Geometric quantization of coadjoint orbits and the Borel–Weil theorem

Intuition Beginner

Take a compact symmetry group, like the rotations of space. Its Lie algebra has a dual space, and the group acts on that dual space; the orbits of this action are curved surfaces called coadjoint orbits. Each orbit comes equipped, for free, with a symplectic form — the data that turns a surface into a phase space of a classical mechanical system. The orbit is, quite literally, the classical phase space of a spinning system, and its size is set by how far the orbit sits from the origin.

Now run the geometric-quantisation machine on this phase space. The previous units built two ingredients: a prequantum line bundle, which exists only when the orbit has the right quantised size, and a polarisation, a choice of half the directions. For a coadjoint orbit the natural polarisation is a complex structure: the orbit is secretly a complex manifold, the flag manifold, and the polarised sections are its holomorphic functions twisted by the line bundle.

The punchline is striking. When you collect all those holomorphic sections, you do not get a vague infinite cloud — you get exactly one irreducible representation of the group, the one whose label matches the orbit's quantised size. This is the Borel–Weil theorem, and read through the quantisation lens it says: quantising the classical phase space of a spinning symmetry produces precisely one quantum spin state-space. The classical orbit and the quantum representation are two faces of the same integral label.

This correspondence — orbits to representations — is the orbit method of Kirillov and Kostant. It is one of the cleanest places where a classical geometric object and a quantum algebraic object line up perfectly.

Visual Beginner

A sphere sits in the dual space, centred on the origin and drawn at a quantised radius — one of a discrete stack of allowed shells. An arrow points from the sphere across to a column of evenly spaced dots, the basis states of a single quantum representation. The sphere's radius and the number of dots are locked together by one integer.

The two sides of the figure — the classical sphere with its symplectic surface form, and the discrete quantum column — are the same datum seen twice. The integer that quantises the sphere's radius is the same integer that names the representation, and the arrow between them is the geometric-quantisation construction read as the Borel–Weil theorem.

Worked example Beginner

Take the rotation group of three-dimensional space, in its compact form $\mathrm{SU}(2)$. Its Lie algebra dual is ordinary three-dimensional space, and the coadjoint orbits are spheres centred at the origin, one for each radius. A sphere of radius $j$ is the classical phase space of a spin of magnitude $j$, carrying the area form scaled to total area $4\pi j$.

Step 1. Ask which spheres are quantisable. The prequantum line bundle exists only when the symplectic area, divided by $2\pi$, is a whole number. For the radius-$j$ sphere this condition reads $2j$ is a whole number, so the allowed radii are $j=0,\frac{1}{2},1,\frac{3}{2},2,\dots$ — exactly the spin values seen in physics.

Step 2. Give the sphere its complex structure. A round sphere is the same as the Riemann sphere, a complex manifold. The polarisation is this complex structure, and the polarised sections of the line bundle are its holomorphic sections.

Step 3. Count them. The holomorphic sections of the line bundle of quantised size $2j$ on the Riemann sphere are the homogeneous polynomials of degree $2j$ in two variables. There are $2j+1$ of them, with basis the monomials from one variable to the power $2j$ down to the other.

What this tells us: the radius-$j$ sphere quantises to a space of dimension $2j+1$, and this space is the spin-$j$ representation of the rotation group. The integer labelling the sphere and the integer counting the states agree by construction. A sphere of half the allowed radius is simply not quantisable; this is why spin comes in steps of one half.

Check your understanding Beginner

Formal definition Intermediate+

Let $G$ be a compact connected Lie group with Lie algebra $\mathfrak{g}$, maximal torus $T$, and a fixed invariant inner product identifying $\mathfrak{g}\cong {\mathfrak{g}}^{\ast }$. The group acts on ${\mathfrak{g}}^{\ast }$ by the coadjoint action ${\mathrm{Ad}}_g^{\ast }$, and the orbit through $\lambda \in {\mathfrak{g}}^{\ast }$ is

$${\mathcal{O}}_{\lambda }=\{{\mathrm{Ad}}_g^{\ast }\lambda :g\in G\}\cong G/G_{\lambda },$$

where $G_{\lambda }$ is the stabiliser of $\lambda$. For $\lambda$ regular, $G_{\lambda }=T$ and ${\mathcal{O}}_{\lambda }\cong G/T$, the flag manifold. The orbit carries the Kirillov-Kostant-Souriau (KKS) symplectic form from 05.03.01,

$${\omega }_{\lambda }({\mathrm{ad}}_X^{\ast }\lambda ,{\mathrm{ad}}_Y^{\ast }\lambda )=⟨\lambda ,[X,Y]⟩,X,Y\in \mathfrak{g},$$

a closed non-degenerate $G$-invariant $2$-form making $({\mathcal{O}}_{\lambda },{\omega }_{\lambda })$ a homogeneous symplectic manifold.

