Geometric quantization

Geometric quantization is a mathematical framework that constructs quantum-mechanical data from a classical system represented by a symplectic manifold. Its basic input is a manifold (M) equipped with a closed, nondegenerate two-form (\omega), while its principal output is a Hilbert space together with operators associated with a restricted class of classical observables. The construction expresses the correspondence between classical mechanics and quantum mechanics in terms of line bundles, connections, polarizations, and representation theory.

The framework does not define a quantization map for every smooth function while preserving all algebraic relations. Instead, it separates the construction into prequantization, which represents the full Poisson algebra but produces an excessively large state space, and polarization, which reduces that state space by imposing a geometric analogue of dependence on half of the classical variables. Additional structures, including half-form bundles and pairings between polarizations, account for features not captured by prequantization alone.

Symplectic and Hamiltonian structure

A classical phase space is represented by a (2n)-dimensional symplectic manifold ((M,\omega)). Every smooth function (f\in C^\infty(M)) determines a Hamiltonian vector field (X_f) through the convention

[ \iota_{X_f}\omega=-df. ]

The associated Poisson bracket is

[ {f,g}=\omega(X_f,X_g). ]

This bracket turns (C^\infty(M)) into a Lie algebra and describes the infinitesimal action of classical observables on one another. A quantization map seeks operators (\widehat f) satisfying the corresponding commutator relation

[ [\widehat f,\widehat g] =i\hbar,\widehat{{f,g}}, ]

subject to sign conventions determined by the definitions of (X_f) and the curvature of the connection.

The Groenewold–Van Hove theorem prevents this correspondence from extending to all classical observables while also retaining the usual irreducibility and normalization conditions. Geometric quantization therefore identifies subalgebras whose Hamiltonian flows are compatible with the geometric data used to define the quantum state space.

Prequantization

Prequantization begins with a Hermitian complex line bundle

[ L\longrightarrow M ]

equipped with a compatible connection (\nabla). Its curvature is required to satisfy

[ F_\nabla=-\frac{i}{\hbar}\omega. ]

Such a bundle exists precisely when the cohomology class

[ \left[\frac{\omega}{2\pi\hbar}\right] ]

lies in the image of integral cohomology (H^2(M,\mathbb Z)) inside (H^2(M,\mathbb R)). This condition is the integrality condition for the symplectic form. It is also the line-bundle form of the relation between curvature and the first Chern class,

[ c_1(L)=\left[\frac{\omega}{2\pi\hbar}\right]. ]

The prequantum state space consists of square-integrable sections of (L), with the symplectic volume form

[ \frac{\omega^n}{n!} ]

providing the measure when the relevant integrals are finite. A classical observable acts through a first-order differential operator of the form

[ \widehat f_{\mathrm{pre}} =-i\hbar\nabla_{X_f}+f. ]

The curvature identity ensures that these operators reproduce the Poisson bracket through their commutators. Constants act by scalar multiplication, and Hamiltonian flows lift to connection-preserving transformations of the line bundle when the corresponding global conditions hold.

Prequantization retains functions of all phase-space variables and therefore differs from familiar quantum representations. For the phase space (T^*Q), prequantum wavefunctions depend on both position and momentum, whereas the Schrödinger representation uses wavefunctions depending only on position. Polarization removes this excess dependence.

Polarizations

A polarization is an integrable Lagrangian distribution in the complexified tangent bundle (T_{\mathbb C}M). If (P\subset T_{\mathbb C}M) is a polarization, then each fiber has complex dimension (n), the symplectic form vanishes on (P), and sections of (P) are closed under the Lie bracket. Polarized sections satisfy

[ \nabla_Xs=0 \qquad \text{for every }X\in P. ]

This equation restricts variation along the directions represented by (P). The resulting space of sections forms the preliminary quantum state space associated with that polarization.

A real polarization is locally tangent to a foliation by Lagrangian submanifolds. The vertical distribution on a cotangent bundle (T^*Q) is the standard example: polarized sections are constant along the momentum fibers after the prequantum connection has been locally trivialized, so the states descend to objects defined over the configuration manifold (Q). Global real polarizations can contain compact leaves with nontrivial holonomy. Covariantly constant sections then occur only on Bohr–Sommerfeld leaves, and their treatment commonly requires distributional sections or cohomology of the sheaf of polarized sections.

A complex polarization instead selects a complex Lagrangian subbundle. On a Kähler manifold, the antiholomorphic tangent distribution defines a polarization, and polarized sections correspond to holomorphic sections of the prequantum line bundle. Their inner product is determined by the Hermitian structure of the bundle and the Kähler volume form. This construction connects geometric quantization with complex geometry and the theory of positive holomorphic line bundles.

Not every Hamiltonian vector field preserves a fixed polarization. An observable (f) acts directly on polarized sections when

[ [X_f,P]\subseteq P. ]

Observables failing this condition generally require a comparison between the original polarization and its image under the Hamiltonian flow. This limitation is geometric rather than merely analytic: it reflects the dependence of the representation on the selected family of Lagrangian directions.

Half-forms and corrected state spaces

Polarized sections of the prequantum line bundle alone do not always produce the expected spectral shifts or transformation laws. The metaplectic correction modifies the state space by adjoining a square root of a polarization-dependent canonical bundle.

For a complex polarization (P), the relevant canonical line is constructed from the top exterior power of the annihilator of (P). A half-form bundle (\delta_P) satisfies

[ \delta_P^{\otimes 2}\cong K_P, ]

where (K_P) is the corresponding canonical bundle. Corrected quantum states are polarized sections of

[ L\otimes\delta_P. ]

Under a change of coordinates, the half-form factor transforms by a square root of the relevant Jacobian determinant. This supplies the density behavior needed for an invariant inner product and introduces the metaplectic group, a double cover of the symplectic group.

