0
$\begingroup$

Could anyone explain to me how to obtain the Corollaries 2 and 3 in the following paragraph taken by Arnold's book?

The Lie algebra of hamiltonian functions

The hamiltonian vector fields on a symplectic manifold form a subalgebra of the Lie algebra of all fields. The hamiltonian functions also form a Lie algebra: the operation in this algebra is called the Poisson bracket of functions. The first integrals of a hamiltonian phase flow form a subalgebra of the Lie algebra of hamiltonian functions.

A The Poisson bracket of two functions

Let $(M^{2n}, \omega^2)$ be a symplectic manifold. To a given function $H: M^{2n} \to \mathbb{R}$ on the symplectic manifold there corresponds a one-parameter group $g^t_H : M^{2n} \to M^{2n}$ of canonical transformations of $M^{2n}$—the phase flow of the hamiltonian function equal to $H$. Let $F: M^{2n} \to \mathbf{R}$ be another function on $M^{2n}$.

Definition. The Poisson bracket $(F,H)$ of functions $F$ and $H$ given on a symplectic manifold $(M^{2n}, \omega^2)$ is the derivative of the function $F$ in the direction of the phase flow with hamiltonian function $H$: $$ (F,H)(x) = \frac{d}{dt}\Big|_{t=0} F(g^t_H(x)). $$ Thus, the Poisson bracket of two functions on $M$ is again a function on $M$.

Corollary 1. A function $F$ is a first integral of the phase flow with hamiltonian function $H$ if and only if its Poisson bracket with $H$ is identically zero: $(F,H) = 0$.

We can give the definition of Poisson bracket in a slightly different form if we use the isomorphism $I$ between 1-forms and vector fields on a symplectic manifold $(M^{2n},\omega^2)$. This isomorphism is defined by the relation (cf. Section 37 (wherein $\eta$ is a vector in the tangent space and $\omega^1$ is the corresponding 1-form in the cotangent space)) $$ \omega^2(\mathbf{\eta},I \omega^1) = \omega^1(\mathbf{\eta}). $$ The velocity vector of the phase flow $g^t_H$ is $I d H$. This implies

Corollary 2. The Poisson bracket of the functions $F$ and $H$ is equal to the value of the 1-form $dF$ on the velocity vector $I dH$ of the phase flow with hamiltonian function $H$: $$ (F,H) = dF(I dH). $$

Using the preceding formula again, we obtain

Corollary 3. The Poisson bracket of the functions $F$ and $H$ is equal to the "skew scalar product" of the velocity vectors fo the phase flows with hamiltonian functions $H$ and $F$: $$ (F,H) = \omega^2(I dH, I dF).$$

The emphasized explanation of the notation of Sec. 37 was added.

$\endgroup$
2
  • 5
    $\begingroup$ Please use Mathjax. Images cannot be searched, and these is particlular are unreadable. $\endgroup$
    – mike stone
    Commented Jul 24 at 11:03
  • $\begingroup$ Agreed with mike. I'm having trouble confirming which superscripts and subscripts are the same or different because the text is so small. However, to narrow the question down a bit: Do you have trouble getting from the first definition to the second, or from the second definition to Corollary 2 and 3? If going from the first definition to the second is the difficult hurdle, then "Section 37" is probably something we would need to understand in order to help, as it appears to have some derivation there. $\endgroup$
    – Cort Ammon
    Commented Jul 24 at 13:36

1 Answer 1

1
$\begingroup$

Corollary 2 is just working out the chain rule in the derivative that defines the Poisson bracket. In Sec. 38A, Arnol'd defines the one-parameter group of diffeomorphisms as $$ \frac{d}{dt}\Big|_{t=0} g^t\mathbf{x} = I dH(\mathbf{x}). $$ This is a vector field. In general, the derivative of $F$ evaluated at a point $\mathbf{y}(t)$ with respect to the real parameter $t$ is \begin{align} \frac{d}{dt} F(\mathbf{y}(t)) &= dF(\dot{\mathbf{y}}), \end{align} which is just the exterior calculus version of the chain rule. You can guess its form because it has to be a function from the reals to the reals, and there's no other such function that's linear in $F$ and in $\dot{\mathbf{y}}$ that you can make out of the one-form $dF$ and the vector $\dot{\mathbf{y}}$. Plugging these two equations into each other proves the corollary.

Corollary 3 uses the definition of $I$ backwards, where $dF$ takes the place of $\omega^1$ and $I dH$ takes the place of $\eta$.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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