Notation conventions

Throughout this unit:

• $G$ is a compact connected Lie group, $T\subset G$ a maximal torus, $\mathfrak{g}\supset \mathfrak{t}$ their Lie algebras, $W=N(T)/T$ the Weyl group.
• $\Phi \subset {\mathfrak{t}}^{\ast }$ is the root system, ${\Phi }^{+}$ a choice of positive roots, $\rho =\frac{1}{2}{\sum }_{\alpha \in {\Phi }^{+}}\alpha$ the half-sum of positive roots.
• ${\Lambda }_{\mathrm{wt}}\subset {\mathfrak{t}}^{\ast }$ is the weight lattice; $\lambda$ is dominant integral when $⟨\lambda ,{\alpha }^{\vee }⟩\in {\mathbb{Z}}_{\ge 0}$ for all $\alpha \in {\Phi }^{+}$.
• ${\mathcal{O}}_{\lambda }\cong G/T$ is the (regular) coadjoint orbit through $\lambda$, with KKS form ${\omega }_{\lambda }$.
• $L_{\lambda }\to G/T$ is the holomorphic line bundle associated to the character $e^{\lambda }$ of $T$; it is the prequantum line bundle of $({\mathcal{O}}_{\lambda },{\omega }_{\lambda })$ when $\lambda$ is integral.
• $V_{\lambda }$ is the irreducible representation of $G$ of highest weight $\lambda$; $V_{\lambda }^{\ast }$ is its dual.
• $K_{G/T}=\det T^{\ast 1,0}(G/T)$ is the canonical bundle of the flag manifold; $K_{G/T}^{1/2}$ is the half-canonical (half-form) bundle of 05.11.03.

The integral orbit and its prequantum bundle

By the prequantisation criterion of 05.11.02, the orbit $({\mathcal{O}}_{\lambda },{\omega }_{\lambda })$ admits a prequantum line bundle if and only if the cohomology class $[{\omega }_{\lambda }/2\pi ]$ is integral. For a coadjoint orbit this is the integrality (Weil) condition, and it holds precisely when $\lambda$ lies in the weight lattice ${\Lambda }_{\mathrm{wt}}$, equivalently when the character $e^{\lambda }:T\to U(1)$ is well-defined. In that case the prequantum bundle is the homogeneous line bundle

$$L_{\lambda }=G{\times }_T{\mathbb{C}}_{\lambda }⟶G/T,$$

associated to the one-dimensional $T$-representation ${\mathbb{C}}_{\lambda }$ on which $t$ acts by $e^{\lambda }(t)$.

The Kähler polarisation

The flag manifold $G/T$ carries a $G$-invariant complex structure, realised concretely as $G/T\cong G_{\mathbb{C}}/B$ where $G_{\mathbb{C}}$ is the complexification and $B$ a Borel subgroup. With respect to this complex structure ${\omega }_{\lambda }$ is a Kähler form, and the antiholomorphic tangent bundle $\mathcal{P}=T^{0,1}(G/T)$ is a $G$-invariant Kähler polarisation in the sense of 05.11.03. The polarised sections of $L_{\lambda }$ are then its holomorphic sections,

