Symplectic Geometry and Darboux's Theorem

Symplectic Geometry studies manifolds with closed, non-degenerate 2-form ω; foundational for Hamiltonian mechanics and classical physics
Darboux's theorem (1882): all symplectic manifolds locally look identical—near any point, ω = Σ dp_i ∧ dq_i in canonical coordinates
No local invariants exist in symplectic geometry (unlike Riemannian curvature); richness comes from global topology and constraints
Symplectic form encodes Poisson bracket structure relating positions and momenta; preserves phase space under time evolution
Focus shifts from metric properties (distances/angles) to area/volume preservation and Lagrangian submanifold behavior
Symplectic capacity and global phenomena replace local geometric invariants as key objects of study

Symplectic Geometry and Darboux's Theorem

Overview

Symplectic geometry studies manifold|manifolds equipped with a closed, non-degenerate 2-form, providing the natural mathematical framework for Hamiltonian mechanics and many areas of modern physics. Unlike Riemannian geometry, where the metric provides local invariants like curvature, symplectic geometry is remarkably rigid: Darboux's theorem establishes that all symplectic manifolds of the same dimension look locally identical. This absence of local invariants shifts attention to global topology and the emerging field of symplectic topology.

The subject emerged from the geometric reformulation of classical mechanics in the 19th century, when mathematicians recognized that Hamilton's Equations possess an intrinsic geometric structure independent of coordinate choices. The symplectic form $ω$ encodes the fundamental Poisson bracket structure of phase space, relating positions and momenta. While differential geometry traditionally focused on distances and angles via Riemannian metrics, symplectic geometry concerns itself with areas, volumes, and the preservation of phase space structure under time evolution.

Darboux's theorem, proven in 1882, stands as the fundamental local result in symplectic geometry. It asserts that near any point on a symplectic manifold, one can find coordinates in which the symplectic form takes the standard form $\sum_{i = 1}^{n} d p_{i} \land d q_{i}$ . This is simultaneously liberating—allowing free use of canonical coordinates—and constraining, as it implies symplectic geometry has no local invariants analogous to the Riemann curvature tensor. The richness of the subject therefore emerges from global phenomena, constraints on Lagrangian submanifold|Lagrangian submanifolds, and quantitative measures like symplectic capacity.

Symplectic manifolds have no local invariants: Darboux's theorem shows every symplectic manifold locally looks like $(R^{2 n}, \sum d p_{i} \land d q_{i})$ . This means the geometry is entirely global, making symplectic topology—not local differential geometry—the central focus.

Symplectic Manifold: A symplectic manifold is a pair

(M, ω)

where

M

is a smooth manifold and

ω

is a differential form|differential 2-form satisfying:

Closed: $d ω = 0$ (i.e., $ω$ is a de Rham cohomology|cocycle)
Non-degenerate: For each $p \in M$ and each non-zero $v \in T_{p} M$ , there exists $w \in T_{p} M$ such that $ω_{p} (v, w) \neq = 0$

Non-degeneracy implies $M$ has even dimension $2 n$ , and $ω$ induces an isomorphism $T_{p} M \to T_{p}^{*} M$ via $v \mapsto ι_{v} ω = ω (v, \cdot)$ .

Darboux's Theorem: For any symplectic manifold

(M, ω)

and any point

p \in M

, there exists a coordinate chart

(U, ϕ)

centered at

p

with coordinates

(q_{1}, \dots, q_{n}, p_{1}, \dots, p_{n})

such that

ω ∣_{U} = i = 1 \sum n d p_{i} \land d q_{i}

Such coordinates are called canonical or Darboux coordinates.

