This is a special case of this question of mine, which, I think, might have drawn little attention because it was too general. In this question, I would like to consider a specific case.

Take a classical system given by the action $$ S[x] = \int dt \left[ \frac{\alpha}{2} \ddot{x}^2 + \frac{\beta}{2} \dot{x}^2 + \frac{\gamma}{2} x^2 \right]. $$

This action is Lagrangian, but it is not what we are usually dealing with in physics, because the Lagrangian contains second time derivatives.

The equations of motion are:

$$ \alpha \ddddot{x} - \beta \ddot{x} + \gamma x = 0. $$

This can be solved as usual by employing Fourier transform:

$$ x(t) = a_1 e^{i \omega_1 t} + a_1^{*} e^{-i \omega_1 t} + a_2 e^{i \omega_2 t} + a_2^{*} e^{-i \omega_2 t}, $$

where $\omega_1^2$ and $\omega_2^2$ are the two roots of $\alpha \omega^4 + \beta \omega^2 + \gamma$, given by

$$ \omega_{1,2}^2 = \frac{- \beta \pm \sqrt{\beta^2 - 4 \alpha \gamma}}{2 \alpha} .$$

Now by definition the phase space is the space of solutions of equations of motion. In this case it is parametrized by two complex co-ordinates $a_{1,2}$, meaning that it is 4-dimensional.

I would like to know if there's a natural way to associate the symplectic structure (Poisson bracket) on this phase space. I.e.,

$$ \{a_1, a_1^{*}\} = ? $$ $$ \{a_1, a_2^{*}\} = ? $$ $$ \{a_2, a_1^{*}\} = ? $$ $$ \{a_2, a_2^{*}\} = ? $$ $$ \{a_1, a_2\} = ? $$ $$ \{a_1^{*}, a_2^{*}\} = ? $$


FWIW, the Ostrogradsky formalism provides a systematic recipe for obtaining a Hamiltonian phase space formulation of an arbitrary Lagrangian theory with higher-time derivatives. However, it famously leads to quantum mechanical instabilities, cf. e.g. this Phys.SE answer by user genneth.


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.