$${\Gamma }_{\mathcal{P}}(L_{\lambda })=H^0(G/T,L_{\lambda }).$$

Counterexamples to common slips

• Not every $\lambda$ quantises. A label $\lambda$ off the weight lattice gives a symplectic orbit with no prequantum bundle; the orbit is a perfectly good classical phase space but has no quantisation. Integrality is the quantisation gate.
• The orbit must be regular for $G/T$. A singular (non-regular) $\lambda$ has stabiliser strictly larger than $T$, and the orbit is a partial flag manifold $G/P$, not $G/T$; the construction still runs but with a parabolic in place of the Borel.
• Highest weight versus its dual. The naive Borel-Weil statement gives $H^0(G/T,L_{\lambda })\cong V_{\lambda }^{\ast }$, the dual of the highest-weight representation; the lowest-weight vector of the sections is the highest-weight vector of $V_{\lambda }$. Mixing up $V_{\lambda }$ and $V_{\lambda }^{\ast }$ is a frequent sign error.
• Without the half-form, the count is shifted. The plain prequantum holomorphic sections realise the weight $\lambda$; the metaplectic-corrected sections of $L_{\lambda }\otimes K_{G/T}^{1/2}$ realise the shifted weight. Whether a formula carries a $+\rho$ depends on whether the half-form correction has been included.

Key theorem with proof Intermediate+

Theorem (geometric quantisation of an integral coadjoint orbit = Borel–Weil). Let $G$ be a compact connected Lie group with maximal torus $T$, and let $\lambda$ be a dominant weight. Then:

(i) The regular coadjoint orbit $({\mathcal{O}}_{\lambda },{\omega }_{\lambda })\cong G/T$ is prequantisable, with prequantum bundle $L_{\lambda }$, if and only if $\lambda \in {\Lambda }_{\mathrm{wt}}$ is a weight.

(ii) The $G$-invariant Kähler polarisation $\mathcal{P}=T^{0,1}(G/T)$ has polarised section space the holomorphic sections $H^0(G/T,L_{\lambda })$.

(iii) (Borel–Weil) As a representation of $G$, $H^0(G/T, L_\lambda) \cong V_\lambda^$,thedualoftheirreduciblerepresentationofhighestweight$\lambda$.Inparticularthegeometricquantisationof$\mathcal{O}\lambda$istheirreduciblerepresentation$V\lambda$(uptodualising),andallothercohomologyvanishes:$H^i(G/T, L_\lambda) = 0$for$i > 0$.* ^{[Kostant 1970 §4-§5; Borel-Weil 1954; Kirillov 2004 Ch. 5]}.

Proof. (i) The orbit ${\mathcal{O}}_{\lambda }$ is prequantisable iff $[{\omega }_{\lambda }/2\pi ]\in H^2(G/T;\mathbb{Z})$ by 05.11.02. Pulling ${\omega }_{\lambda }$ back along the projection $G\to G/T$ and restricting to the torus directions, the period of ${\omega }_{\lambda }$ over the integral $2$-spheres ${\mathbb{CP}}_{\alpha }^1\subset G/T$ (one for each positive root $\alpha$) equals $⟨\lambda ,{\alpha }^{\vee }⟩$. These periods are integers for every $\alpha$ precisely when $\lambda$ pairs integrally with every coroot, which is the defining condition of the weight lattice. Hence prequantisability is equivalent to $\lambda \in {\Lambda }_{\mathrm{wt}}$, and the bundle is $L_{\lambda }=G{\times }_T{\mathbb{C}}_{\lambda }$.

(ii) With the $G$-invariant complex structure on $G/T\cong G_{\mathbb{C}}/B$, the polarisation $\mathcal{P}=T^{0,1}(G/T)$ is integrable (Newlander-Nirenberg holds since the structure is genuinely complex) and Lagrangian for ${\omega }_{\lambda }$ (a Kähler form is of type $(1,1)$, so its restriction to the antiholomorphic distribution vanishes). The polarised condition ${\nabla }_{X}s=0$ for $X\in \mathcal{P}$ is the Cauchy-Riemann equation for the Chern connection on the Hermitian holomorphic bundle $L_{\lambda }$, whose solutions are the holomorphic sections $H^0(G/T,L_{\lambda })$ by 05.11.03.