Core Theory

Darboux's Theorem

Theorem [1]: Let

(M, ω)

be a

2 n

-dimensional symplectic manifold and

p \in M

. Then there exists a neighborhood

U

p

and a diffeomorphism

ϕ : U \to V \subset R^{2 n}

with

ϕ (p) = 0

such that

ϕ^{*} ω_{0} = ω ∣_{U}

where

ω_{0} = \sum_{i = 1}^{n} d x_{i} \land d y_{i}

is the standard symplectic form on

R^{2 n}

. Theorem [2] (Relative Darboux): Let

(M, ω)

be symplectic and

L \subset M

a Lagrangian submanifold (i.e.,

dim L = n

and

ω ∣_{L} = 0

). For any

p \in L

, there exist Darboux coordinates

(q_{1}, \dots, q_{n}, p_{1}, \dots, p_{n})

near

p

such that

L

is locally given by

{p_{1} = \dots = p_{n} = 0}

We prove Theorem [1] using Moser's method, which relies on the Moser trick for deforming symplectic forms.

Step 1 (Linear Darboux theorem): By the linear algebra of skew-symmetric form|skew-symmetric forms, there exists a linear isomorphism

A : T_{p} M \to R^{2 n}

such that

A^{*} ω_{0} = ω_{p}

. Choose coordinates

x

on a neighborhood

U

p

such that

d x_{p} = A

In these coordinates, write $ω = ω_{0} + α$ where $α_{p} = 0$ (since both forms agree at $p$ after our choice of coordinates). We have $d ω = 0$ and $d ω_{0} = 0$ , hence $d α = 0$ .

Step 2 (Poincaré lemma near

p

): Since

α_{p} = 0

and

d α = 0

, by the Poincaré lemma on a contractible neighborhood, we can write

α = d β

for some 1-form

β

with

β_{p} = 0

. Step 3 (Moser's trick): Define a family of 2-forms

ω_{t} = ω_{0} + t α = ω_{0} + t d β

for

t \in [0, 1]

. Each

ω_{t}

is closed. For small enough

U

, each

ω_{t}

is non-degenerate (non-degeneracy is an open condition and

ω_{1} = ω

is non-degenerate).

We seek a family of diffeomorphisms $ψ_{t} : U \to U$ with $ψ_{0} = id$ such that $ψ_{t}^{*} ω_{t} = ω_{0}$ . Differentiating:

\frac{d}{d t} ψ_{t}^{*} ω_{t} = ψ_{t}^{*} (L_{X_{t}} ω_{t} + \frac{\partial ω _{t}}{\partial t}) = 0

where

X_{t} = \frac{d ψ _{t}}{d t} \circ ψ_{t}^{- 1}

is the generating vector field.

We need $L_{X_{t}} ω_{t} + d β = 0$ . By Cartan's formula, $L_{X_{t}} ω_{t} = d (ι_{X_{t}} ω_{t}) + ι_{X_{t}} d ω_{t} = d (ι_{X_{t}} ω_{t})$ .

Thus we require: $d (ι_{X_{t}} ω_{t} + β) = 0$ , or equivalently, $ι_{X_{t}} ω_{t} = - β$ .

Step 4 (Solving for

X_{t}

): Since

ω_{t}

is non-degenerate, it establishes an isomorphism

T M \to T^{*} M

, so we can uniquely solve for

X_{t}

via:

X_{t} = - ω_{t}^{- 1} (β)

where

ω_{t}^{- 1} : T^{*} M \to T M

is the inverse of

v \mapsto ι_{v} ω_{t}

Note that $X_{t} (p) = 0$ since $β_{p} = 0$ . Integrating the time-dependent vector field $X_{t}$ gives $ψ_{t}$ , which exists on a neighborhood of $p$ for $t \in [0, 1]$ .

Then $ψ_{1}^{*} ω = ψ_{1}^{*} ω_{1} = ω_{0}$ , giving us the desired Darboux coordinates. □

Common Misconception: Students often think Darboux's theorem means "symplectic geometry is trivial" because there are no local invariants. This is backwards! The absence of local invariants means all interesting phenomena are global. Unlike Riemannian geometry where local curvature dominates, symplectic geometry is inherently topological. For instance, Gromov's non-squeezing theorem shows that the standard symplectic ball cannot be symplectically embedded into a cylinder of smaller radius, despite both spaces being locally identical. Gaston Darboux (1882) — Proved that all symplectic structures are locally equivalent, establishing the foundational local triviality result that distinguishes symplectic geometry from Riemannian geometry.

