$ . The commutator identity $[\hat{f}, \overset{g}{^}] = - i ℏ {f, g}$ is the bridge from classical Hamiltonian dynamics on $(M, ω)$ to the quantum operator algebra.
CCR algebra and Weyl algebra 12.14.01. The prequantum representation realises the canonical commutation relation $[\overset{q}{^}, \overset{p}{^}] = i ℏ$ as the operator-algebra image of the Poisson bracket ${q, p} = - 1$ . The algebraic-QFT framework treats the prequantum bundle as the geometric origin of the CCR before any Hilbert-space representation is chosen. The integrality condition in the QFT context is the Dirac-Kostant generalisation of the magnetic-charge quantisation rule.

Historical & philosophical context Master

Jean-Marie Souriau introduced the prequantum line bundle and the integrality condition in 1966 in the foundational paper Quantification géométrique in Communications in Mathematical Physics ^{[Souriau 1966]}, with the full framework developed in his 1970 monograph Structure des Systèmes Dynamiques ^{[Souriau 1970]}. Souriau's formulation emphasised the topological character of the condition: only symplectic manifolds with integral cohomology class admit prequantum bundles, and this constraint is the geometric face of the discreteness of quantum spectra.

Bertram Kostant independently arrived at the construction in his 1970 Lecture Notes in Mathematics 170 paper Quantization and Unitary Representations ^{[Kostant 1970]}, in the framework of representation theory of Lie groups. Kostant's formulation embedded the prequantum bundle into the orbit-method programme: integral coadjoint orbits of a compact group are prequantisable, and the resulting bundles realise irreducible representations via Borel-Weil. The Kostant and Souriau papers together established geometric quantisation as a discipline; the integrality condition is sometimes called the Kostant-Souriau condition in the modern literature.

The integrality lemma itself — that a closed $2$ -form on a smooth manifold is the curvature of a Hermitian line bundle iff its de Rham class is integral — predates the application to quantisation by over a decade. André Weil proved the version for compact Kähler manifolds in his 1958 Introduction à l'étude des variétés kählériennes ^{[Weil 1958]}; the smooth-manifold version follows the same Čech-de Rham argument used here. The Bohr-Sommerfeld rules of the old quantum theory — Bohr 1913, Sommerfeld 1916 — are the integrality condition fibre-by-fibre on a Lagrangian fibration, recognised as such by Souriau and Kostant simultaneously. Dirac's 1931 monopole-quantisation argument, in the Proceedings of the Royal Society A 133, 60–72, is the same condition applied to $M = S^{2}$ with the magnetic-field curvature.

Bibliography Master

@book{Souriau1970,
  author    = {Souriau, Jean-Marie},
  title     = {Structure des Syst\`emes Dynamiques},
  publisher = {Dunod},
  address   = {Paris},
  year      = {1970},
  note      = {English translation: \emph{Structure of Dynamical Systems: A Symplectic View of Physics}, Birkh\"auser Progress in Mathematics 149, 1997},
}

@incollection{Kostant1970,
  author    = {Kostant, Bertram},
  title     = {Quantization and unitary representations},
  booktitle = {Lectures in Modern Analysis and Applications III},
  series    = {Lecture Notes in Mathematics},
  volume    = {170},
  publisher = {Springer},
  year      = {1970},
  pages     = {87--208},
}

@article{Souriau1966,
  author    = {Souriau, Jean-Marie},
  title     = {Quantification g\'eom\'etrique},
  journal   = {Communications in Mathematical Physics},
  volume    = {1},
  year      = {1966},
  pages     = {374--398},
}

@book{Weil1958,
  author    = {Weil, Andr\'e},
  title     = {Introduction \`a l'\'etude des vari\'et\'es k\"ahl\'eriennes},
  publisher = {Hermann},
  address   = {Paris},
  year      = {1958},
}

@book{Woodhouse1992,
  author    = {Woodhouse, Nicholas M. J.},
  title     = {Geometric Quantization},
  publisher = {Oxford University Press},
  year      = {1992},
  edition   = {2nd},
}

@book{BatesWeinstein1997,
  author    = {Bates, Sean and Weinstein, Alan},
  title     = {Lectures on the Geometry of Quantization},
  series    = {Berkeley Mathematics Lecture Notes},
  volume    = {8},
  publisher = {American Mathematical Society},
  year      = {1997},
}

@article{Dirac1931,
  author    = {Dirac, Paul A. M.},
  title     = {Quantised singularities in the electromagnetic field},
  journal   = {Proceedings of the Royal Society A},
  volume    = {133},
  year      = {1931},
  pages     = {60--72},
}

@incollection{KirillovEMS,
  author    = {Kirillov, Alexandre A.},
  title     = {Geometric Quantization},
  booktitle = {Dynamical Systems IV: Symplectic Geometry and Its Applications},
  editor    = {Arnol'd, V. I. and Novikov, S. P.},
  series    = {Encyclopaedia of Mathematical Sciences},
  volume    = {4},
  publisher = {Springer},
  year      = {2001},
  edition   = {2nd},
}

@book{Kirillov2004,
  author    = {Kirillov, Alexandre A.},
  title     = {Lectures on the Orbit Method},
  series    = {Graduate Studies in Mathematics},
  volume    = {64},
  publisher = {American Mathematical Society},
  year      = {2004},
}

Prerequisites

05.01.02
05.03.01
03.05.20

Tier anchors

beginner: Woodhouse Geometric Quantization Ch. 8 informal; Bates-Weinstein Lectures on the Geometry of Quantization Ch. 6 informal
intermediate: Woodhouse Geometric Quantization Ch. 8; Bates-Weinstein Lectures on the Geometry of Quantization Ch. 6; Kirillov EMS Vol 4 Part II Ch. 1
master: Souriau 1970 Structure des Systèmes Dynamiques Ch. V (originator); Kostant 1970 Quantization and Unitary Representations LNM 170 (originator); Weil 1958 Introduction à l'étude des variétés kählériennes (integrality lemma); Woodhouse 1992 Geometric Quantization Ch. 8; Kirillov EMS Vol 4 Part II Ch. 1

References

Souriau 1970 — Structure des Systèmes Dynamiques · Ch. V — originator of the prequantum line bundle and the integrality condition
Kostant 1970 — Quantization and unitary representations · Lecture Notes in Mathematics 170 — independent originator of the prequantum construction
Souriau 1966 — Quantification géométrique · Comm. Math. Phys. 1, 374–398 — earliest published statement of the integrality condition
Weil 1958 — Introduction à l'étude des variétés kählériennes · Hermann §VI — integrality lemma for closed 2-forms with integral periods
Woodhouse 1992 — Geometric Quantization · Oxford University Press, 2nd ed., Ch. 8 — modern textbook exposition of prequantisation
Bates-Weinstein 1997 — Lectures on the Geometry of Quantization · Berkeley Math. Lecture Notes 8, Ch. 6 — Čech-cocycle construction of the prequantum bundle
Kirillov EMS Vol 4 — Geometric Quantization (Part II) · Ch. 1 — the orbit-method framing of the prequantum line bundle
Kostant 1970 — Quantization and unitary representations · Lecture Notes in Mathematics 170, §0 (the prequantum representation $\hat f = -i\hbar \nabla_{X_f} + f$)

Estimated time

beginner: 18m
intermediate: 45m
master: 90m