(iii) The group $G$ acts on $H^0(G/T,L_{\lambda })$ by pullback, a finite-dimensional representation since $G/T$ is compact. Restricting a holomorphic section to the base point $eB$ and using $B$-equivariance, a holomorphic section is determined by a $B$-eigenvector data: the highest-weight line. By the highest-weight theory of compact groups the space of such sections is non-zero exactly when $\lambda$ is dominant, and then it contains a unique (up to scale) lowest-weight vector of weight $-\lambda$. The $G_{\mathbb{C}}$-orbit of this vector spans an irreducible subrepresentation; a dimension count via the Weyl character formula shows the section space is exactly the irreducible $V_{\lambda }^{\ast }$ with no remainder, and the Kodaira vanishing theorem (or Bott's computation) kills $H^i$ for $i>0$. Thus $H^0(G/T,L_{\lambda })\cong V_{\lambda }^{\ast }$. $\square$

Bridge. This theorem is the bridge between the symplectic and the representation-theoretic faces of a single integer. The integrality condition of 05.11.02 is exactly the statement that $\lambda$ is a weight, so the prequantisation gate and the highest-weight label are the foundational reason the orbit method works at all; this is exactly the Borel–Weil theorem of 07.06.09 re-read as "quantisation = the irrep". The construction builds toward the full orbit method: the rule "integral orbit $\mapsto$ irreducible representation" generalises from compact groups to nilpotent groups (where Kirillov's original exponential-map version holds with no integrality obstruction) and to solvable and reductive groups (where it becomes the Kirillov-Kostant-Duflo correspondence with subtle corrections). Putting these together, the Kähler polarisation 05.11.03 is the central insight that turns the abstract orbit into a complex flag manifold whose holomorphic functions are countable and group-organised. The role of the half-form correction — quantising $L_{\lambda }\otimes K_{G/T}^{1/2}$ rather than $L_{\lambda }$ — appears again in the character formula, where it is dual to the $\rho$-shift in the Weyl denominator: the metaplectic correction is the geometric origin of the half-sum of positive roots, and this same $\rho$ recurs in the higher-cohomology Borel-Weil-Bott statement and in the Weyl dimension count computed via 03.06.20.

Exercises Intermediate+

Advanced results Master

The orbit method as a dictionary. The Borel-Weil realisation is the compact-group instance of Kirillov's orbit method, the heuristic that unitary irreducible representations of a Lie group $G$ are parametrised by integral coadjoint orbits in ${\mathfrak{g}}^{\ast }$ ^{[Kirillov 1962; Kirillov 2004]}. For a simply-connected nilpotent group the dictionary is exact and free of integrality conditions: every coadjoint orbit quantises (the exponential map is a diffeomorphism, the orbits are even-dimensional affine subspaces, and there is no topology to obstruct prequantisation), and the representation is induced from a character on the polarising subgroup. For compact groups the dictionary acquires the integrality condition $\lambda \in {\Lambda }_{\mathrm{wt}}$ and the $\rho$-shift, and the representation is realised by Borel-Weil. For general reductive and solvable groups the correspondence becomes the Kirillov-Kostant-Duflo map, with the Duflo isomorphism correcting the naive symbol calculus by the square root of the Jacobian of the exponential — a higher avatar of the same half-form correction that produces $\rho$ here.

The metaplectic correction and the half-sum of positive roots. The canonical bundle of the flag manifold is $K_{G/T}=L_{-2\rho }$, computed from the weights of the holomorphic cotangent bundle $T^{\ast 1,0}(G/T)={⨁}_{\alpha \in {\Phi }^{+}}{\mathbb{C}}_{-\alpha }$, whose determinant has weight $-{\sum }_{\alpha \in {\Phi }^{+}}\alpha =-2\rho$. The half-form bundle is therefore $K_{G/T}^{1/2}=L_{-\rho }$, and the metaplectic-corrected geometric quantisation of the orbit through $\mu$ uses sections of $L_{\mu }\otimes L_{-\rho }=L_{\mu -\rho }$. Setting $\mu =\lambda +\rho$ recovers $L_{\lambda }$ and hence $V_{\lambda }^{\ast }$. This is the precise sense in which the orbit method labels representations by orbits through $\lambda +\rho$: the $\rho$ is not a convention but the weight of the half-canonical bundle, and the metaplectic correction of 05.11.03 is the geometric source of the half-sum of positive roots that appears throughout highest-weight theory and in the Weyl character denominator ^{[Kostant 1970; Woodhouse 1992 Ch. 8]}.

