From action-angle coordinates to geometric quantization: a round trip

Eva Miranda

The philosophy of geometric quantization is to find and understand a “(one-way)

dictionary” that “translates” classical systems into quantum systems . In this way,
a quantum system is associated to a classical system in which observables (smooth
functions) become operators of a Hilbert space and the classical Poisson bracket
becomes the commutator of operators. In this process, the choice of additional
geometric structures (polarizations) plays an important rôle. A desired property
is that the quantization obtained does not depend on the polarization. Another
rule in the game is that of keeping track of the symmetries on both sides. This is
the deep link of geometric quantization with representation theory. The quanti-
zation commutes with reduction “principle” becomes realistic in some geometric
quantization set-ups.
Our point of view in this big endeavour is very modest. We plan to construct a
“representation space” in the case the polarization is given by a real polarization.
For this, we follow the definition of Kostant of the representation spaces via higher
cohomology groups with coefficients in the sheaf of flat sections. In this short
note, we will not discuss either the (pre)Hilbert structure of this space nor the
quantization rules.

1. Quantization via real polarizations

2n
Let (M , ω) be a symplectic manifold such that [ω] is integral. Under these
circumstances (see for instance [14] or [6]), there exists a complex line bundle L
with a connection ∇ over M such that curv(∇) = ω. The symplectic manifold
(M 2n , ω) is called prequantizable and the pair (L, ∇) is called a prequantum line
bundle of (M 2n , ω). In order to construct the representation space we need to
restrict the space of sections to a subspace of sections which are flat in “priv-
iledged” directions given by a polarization. In this note we will just consider a
real polarization. A real polarization P is a foliation whose leaves are Lagrangian
submanifolds. Integrable systems provide natural examples of real polarizations.
If the manifold M is compact the “moment map”: F : M 2n → Rn has singularities
that correspond to equilibria. Consider the following:
Example 1.1. Consider M = S 1 ×R and ω = dt∧dθ. Take as L the trivial bundle
∂
with connection 1-form Θ = tdθ. Now, let P =< ∂θ > then flat sections satisfy,
itθ
∇X σ = X(σ) − i < θ, X > σ. Thus σ(t, θ) = a(t).e and Bohr-Sommerfeld leaves
are given by the condition t = 2πk, k ∈ Z.
This example shows that flat sections are not globally defined but they exist
along a subset of leaves of the polarization. These are called Bohr-Sommerfeld
leaves. The characterization of Bohr-Sommerfeld leaves for regular fibrations un-
der some conditions is a well-known result by Guillemin and Sternberg ([4]). In
particular the set of Bohr-Sommerfeld leaves is discrete and is given by “action”
coordinates.
1
Theorem 1.1 (Guillemin-Sternberg). If the polarization is a regular fibration
with compact leaves over a simply connected base B, then the Bohr-Sommerfeld
set is discrete and assuming that the zero-fiber is a Bohr-Sommerfeld leaf, the
Bohr-Sommerfeld set is given by, BS = {p ∈ M, (f1 (p), . . . , fn (p)) ∈ Zn } where
f1 , . . . , fn are global action coordinates on B.
This result connects with Arnold-Liouville-Mineur theorem for action-angle co-
ordinates for integrable systems. When we consider a toric manifolds the base B
may be identified with the image of the moment map by the toric action (Delzant
polytope).
In view of the previous theorem, it would make sense to “quantize” these sys-
tems counting Bohr-Sommerfeld leaves. When the polarization is an integrable
system with global action-angle coordinates, Bohr-Sommerfeld leaves are just “in-
tegral” Liouville tori. But why? Following the idea of Kostant [7], in the case
there are no global sections denote by J the sheaf of flat sections
L along the polar-
ization, we can then define the quantization as Q(M ) = k≥0 H k (M, J ). Then
quantization is given by precisely the following theorem of Sniatycki [13]:
Theorem 1.2 (Sniatycki). If the leaf space B n is a Hausdorff manifold and the
natural projection π : M 2n → B n is a fibration with compact fibres, then all the
cohomology groups vanish except for degree half of the dimension of the manifold.
Furthermore, Q(M 2n ) = H n (M 2n , J ), and the dimension of H n (M 2n , J ) is the
number of Bohr-Sommerfeld leaves.
There are two different approaches to compute this sheaf cohomology:
(1) Using a fine resolution of the complex: Namely, we can define the sheaf:
ΩiP (U ) = Γ(U, ∧i P). and C to be the sheaf of complex-valued functions
that are locally constant along P. Consider the natural (fine) resolution
i d d d d
0 → C → Ω0P →
P
Ω1P →
P
Ω1P →
P
Ω2P →
P
···
The differential operator dP is the restriction of the exterior differential
to the directions of the distributions (as in foliated cohomology). We can
use this resolution to obtain a fine resolution of J by twisting the previous
resolution with the sheaf J .
(2) A different approach used in [2] and [5] is the one of Čech cohomology
which turns out to be useful when we consider integrable systems with
singularities.