Examples and Applications

1

Cotangent Bundle: Let

Q

be any

n

-dimensional manifold (configuration space). The cotangent bundle

M = T^{*} Q

has a canonical symplectic structure.

Define the tautological 1-form $θ$ on $T^{*} Q$ by: for $(q, ξ) \in T^{*} Q$ (where $q \in Q$ and $ξ \in T_{q}^{*} Q$ ) and $v \in T_{(q, ξ)} (T^{*} Q)$ ,

θ_{(q, ξ)} (v) = ξ (π_{*} v)

where

π : T^{*} Q \to Q

is the bundle projection.

In local coordinates $(q^{1}, \dots, q^{n})$ on $Q$ , inducing coordinates $(q^{1}, \dots, q^{n}, p_{1}, \dots, p_{n})$ on $T^{*} Q$ where $p_{i}$ are the coefficients of $ξ = \sum p_{i} d q^{i}$ , we have:

θ = i = 1 \sum n p_{i} d q^{i}

Define the canonical symplectic form $ω = - d θ$ :

ω = - d θ = i = 1 \sum n d q^{i} \land d p_{i} = i = 1 \sum n d p_{i} \land d q^{i}

This is manifestly in Darboux form, showing $(T^{*} Q, ω)$ is symplectic. The coordinates $(q^{i}, p_{i})$ are precisely the position-momentum coordinates of Hamiltonian mechanics. Note that $ω$ is independent of the choice of coordinates on $Q$ —different coordinate charts give related Darboux coordinates via symplectomorphism.

2

Complex Projective Space: Consider

CP^{n}

with the Fubini-Study form

ω_{F S}

. In homogeneous coordinates

[z_{0} : \dots : z_{n}]

CP^{n}

ω_{F S} = \frac{i}{2 π} \partial \overset{ˉ}{\partial} lo g (j = 0 \sum n ∣ z_{j} ∣^{2})

This is a Kähler form, hence closed. It's non-degenerate, making $(CP^{n}, ω_{F S})$ symplectic.

On the standard chart $U_{0} = {z_{0} \neq = 0}$ with coordinates $w_{j} = z_{j} / z_{0}$ for $j = 1, \dots, n$ , write $w_{j} = x_{j} + i y_{j}$ :

ω_{F S} ∣_{U_{0}} = \frac{i}{2 π ( 1 + ∣ w ∣ ^{2} ) ^{2}} (1 + ∣ w ∣^{2}) j \sum d w_{j} \land d \overset{w}{ˉ}_{j} - j, k \sum \overset{w}{ˉ}_{j} w_{k} d w_{j} \land d \overset{w}{ˉ}_{k}

At the point $[1 : 0 : \dots : 0]$ (where $w = 0$ ), this simplifies to:

ω_{F S} ∣_{w = 0} = \frac{1}{π} j = 1 \sum n d x_{j} \land d y_{j}

By Darboux's theorem, there exist coordinates near any point making $ω_{F S}$ standard. However, globally, $CP^{n}$ is compact while $R^{2 n}$ is not, showing symplectic manifolds can differ drastically in their global properties despite local equivalence.

Dropping Closedness: Consider

R^{2}

with the 2-form

α = (1 + x) d x \land d y

. This is non-degenerate everywhere (since

1 + x > 0

for

x > - 1

), but:

d α = d x \land d x \land d y = 0

Actually, $α$ IS closed (as any 2-form in dimension 2). Let's instead consider $R^{4}$ with coordinates $(x, y, p, q)$ and:

α = d x \land d y + x d p \land d q

Check: $d α = d x \land d p \land d q \neq = 0$ , so $α$ is not closed. This form is non-degenerate (checking the induced map $T R^{4} \to T^{*} R^{4}$ ).