Borel-Weil-Bott and the higher cohomology. Bott's 1957 theorem completes the picture for non-dominant weights: for any weight $\lambda$ with $\lambda +\rho$ regular, the line bundle $L_{\lambda }$ has cohomology concentrated in the single degree $\ell (w)$, where $w$ is the Weyl element carrying $\lambda +\rho$ to the dominant chamber, and there $H^{\ell (w)}(G/T,L_{\lambda })\cong V_{w\cdot \lambda }^{\ast }$ for the dot-action $w\cdot \lambda =w(\lambda +\rho )-\rho$; if $\lambda +\rho$ is singular all cohomology vanishes ^{[Bott 1957]}. Read through geometric quantisation, this says that a "wrong-sign" polarisation — one for which $L_{\lambda }$ has no holomorphic sections — still quantises the orbit, but the quantum states now live in a higher Dolbeault cohomology group rather than in degree zero. The Euler characteristic $\chi (G/T,L_{\lambda })={\sum }_i(-1{)}^i\dim H^i=\dim V_{\lambda }$ is polarisation-independent and equals the Weyl dimension, exactly as the index-theoretic formulation of geometric quantisation (the spin${}^c$-Dirac index) predicts; the cohomology degree is the only polarisation-dependent datum.

The worked $\mathrm{SU}(2)$ orbit in full. For $G=\mathrm{SU}(2)$ the orbit ${\mathcal{O}}_{\lambda }$ of the weight $\lambda =2j\varpi$ is the sphere $S^2$ of radius $j$, with KKS form of symplectic area giving period $2j$. Integrality forces $2j\in \mathbb{Z}$, the half-integer spin quantisation. The Kähler polarisation is the Riemann-sphere complex structure, the prequantum bundle is $\mathcal{O}(2j)$, and its holomorphic sections are the degree-$2j$ polynomials — the spin-$j$ representation of dimension $2j+1$. With the half-form correction the bundle is $\mathcal{O}(2j)\otimes \mathcal{O}(-1)=\mathcal{O}(2j-1)$, of section-space dimension $2j$, which is the spin-$(j-\frac{1}{2})$ representation: the $\rho$-shift visibly converts the integer-labelled prequantum count to the half-integer-labelled physical spin, and the orbit through $\lambda +\rho$ at radius $j+\frac{1}{2}$ is the one whose corrected quantisation is the genuine spin-$j$ state-space ^{[Bates-Weinstein 1997 Ch. 8]}.

Synthesis. The geometric quantisation of integral coadjoint orbits is the foundational reason the orbit method exists: it shows that the classical phase space attached to a symmetry — an integral coadjoint orbit with its KKS form — quantises to exactly one irreducible representation, and this is exactly the Borel–Weil theorem 07.06.09 read through the prequantisation 05.11.02 and polarisation 05.11.03 machinery. Putting these together, three integers from different worlds turn out to be one integer seen three ways: the period of the symplectic form (from 05.03.01), the highest weight $\lambda$, and the first Chern number of the prequantum bundle coincide, their common value the integrality condition. The construction generalises in two directions that the rest of the corpus develops. Upward in degree, Borel-Weil-Bott shows the quantisation is polarisation-independent at the level of the Euler characteristic, which is dual to the index-theoretic spin${}^c$-Dirac formulation and recurs in the Weyl dimension count of 03.06.20; the central insight there is that only the cohomological degree, not the dimension, depends on the choice of polarisation. Outward in generality, the orbit-to-representation dictionary generalises from compact groups to nilpotent, solvable, and reductive groups, where the half-form $\rho$-shift becomes the Duflo correction and the bridge is always the same metaplectic half-canonical bundle. The half-sum of positive roots $\rho$, which appears here as the weight of $K_{G/T}^{1/2}$, is the single thread tying the metaplectic correction, the orbit-method shift, and the Weyl denominator into one structure.

Full proof set Master

Proposition (integrality of a coadjoint orbit is the weight condition). Let $G$ be compact connected with maximal torus $T$, and $\lambda \in \mathfrak{t}^ \subset \mathfrak{g}^$aregularelement.Theorbit$(\mathcal{O}\lambda, \omega\lambda) \cong G/T$isprequantisableifandonlyif$\lambda$liesintheweightlattice$\Lambda_{\mathrm{wt}}$.