The correction accounts for the familiar zero-point displacement in the quantization of the harmonic oscillator. It also modifies the orbit-method correspondence by replacing an unshifted weight with the appropriate shift involving the Weyl vector. These effects arise from the transformation of half-forms rather than from an alteration of the classical symplectic form.

Pairings between differently polarized spaces are described by the Blattner–Kostant–Sternberg pairing. When defined, such a pairing compares states associated with transverse or suitably compatible polarizations and supplies operators for Hamiltonian flows that do not preserve a fixed polarization. Analytic difficulties occur when the intersection of the polarizations changes dimension or when the resulting integral kernel fails to define a bounded map.

Development

The geometric formulation emerged from several related developments in the representation theory of Lie groups and the global geometry of classical mechanics. André Weil related integral differential forms to line bundles with connection, while Irving Segal examined unitary representations arising from classical phase spaces. Jean-Marie Souriau formulated quantization using symplectic manifolds and group actions, and Bertram Kostant developed the line-bundle and polarization framework in a representation-theoretic form.

During the early 1970s, You Watanabe developed a cohomological description of polarized sections supported on Bohr–Sommerfeld leaves for compact real polarizations. This treatment placed the failure of ordinary smooth polarized sections within the sheaf cohomology of the polarization and identified the degree in which nontrivial quantized states occur. The construction became part of the analysis of real polarizations with compact leaves.

Elsewhere in the same period, Louis Auslander and Bertram Kostant related quantization to representations of solvable Lie groups. George Mackey developed the theory of induced representations that supplied a parallel algebraic description, while Alexandre Kirillov associated irreducible representations with symplectic coadjoint orbits. David Blattner and Shlomo Sternberg contributed to the comparison of polarized representations through the pairing that bears their names together with Kostant.

These developments established a common structure in which symplectic manifolds encode classical systems, integral line bundles encode the quantum phase, and polarizations encode the form of the representation.

Cotangent bundles

For a smooth manifold (Q), the cotangent bundle (T^*Q) carries the canonical one-form

[ \theta=p_i,dq^i ]

in local coordinates, with symplectic form

[ \omega=-d\theta=dq^i\wedge dp_i. ]

Because (\omega) is exact, the prequantum line bundle can be taken to be topologically trivial. A connection with local form determined by (\theta) has the required curvature.

The vertical polarization is generated locally by the vector fields (\partial/\partial p_i). Its polarized sections depend only on the position variables after the connection term has been incorporated. With the half-form correction, the quantum states transform as half-densities on (Q), which gives an inner product independent of a separately selected coordinate volume element.

Functions affine in momentum,

[ f(q,p)=a^i(q)p_i+b(q), ]

preserve the vertical polarization and therefore act directly on the polarized space. Their quantizations combine differentiation along the vector field (a^i\partial/\partial q^i), multiplication by (b), and a divergence term supplied by the half-form correction. General functions of higher polynomial degree in momentum do not preserve the polarization, in agreement with the obstruction to quantizing the complete Poisson algebra.

Coadjoint orbits

Let (G) be a Lie group with Lie algebra (\mathfrak g). Each orbit (\mathcal O\subset\mathfrak g^*) of the coadjoint action possesses the Kirillov–Kostant–Souriau form,

[ \omega_\lambda(X^#,Y^#) =\langle\lambda,[X,Y]\rangle, ]

where (X^#) and (Y^#) are the infinitesimal vector fields generated by (X,Y\in\mathfrak g). This form is closed and nondegenerate on the orbit.

When its cohomology class satisfies the integrality condition, the orbit admits a prequantum line bundle compatible with the (G)-action. A suitable polarization then produces a representation of (G), often through holomorphic sections when the orbit carries an invariant Kähler structure. For compact connected Lie groups, this construction is closely related to the Borel–Weil theorem, which realizes irreducible representations as spaces of holomorphic sections over flag manifolds.

The half-form correction changes the relation between the orbit parameter and the highest weight by the standard Weyl shift. Consequently, the geometry of the canonical bundle participates directly in the representation assigned to the orbit.

Symmetry and reduction

Suppose a Lie group (G) acts symplectically on (M) with an equivariant moment map

[ \mu:M\longrightarrow\mathfrak g^*. ]

At a regular value, the symplectic reduction

[ M/!/G=\mu^{-1}(0)/G ]

inherits a symplectic form under the usual freeness and properness conditions. When the group action lifts to the prequantum line bundle, the bundle descends to the reduced space.

The relation between quantizing the reduced space and taking invariant vectors after quantizing the original space is expressed by the principle that quantization commutes with reduction. In the compact Kähler setting, the resulting correspondence identifies the quantization of the quotient with the invariant subspace of the unreduced quantization, subject to the hypotheses required for the quotient and line bundle. This relation connects geometric quantization with equivariant index theory and geometric invariant theory.

Scope and structural limitations

Geometric quantization depends on global data that may fail to exist. The symplectic form can violate the integrality condition, a polarization can be unavailable, or a half-form bundle can encounter a topological obstruction. Even when all structures exist, different polarizations can produce spaces whose equivalence requires nontrivial analytic arguments.

The construction also distinguishes between exact operator identities and semiclassical correspondence. Prequantization represents the full Poisson algebra but is reducible and physically oversized. Polarization yields a more restricted representation, although only observables compatible with the polarization act without additional machinery. Pairings and index-theoretic formulations extend the framework without removing the Groenewold–Van Hove obstruction.

Modern formulations frequently replace the literal vector space of polarized sections with a Spin(^c) structure, a Dirac operator, or an equivariant index. In that setting, quantization is represented by a virtual vector space or a group representation. This preserves the global topological information of the line bundle while reducing dependence on singularities of a particular polarization.