MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Let $g$ be a positive definite symmetric form on a $2n$-dimensional vector space (a metric), $\mathbb{J}$ the symplectic unit and $c$ a real number. Define the symmetric form

$h = g + c[g,\mathbb{J}],$

where $[\cdot,\cdot]$ is the commutator. Have you ever encountered this or similar objects? If so, what is the geometrical meaning of requiring it to be positive definite,



share|cite|improve this question
how do you define the commutator between $g$ and $\mathbb{J}$? As operators that doesn't make sense, so you'd be doing this as matrices w.r.t. a fixed basis? If so the result would be basis dependent. Or do you mean $[g,\mathbb{J}]$ = g\mathbb{J} - \mathbb{J}^Tg$? – Willie Wong Apr 23 '11 at 12:06
(Also, can you give the context in which you encountered this object?) – Willie Wong Apr 23 '11 at 12:07
What I am trying to get you to realise is that the fact $\mathbb{J}^2 = -\mathbb{I}$ tells you that $\mathbb{J}:V\to V$, whereas $g$, a symmetric bilinear form, maps $V\otimes V \to \mathbb{R}$, or equivalently $g: V\to V^*$. Hence $\mathbb{J}g$ is ill defined as a composition of maps. What you really want is probably the induced action $\mathbb{J}^T: V^*\to V^*$ instead in that direction. If $g$ is positive definite and symmetric on a finite dimensional vector space, there always exists (Graham-Schmidt) an orthonormal basis. In this basis, $g = \mathbb{I}$, and the matrix commutator of – Willie Wong Apr 23 '11 at 21:52
any matrix $\Omega$, completely independent of whatever property $\Omega$ has, against $\mathbb{I}$, is $[\mathbb{I},\Omega] = 0$. If your statement is completely dependent on the choice of vector basis, it can have no natural geometric interpretation! On the other hand, if what you want is actually $\mathbb{J}^Tg - g \mathbb{J}$, this object is geometric, and in the simplest case where $g = \mathbb{I}$ and $\mathbb{J}$ is the standard $[ 0,-\mathbb{I};\mathbb{I},0]$ matrix, you get something nontrivial: the commutator is now $\pm 2\mathbb{J}$ (as a matrix) or in fact the antisymmetric form – Willie Wong Apr 23 '11 at 21:56
induced by $2g\mathbb{J}$. In this sense the computation is coordinate invariant and may admit a nice geometrical interpretation. – Willie Wong Apr 23 '11 at 21:57

I have met the combination of complex structures, symplectic forms, scalar products on the same real vector space in the following context.

Let $V$ be a real vector space.
The complex vector space structures compatible with the assigned real vector space structure of $V$ are in correspondence one-to-one with the complex operators on $V$, i.e. the linear operators $J$ on $V$ such that $\mathbb{J}^2=-id_V\equiv \mathbb{I}$; this correspondence is realized through the relation $(a+ib).v=av+b\mathbb{J}v$, for any $a,b\in\mathbb{R}$, and $v\in V$.
Obviously such structures $(V,\mathbb{J})$ exist if and only if $\dim{V}$ is pair.

Definition. Let $\Omega$ be a symplectic form and $\mathbb{J}$ a complex operator on $V$. The complex operator $\mathbb{J}$ is said to be adapted to $\Omega$ when there exists a pseudo-hermitian form $\eta$ on the complex vector space $(V,\mathbb{J})$ such that $\Im\eta=\Omega$.

Theorem $\mathbb{J}$ is adapted to $\Omega$ if and only if $\mathbb{J}$ is an isomorphism of $(V,\Omega)$, i.e. $\mathbb{J}^T\circ\Omega^\flat\circ\mathbb{J}=\mathbb{J}\circ\Omega\in L(V,V^*)$; and in the affirmative case there is a unique hermitian form $\eta$ on $(V,\Omega)$ with $\Im\eta=\Omega$, it is given by $\eta=(\Omega^\flat\circ\mathbb{J})^\sharp+i\Omega$.

share|cite|improve this answer
This seems to go in my direction. I'll try to think about it. – tomate Apr 24 '11 at 10:40

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.