Proof. By 05.11.02 prequantisability is equivalent to $[{\omega }_{\lambda }]\in 2\pi H^2(G/T;\mathbb{Z})$. The second homology $H_2(G/T;\mathbb{Z})$ is free abelian, generated by the embedded rational curves ${\mathbb{CP}}_{\alpha }^1$ for the simple roots $\alpha$, each arising from the root $\mathrm{SU}(2{)}_{\alpha }\hookrightarrow G$. The line bundle $L_{\lambda }=G{\times }_T{\mathbb{C}}_{\lambda }$ restricts over ${\mathbb{CP}}_{\alpha }^1$ to the bundle of degree $⟨\lambda ,{\alpha }^{\vee }⟩$, since the character $e^{\lambda }{|}_{T_{\alpha }}$ has $\mathrm{SU}(2{)}_{\alpha }$-weight $⟨\lambda ,{\alpha }^{\vee }⟩$. The curvature of the prequantum connection is $-i{\omega }_{\lambda }$, so $\frac{1}{2\pi }{\int }_{{\mathbb{CP}}_{\alpha }^1}{\omega }_{\lambda }=⟨\lambda ,{\alpha }^{\vee }⟩$. These periods are integers for all $\alpha$ precisely when $\lambda$ pairs integrally with every coroot, the defining condition of ${\Lambda }_{\mathrm{wt}}$. Conversely, when $\lambda \in {\Lambda }_{\mathrm{wt}}$ the character $e^{\lambda }$ exists on $T$, the associated bundle $L_{\lambda }$ is defined, and its first Chern class is $[{\omega }_{\lambda }/2\pi ]$, giving the prequantum data. $\square$

Proposition (the Kähler polarisation is admissible). The antiholomorphic tangent bundle $\mathcal{P}=T^{0,1}(G/T)$, for the $G$-invariant complex structure of $G/T\cong G_{\mathbb{C}}/B$, is an integrable Lagrangian polarisation for ${\omega }_{\lambda }$ in the sense of 05.11.03.

Proof. Integrability of $\mathcal{P}$ is the integrability of the complex structure, which holds because $G/T\cong G_{\mathbb{C}}/B$ is a genuine complex (indeed projective) manifold; the Nijenhuis tensor vanishes and Newlander-Nirenberg applies. For the Lagrangian condition, ${\omega }_{\lambda }$ is the imaginary part of a $G$-invariant Kähler metric on $G/T$, hence a form of Hodge type $(1,1)$. A $(1,1)$-form pairs to zero on two antiholomorphic vectors: ${\omega }_{\lambda }(X,Y)=0$ for $X,Y\in T^{0,1}$, since the form has no $(0,2)$-component. Thus $\mathcal{P}$ is isotropic, and being of complex dimension equal to half the real dimension of $G/T$ it is Lagrangian. The intersection $\mathcal{P}\cap \overline{\mathcal{P}}=T^{0,1}\cap T^{1,0}=0$ is the Kähler (purely complex polarisation) case. $\square$

Proposition (Borel-Weil: holomorphic sections give the irreducible). For $\lambda$ dominant integral, $H^0(G/T, L_\lambda) \cong V_\lambda^$as$G$-representations,and$H^i(G/T, L_\lambda) = 0$for$i > 0$.*

Proof. The group $G$ acts on the finite-dimensional space $S=H^0(G/T,L_{\lambda })$ by pullback. Evaluation of a holomorphic section at the base point $o=eB$ together with the $B$-equivariance of $L_{\lambda }$ shows that a section is determined by its restriction along the $B$-orbit through $o$, which is dense (the big Bruhat cell); a holomorphic section is thus a highest-weight datum, and the space of $B$-eigenvectors of weight $-\lambda$ in $S$ is one-dimensional. By complete reducibility $S$ decomposes into irreducibles, each with a unique highest-weight line; since $S$ has a single lowest-weight line of weight $-\lambda$, it contains $V_{\lambda }^{\ast }$ with multiplicity one and no other irreducible whose extreme weight exceeds $-\lambda$. A character computation — equating the holomorphic Lefschetz number ${\sum }_i(-1{)}^i\mathrm{tr}(g\mid H^i(G/T,L_{\lambda }))$ with the Weyl character ${\chi }_{\lambda }$ via the Atiyah-Bott fixed-point formula at the isolated $T$-fixed points $wB$ — shows the alternating sum of characters is exactly ${\chi }_{V_{\lambda }^{\ast }}$. Kodaira vanishing for the positive bundle $L_{\lambda }\otimes K_{G/T}^{-1}$ (positive since $\lambda +2\rho$ is dominant regular) forces $H^i=0$ for $i>0$, so the alternating sum collapses to $H^0$, giving $H^0(G/T,L_{\lambda })\cong V_{\lambda }^{\ast }$. $\square$

Proposition ($\rho$-shift reconciliation). The half-canonical bundle of $G/T$ is $K_{G/T}^{1/2}=L_{-\rho }$, and the metaplectic-corrected quantisation of the orbit through $\lambda +\rho$ equals the Borel-Weil quantisation of $L_{\lambda }$.