1.1. Applications to the general case of Lagrangian foliations. This fine

resolution approach can be useful to compute this geometric quantization for reg-
ular foliations (including those not coming from integrable systems like irrational
slope on the torus).
In [9] we use the classification of foliations on the torus (Kneser-Denjoy-Schwartz
theorem) together with basic properties of this sheaf cohomology to compute the
geometric quantization of a torus. In the case of irrational slope we can compute
2
the quantization (see [9]) and we obtain that the quantization space is always infi-
nite dimensional. However, if we compute the limit case of the foliated cohomology
(ω = 0), we obtain that this foliated cohomology is finite dimensional if the irra-
tionality measure of η and is infinite dimensional if the irrationality measure of η
is infinite. The results contained in [9] seem to generalize a result of El Kacimi [8]
for foliated cohomology.
Most computations in [9] rely on what we call “geometric quantization compu-
tation kit” (essentially a Künneth formula and a Mayer-Vietoris theorem in this
context). This Künneth formula is very helpful to extend results to higher dimen-
sion by reduction to the 2-dimensional case (whenever the corresponding theorem
for reduction also holds within the category of foliations considered).

2. Quantization using singular action-angle coordinates

Consider the case of rotations of the sphere. There are two leaves of the po-
larization which are singular and correspond to fixed points of the action. What
happens if we go to the edges and vertexes of Delzant’s polytope? This case and,
more generally, that of toric manifolds was considered by Mark Hamilton in [2].
Theorem 2.1 (Hamilton). For a 2n-dimensional compact toric manifold and
let n ∼
L BSr be the set of regular Bohr-Sommerfeld leaves, Q(M ) = H (M ; J ) =
l∈BSr C

Then this geometric quantization does not see the singular elliptic points. In
the example of the sphere Bohr-Sommerfeld leaves are given by integer values of
height (or, equivalently) leaves which divide out the manifold in integer areas.
In order to consider more general singularities, we need to review some results
for normal forms of integrable system. The theorem of Guillemin-Marle-Sternberg
gives normal forms in a neighbourhood of fixed points of a toric action. This can
be generalized to normal forms of integrable systems (not always toric) that we
call non-degenerate. A proof of this theorem in the elliptic case can be found in
[1]. For the other cases see the author’s thesis [10] where the idea of symplectic
orthogonal decomposition is used and the paper [12].
Theorem 2.2 (Eliasson-Miranda). There exists symplectic Morse normal forms
for integrable systems with non-degenerate singularities.
Pk
The local model is given by N = Dk × Tk × D2(n−k) and ω = i=1 dpi ∧ dθi +
Pn−k
i=1 dxi ∧ dyi . and the components of the moment map are:
(1) Regular fi = pi for i = 1, ..., k;
(2) Elliptic fi = x2i + yi2 for i = k + 1, ..., ke ;
(3) Hyperbolic fi = xi yi for i = ke + 1, ..., ke + kh ;
(4) focus-focus fi = xi yi+1 − xi+1 yi , fi+1 = xi yi + xi+1 yi+1 for i = ke + kh +
2j − 1, j = 1, ..., kf .
We can use these models to compute geometric quantization in these cases. In
the case of non-degenerate singularities in dimension 2 (only elliptic and hyperbolic
singularities), we [5] obtain the following:
3
Theorem 2.3 (Hamilton and Miranda). The quantization of a compact surface
endowed with an integrable system with non-degenerate singularities is given by,
M M
Q(M ) = H 1 (M ; J ) ∼
= (CN ⊕ CN ) ⊕ C,
p∈H l∈BSr

