I have to find the value $\alpha\in\mathbb{R}$ such that the following change of coordinates is symplectic:

$ \varphi(p,q)\rightarrow (P,Q)$


$Q = q^2 + \alpha\sqrt{q^2+p}$

$P = q + \sqrt{q^2+p}$

My first approach to solve it was to directly verify for which $\alpha$ it holds $dp\wedge dq = dP \wedge dQ$ and if comes out $\alpha=2$, which should be right.

But on some notes, I found that to prove that this holds it could be even imposed that the determinant of the Jacobian of the transformation $\varphi$ is equal 1.

Is this related to Liouville volume preserving theorem? I really don't get why we should have that this last verification implies that $\varphi$ is a symplectomorphism.


I one sentence: You are in 2D, the symplectic form is the volume form.

In more words: Jacobian determinant measures (local) volume stretching, so is being 1 at every point means volume is preserved, but in 2D this is the same as being a symplectomorphism. You can see it very explicitly: one one hand, Jacobian determinant is $\frac{\partial{P}}{\partial{p}} \frac{\partial{Q}}{\partial{q}} - \frac{\partial{Q}}{\partial{p}} \frac{\partial{P}}{\partial{q}}$; on the other hand $dP= \frac{\partial{P}}{\partial{p}} dp+ \frac{\partial{P}}{\partial{q}} dq$ and $dQ= \frac{\partial{Q}}{\partial{p}} dp+ \frac{\partial{Q}}{\partial{q}} dq$ so $dP\wedge dQ= (\frac{\partial{P}}{\partial{p}} \frac{\partial{Q}}{\partial{q}} - \frac{\partial{Q}}{\partial{p}} \frac{\partial{P}}{\partial{q}}) dp \wedge dq$. Hence $dP\wedge dQ=dp\wedge dq$ is the same as Jacobian 1.

In higher dimensions, the volume form is $\omega^n$. Preserving $\omega$ (symplectomorphism) implies preserving volume $\omega^n$. This is Liouville theorem. However, when $n\geq 2$ (i.e. in 4D and up) the converse is not true.


To begin just a definition: A symplectic linear structure on $\mathbb{R}^{2n}$ is a non-degenerate bilinear skew symmetric 2-form. This form is called the skew-scalar product such that $[a,b]=-[b,a]$.

Now a transformation $S:\mathbb{R}^{2n} \rightarrow \mathbb{R}^{2n}$ of the symplectic space is symplectic iff it preserves the differential 2-Form. In your case where $n=1$ this means that $dp \wedge dq$ is preserved and maps to $dP \wedge dQ$. Which is equivelent to $[Sa,Sb]=[a,b]$

Moreover a transformation is symplectic iff the Matrix Representation of S satisfies $S^T\Omega S=\Omega$ where $\Omega$ is the block matrix

\begin{bmatrix} 0 & -I \\ I & 0 \end{bmatrix}

and I is the identity matrix; becasue: $[\Omega Sa,Sb]=[\Omega a,b]=[S^T \Omega Sa,Sb]$.

Now then you could find the matrix representation of the transformation $\phi$ and then compute $\phi ^T \Omega \phi$, let it equal to $\Omega$ and solve for $\alpha$.

Furthermore I recommend the chapter on Symplectic Manifolds from V.I Arnold's book on Classical Mechanics.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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