Proof. The holomorphic cotangent space at $o=eB$ is ${\mathfrak{n}}^{\ast }\cong {⨁}_{\alpha \in {\Phi }^{+}}{\mathbb{C}}_{-\alpha }$ as a $T$-representation, since $T_o^{1,0}(G/T)\cong {\mathfrak{g}}_{\mathbb{C}}/\mathfrak{b}\cong {⨁}_{\alpha \in {\Phi }^{+}}{\mathfrak{g}}_{\alpha }$. The canonical bundle $K_{G/T}=\det T^{\ast 1,0}$ is the homogeneous bundle of the determinant character, of weight $-{\sum }_{\alpha \in {\Phi }^{+}}\alpha =-2\rho$, so $K_{G/T}=L_{-2\rho }$ and the square root is $K_{G/T}^{1/2}=L_{-\rho }$ (which exists as a genuine line bundle because $\rho$ is a weight for compact simply-connected $G$). The metaplectic-corrected quantisation of ${\mathcal{O}}_{\lambda +\rho }$ uses sections of $L_{\lambda +\rho }\otimes K_{G/T}^{1/2}=L_{\lambda +\rho }\otimes L_{-\rho }=L_{\lambda }$, whose holomorphic sections are $V_{\lambda }^{\ast }$ by the Borel-Weil proposition. Hence the half-form construction on the shifted orbit reproduces the same representation as the uncorrected construction on $L_{\lambda }$, and the orbit-method label $\lambda +\rho$ is the metaplectic shift of the highest weight $\lambda$. $\square$

Connections Master

• Prequantisation of the spin coadjoint orbit 05.11.02. This unit takes the prequantisation criterion proved there and reads it on a general coadjoint orbit: the orbit is prequantisable iff its label $\lambda$ is a weight, which is the integrality (Weil) condition specialised to $G/T$. The spin orbit $S^2$ treated in that unit is the $G=\mathrm{SU}(2)$ instance of the general construction here, and the prequantum bundle $\mathcal{O}(2j)$ is $L_{\lambda }$ for $\lambda =2j\varpi$.

• Polarisation, half-densities, and metaplectic correction 05.11.03. The Kähler polarisation $T^{0,1}(G/T)$ used here is the flag-manifold instance of the complex polarisation defined there; the polarised sections are holomorphic sections, and the half-form correction $K_{G/T}^{1/2}=L_{-\rho }$ is the metaplectic correction read on the orbit, supplying the half-sum of positive roots $\rho$ as the weight of the half-canonical bundle. The $\rho$-shift reconciling Borel-Weil with the Kostant count is exactly the metaplectic shift discussed in that unit.

• Coadjoint orbit and the KKS symplectic form 05.03.01. The symplectic manifold being quantised is the coadjoint orbit with its Kirillov-Kostant-Souriau form constructed there; the integrality of that form's cohomology class is the input to the present quantisation, and the period of ${\omega }_{\lambda }$ over the root spheres is the highest weight $\lambda$ paired with coroots.

• The Borel-Weil theorem 07.06.09. The representation-theoretic statement that holomorphic sections of $L_{\lambda }$ on $G/T$ realise $V_{\lambda }^{\ast }$ is supplied there; this unit re-reads it as the assertion "geometric quantisation of the integral orbit = the irreducible representation", making Borel-Weil the output of the prequantisation-polarisation pipeline rather than a standalone algebraic fact.

• Borel-Hirzebruch cohomology of G/T 03.06.20. The cohomology ring of the flag manifold and the Chern classes of the bundles $L_{\lambda }$ computed there control the Euler characteristic $\chi (G/T,L_{\lambda })=\dim V_{\lambda }$ via Borel-Weil-Bott, and the $\rho$ appearing in the geometric-quantisation half-form correction is the same $\rho$ in the Weyl denominator computed in that setting. The Weyl dimension formula is the integral of the Chern character against the Todd class of $G/T$.

