ProofComplete

Proof of Symplectic Basis Theorem

ProofConstruction of Symplectic Basis

We prove by induction on n=12dim⁑Vn = \frac{1}{2}\dim V that every symplectic vector space (V,Ο‰)(V, \omega) admits a symplectic basis.

Base Case (n=1n = 1, dim⁑V=2\dim V = 2):

Since Ο‰\omega is non-degenerate, there exist v,w∈Vv, w \in V with Ο‰(v,w)β‰ 0\omega(v, w) \neq 0. Rescaling, we may assume Ο‰(v,w)=1\omega(v, w) = 1. Set e1=ve_1 = v and f1=wf_1 = w. Then:

Ο‰(e1,f1)=1,Ο‰(e1,e1)=0,Ο‰(f1,f1)=0\omega(e_1, f_1) = 1, \quad \omega(e_1, e_1) = 0, \quad \omega(f_1, f_1) = 0

So {e1,f1}\{e_1, f_1\} is a symplectic basis.

Inductive Step: Assume the theorem holds for symplectic spaces of dimension 2(nβˆ’1)2(n-1).

Let (V,Ο‰)(V, \omega) have dimension 2n2n. Choose any non-zero v1∈Vv_1 \in V. By non-degeneracy, there exists w∈Vw \in V with Ο‰(v1,w)β‰ 0\omega(v_1, w) \neq 0. Normalizing, choose f1f_1 so that Ο‰(v1,f1)=1\omega(v_1, f_1) = 1. Set e1=v1e_1 = v_1.

Define W=span{e1,f1}W = \text{span}\{e_1, f_1\}. Then:

  • WW is a symplectic subspace: Ο‰βˆ£W\omega|_W is non-degenerate
  • dim⁑W=2\dim W = 2

Consider the symplectic complement:

WΟ‰={v∈V:Ο‰(v,w)=0Β forΒ allΒ w∈W}W^\omega = \{v \in V : \omega(v, w) = 0 \text{ for all } w \in W\}

We claim that dim⁑WΟ‰=2nβˆ’2\dim W^\omega = 2n - 2 and V=WβŠ•WΟ‰V = W \oplus W^\omega.

Proof of claim:

  1. Non-degeneracy of Ο‰βˆ£WΟ‰\omega|_{W^\omega}: Suppose v∈WΟ‰v \in W^\omega satisfies Ο‰(v,u)=0\omega(v, u) = 0 for all u∈WΟ‰u \in W^\omega. Then Ο‰(v,w)=0\omega(v, w) = 0 for all w∈Ww \in W (by definition of WΟ‰W^\omega) and for all u∈WΟ‰u \in W^\omega, so Ο‰(v,x)=0\omega(v, x) = 0 for all x∈Vx \in V. By non-degeneracy, v=0v = 0.

  2. Direct sum decomposition: We have W∩WΟ‰βŠ†WW \cap W^\omega \subseteq W and W∩WΟ‰βŠ†WΟ‰W \cap W^\omega \subseteq W^\omega. If v∈W∩WΟ‰v \in W \cap W^\omega, then v∈Wv \in W and Ο‰(v,w)=0\omega(v, w) = 0 for all w∈Ww \in W. Since Ο‰βˆ£W\omega|_W is non-degenerate, v=0v = 0. Thus W∩WΟ‰={0}W \cap W^\omega = \{0\}.

  3. Dimension count: By linear algebra, dim⁑WΟ‰=2nβˆ’dim⁑W=2nβˆ’2\dim W^\omega = 2n - \dim W = 2n - 2. Thus:

dim⁑(W+WΟ‰)=dim⁑W+dim⁑WΟ‰=2+(2nβˆ’2)=2n=dim⁑V\dim(W + W^\omega) = \dim W + \dim W^\omega = 2 + (2n-2) = 2n = \dim V

Therefore V=WβŠ•WΟ‰V = W \oplus W^\omega.

By the inductive hypothesis, (WΟ‰,Ο‰βˆ£WΟ‰)(W^\omega, \omega|_{W^\omega}) admits a symplectic basis (e2,…,en,f2,…,fn)(e_2, \ldots, e_n, f_2, \ldots, f_n).

The union (e1,e2,…,en,f1,f2,…,fn)(e_1, e_2, \ldots, e_n, f_1, f_2, \ldots, f_n) is a symplectic basis for VV:

  • Ο‰(ei,ej)=0\omega(e_i, e_j) = 0 for all i,ji, j (since each eie_i is in either WW or WΟ‰W^\omega, both isotropic)
  • Ο‰(fi,fj)=0\omega(f_i, f_j) = 0 for all i,ji, j (similarly)
  • Ο‰(ei,fj)=Ξ΄ij\omega(e_i, f_j) = \delta_{ij} (by construction for i=1i=1 or j=1j=1, and by induction for i,jβ‰₯2i, j \geq 2)

This completes the induction.

β– 
Remark

The proof is constructive: given any non-zero vector, we can algorithmically construct a symplectic basis. The key insight is that symplectic complements of symplectic subspaces are again symplectic, allowing induction.

DefinitionSymplectic Gram-Schmidt Process

The proof above can be viewed as a symplectic Gram-Schmidt process: starting from any basis {v1,…,v2n}\{v_1, \ldots, v_{2n}\}, we construct a symplectic basis by:

  1. Normalize v1v_1 and find f1f_1 with Ο‰(v1,f1)=1\omega(v_1, f_1) = 1
  2. Orthogonalize remaining vectors with respect to span{v1,f1}\text{span}\{v_1, f_1\} using WωW^\omega
  3. Repeat inductively

Unlike the usual Gram-Schmidt for inner products, the symplectic version maintains isotropy of subspaces.

ExampleExplicit Construction in $\mathbb{R}^4$

Consider R4\mathbb{R}^4 with Ο‰=dx1∧dx2+dx3∧dx4\omega = dx_1 \wedge dx_2 + dx_3 \wedge dx_4. Start with v1=(1,0,0,0)v_1 = (1, 0, 0, 0).

We need ww with Ο‰(v1,w)β‰ 0\omega(v_1, w) \neq 0. Try w=(0,1,0,0)w = (0, 1, 0, 0): Ο‰(v1,w)=1\omega(v_1, w) = 1. Set e1=(1,0,0,0)e_1 = (1,0,0,0), f1=(0,1,0,0)f_1 = (0,1,0,0).

The symplectic complement is:

Wω={(x1,x2,x3,x4):ω((x1,x2,x3,x4),(1,0,0,0))=0 and ω((x1,x2,x3,x4),(0,1,0,0))=0}W^\omega = \{(x_1, x_2, x_3, x_4) : \omega((x_1,x_2,x_3,x_4), (1,0,0,0)) = 0 \text{ and } \omega((x_1,x_2,x_3,x_4), (0,1,0,0)) = 0\}

This gives x2=0x_2 = 0 and βˆ’x1=0-x_1 = 0, so WΟ‰=span{(0,0,1,0),(0,0,0,1)}W^\omega = \text{span}\{(0,0,1,0), (0,0,0,1)\}.

Set e2=(0,0,1,0)e_2 = (0,0,1,0) and f2=(0,0,0,1)f_2 = (0,0,0,1). Then (e1,e2,f1,f2)(e_1, e_2, f_1, f_2) is a symplectic basis.

Remark

The uniqueness of the standard form (up to symplectomorphism) contrasts sharply with Riemannian geometry, where metrics on a fixed vector space form an infinite-dimensional cone. This rigidity makes symplectic geometry both simpler and more constrained than Riemannian geometry.