where H is the set of hyperbolic singularities.

In particular, this theorem shows that this quantization depends strongly on
the polarization (for more details see [5]).
2.1. New directions. The case of general non-degenerate singularities in higher
dimensions is a joint work of the author with Romero Solha and uses the above-
mentioned “geometric quantization computation kit” together with the results in
[11] and [10]. For these singular real polarizations a “quantization commutes with
reduction” principle seems to hold.
Finally, we have learned from the symplectic case that action-angle coordinates
are useful to compute geometric quantization. We can use the existence of (par-
tial) action-angle coordinates for Poisson manifolds (recently explored in [3]) to
compute geometric quantization in the Poisson context. This is a joint project
with Mark Hamilton.

References
[1] L. H. Eliasson, “Normal forms for Hamiltonian systems with Poisson commuting integrals—
elliptic case,” Comment. Math. Helv. 65 (1990), no. 1, 4–35.
[2] M. Hamilton, Locally toric manifolds and singular Bohr-Sommerfeld leaves, Mem. Amer.
Math. Soc. 207 (2010), no. 971, vi+60 pp. ISBN: 978-0-8218-4714-5.
[3] C. Laurent-Gengoux, E. Miranda and P. Vanhaecke , Action-angle coordinates for integrable
systems on Poisson manifolds, appeared in IMRN online, July 13, 2010.
[4] V. Guillemin and S. Sternberg, The Gel’fand-Cetlin system and quantization of the complex
flag manifolds, J. Funct. Anal., 52(1):106–128, 1983.
[5] M. Hamilton and E. Miranda, Geometric quantization of integrable systems with hyperbolic
singularities, Annales de l’Institut Fourier, vol. 60 (1) p. 51–85, 2010.
[6] Kirillov, Geometric quantization, Dynamical systems. IV. Symplectic geometry and its ap-
plications, Encycl. Math. Sci. 4, 137-172 (1990); translation from Itogi Nauki Tekh., Ser.
Sovrem. Probl. Mat., Fundam. Napravleniya 4, 141-178 (1985).
[7] B. Kostant, “On the Definition of Quantization,” Géométrie Symplectique et Physique
Mathématique, Coll. CNRS, No. 237, Paris, (1975), 187-210.
[8] El Kacimi-Alaoui, Sur la cohomologie feuilletée, Compositio Math. 49 (1983), no. 2,195–215.
[9] E. Miranda and F. Presas, Lagrangian foliations and geometric quantization via sheaves,
preprint, 2010.
[10] E. Miranda, On symplectic linearization of singular Lagrangian foliations, Ph.D. Thesis,
Univ. de Barcelona, 2003.
[11] E. Miranda, San Vu Ngoc, A singular Poincaré lemma, IMRN, 1, 27-46, (2005).
[12] E. Miranda, N.T. Zung, Equivariant normal form for nondegenerate singular orbits of
integrable Hamiltonian systems, Ann. Sci. Ecole Norm. Sup.,37 (2004), no. 6, 819–839.
[13] J. Śniatycki, On Cohomology Groups Appearing in Geometric Quantization, Differential
Geometric Methods in Mathematical Physics (1975).
[14] N.M.J. Woodhouse, Geometric quantization, second edition. Oxford Mathematical Mono-
graphs, Oxford Science Publications, The Clarendon Press, Oxford University Press, New
York, 1992.
4