Historical & philosophical context Master

The correspondence between coadjoint orbits and irreducible representations originates with Alexandre Kirillov's 1962 paper Unitary representations of nilpotent Lie groups (Uspekhi Mat. Nauk 17) ^{[Kirillov 1962]}, where the unitary dual of a simply-connected nilpotent group was shown to be in bijection with the coadjoint orbits, each representation induced from a character along a polarising subalgebra. This was the first appearance of the orbit method, and its success in the nilpotent case — where the geometry is free of integrality obstructions and the exponential map is a global diffeomorphism — motivated the search for a general dictionary. Bertram Kostant's 1970 Quantization and unitary representations (Lecture Notes in Mathematics 170, §4-§5) ^{[Kostant 1970]} recast the construction as geometric quantisation: the coadjoint orbit is a symplectic manifold, its prequantisation requires the integrality of the symplectic class, and a polarisation realises the representation as polarised sections. For compact groups this reproduces and conceptually explains the Borel–Weil theorem, announced by Armand Borel and André Weil in 1954 and disseminated through Serre's Séminaire Bourbaki exposé 100 ^{[Borel-Weil 1954]}, which realised the irreducible representation of highest weight $\lambda$ as the holomorphic sections of a line bundle on the flag manifold $G/T$.

Raoul Bott's 1957 Homogeneous vector bundles (Annals of Mathematics 66) ^{[Bott 1957]} extended the theorem to all weights, locating the irreducible representation in the unique non-vanishing cohomology degree determined by the Weyl-group dot-action — the Borel-Weil-Bott theorem — and thereby showing that geometric quantisation of an orbit is polarisation-independent at the level of the Euler characteristic even when the holomorphic-section count fails. The philosophical content of the orbit method is its claim that representation theory is a form of quantum mechanics: an irreducible representation is the quantum state-space of a classical system whose phase space is a coadjoint orbit, and the integrality of the orbit is the Bohr-Sommerfeld quantisation condition. Kirillov's later Lectures on the Orbit Method (Graduate Studies in Mathematics 64, 2004) ^{[Kirillov 2004]} codified the dictionary into a "user's guide" of correspondence rules — including the $\rho$-shift, which the geometric-quantisation viewpoint identifies as the metaplectic half-form correction — and traced its reach and its limits across nilpotent, compact, solvable, and reductive groups. Guillemin and Sternberg's 1982 Geometric quantization and multiplicities of group representations (Inventiones Mathematicae 67) ^{[Guillemin-Sternberg 1982]} embedded the orbit method into the broader $[Q,R]=0$ principle linking quantisation to symplectic reduction, situating Borel-Weil as the simplest case of a functorial correspondence.

Bibliography Master

@article{Kirillov1962,
  author    = {Kirillov, Alexandre A.},
  title     = {Unitary representations of nilpotent Lie groups},
  journal   = {Uspekhi Matematicheskikh Nauk},
  volume    = {17},
  number    = {4},
  year      = {1962},
  pages     = {57--110},
}

@incollection{Kostant1970bw,
  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},
}

@incollection{BorelWeil1954,
  author    = {Serre, Jean-Pierre},
  title     = {Repr\'esentations lin\'eaires et espaces homog\`enes k\"ahl\'eriens des groupes de Lie compacts (d'apr\`es Borel et Weil)},
  booktitle = {S\'eminaire Bourbaki},
  number    = {100},
  year      = {1954},
}

@article{Bott1957,
  author    = {Bott, Raoul},
  title     = {Homogeneous vector bundles},
  journal   = {Annals of Mathematics},
  volume    = {66},
  number    = {2},
  year      = {1957},
  pages     = {203--248},
}

@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},
}

@book{Woodhouse1992bw,
  author    = {Woodhouse, Nicholas M. J.},
  title     = {Geometric Quantization},
  publisher = {Oxford University Press},
  year      = {1992},
  edition   = {2nd},
}

@article{GuilleminSternberg1982,
  author    = {Guillemin, Victor and Sternberg, Shlomo},
  title     = {Geometric quantization and multiplicities of group representations},
  journal   = {Inventiones Mathematicae},
  volume    = {67},
  number    = {3},
  year      = {1982},
  pages     = {515--538},
}

@book{BatesWeinstein1997bw,
  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},
}