Why Darboux fails: The proof requires writing

α = ω_{0} + d β

after adjusting coordinates. But if

d α \neq = 0

, the deformation

ω_{t} = ω_{0} + t α

doesn't consist of closed forms, and the Moser argument breaks down—we cannot solve

L_{X_{t}} ω_{t} = - \partial_{t} ω_{t}

while maintaining

d ω_{t} = 0

Without closedness, there's no guarantee of local normal form. The condition $d ω = 0$ is essential.

Physical Picture: In classical mechanics, the symplectic form encodes the fundamental relationship between positions

q_{i}

and momenta

p_{i}

. Darboux's theorem says: no matter how complicated your mechanical system looks globally, locally you can always find canonical coordinates where Hamilton's equations take their standard form. Geometric Meaning: Symplectic geometry is about area preservation, not distance. The 2-form

ω

measures oriented areas of 2-dimensional surfaces in phase space. Darboux coordinates are those in which area measurement looks standard: the area of a parallelogram spanned by

\partial / \partial q_{i}

and

\partial / \partial p_{i}

is exactly 1. Contrast with Riemannian Geometry: A Riemannian metric can have curvature—locally distinguishing different manifolds. But symplectic forms have no "symplectic curvature." Think of it like this: all Riemannian manifolds have curvature invariants (Gaussian curvature, sectional curvature), but all symplectic manifolds are locally flat. The "interesting stuff" migrates entirely to topology.

Historical Development

Symplectic geometry has its roots in the 19th-century development of analytical mechanics. William Rowan Hamilton's reformulation of Newtonian mechanics (1833) introduced the Hamiltonian formalism, though the underlying geometric structure wasn't yet recognized. Carl Gustav Jacobi developed the theory of Canonical Transformations, noting certain coordinate changes preserve the form of Hamilton's equations.

Gaston Darboux proved his theorem in 1882 in the paper "Sur le problème de Pfaff," establishing that all symplectic forms are locally equivalent. The term "symplectic" itself was coined much later by Hermann Weyl (1939), derived from the Greek word for "complex" (συμπλέκειν), reflecting the connection to complex geometry through Kähler manifold|Kähler manifolds.

The modern categorical and global perspective emerged in the mid-20th century. The work of Jean-Marie Souriau (1970s) and others reformulated mechanics geometrically, treating symplectic manifolds as objects of study independent of their physical origins. Jürgen Moser's 1965 proof of Darboux's theorem via his deformation technique became standard, as it generalizes to many relative versions and perturbation results.

The 1985 work of Mikhail Gromov on pseudoholomorphic curves revolutionized the field, creating symplectic topology and showing that symplectic geometry has surprising rigidity phenomena (like Gromov's non-squeezing theorem) despite Darboux's local flexibility. This led to the development of Floer homology, symplectic field theory, and connections to mirror symmetry in string theory.

Connections to Other Areas

Hamiltonian Mechanics and Quantum Theory: The passage from classical to quantum mechanics is formalized through geometric quantization, which assigns a Hilbert space to a symplectic manifold. The closedness condition

d ω = 0

ensures the existence of a prequantum line bundle, and Darboux's theorem guarantees that locally, quantization looks like the standard Schrödinger representation with position and momentum operators satisfying

[q_{i}, p_{j}] = i ℏ δ_{ij}

. The ambiguity in choosing Darboux coordinates corresponds to unitary equivalence of quantum representations, while global topological constraints (like Chern class|Chern classes) impose quantization conditions (e.g., Bohr-Sommerfeld rules). The WKB approximation in quantum mechanics is precisely the statement that in the semiclassical limit, quantum dynamics follows Hamiltonian flow on the symplectic phase space.

Symplectic Geometry and Darboux's Theorem