3
$\begingroup$

Let $f=f(q,p)$, $g=g(q,p)$ and Possion bracket $$\{f,g\}=\frac{\partial f}{\partial q}\frac{\partial g}{\partial p}-\frac{\partial f}{\partial p}\frac{\partial g}{\partial q}. \tag{1}$$ Then Hamilton's equations for Hamiltonian $H=H(q,p)$ are $$\dot{q}=\{q,H(q,p)\}, \tag{2}$$ $$\dot{p}=\{p,H(q,p)\}. \tag{3}$$ Now let $\mathcal{F}(\mathbb{R})$ be a space of function on $\mathbb{R}$ so that for $u\in\mathcal{F}$ we have $u$ and its derivatives decay at $\pm\infty$. I am trying to construct Hamilton's equations for Hamiltonian $$H=\int_{-\infty}^{\infty}\left(\frac{1}{2}u_x^2-u^3\right)\,dx \tag{4}$$ using Poisson bracket $$\{F,G\}(u)=\int_{-\infty}^{\infty}\frac{\delta F}{\delta u}\frac{d}{dx}\left(\frac{\delta G}{\delta u}\right)\,dx. \tag{5}$$ Is it correct if I take $H=H(u,v)$ with $u=u$ and $v=u_x$? If so then should I leave the Hamilton's equations as $$\dot{u}=\{u,H\}=-\int_{-\infty}^\infty\left(6uu_x+u_{xxx}\right)\,dx=-\int_{-\infty}^\infty\left(6uv+v_{xx}\right)\,dx ,\tag{6}$$ $$\dot{v}=\{v,H\}=\int_{-\infty}^\infty u_{xx}\,dx=\int_{-\infty}^\infty v_{x}\,dx ,\tag{7}$$ or evaluate the integrals?

$\endgroup$
4
$\begingroup$
  1. Hint: Eq. (6) in its current form (v4) is meaningless since the lhs. depends on $x$, while the rhs. is integrated over $x$. The functional derivative $$\tag{A}\frac{\delta F}{\delta u(x^{\prime})}~\stackrel{(B)}{=}~\frac{\delta u(x)}{\delta u(x^{\prime})}~=~\delta(x\!-\!x^{\prime})$$ in eq. (6) for the functional $$\tag{B} F[u]~:=~u(x)~=~\int \! dx^{\prime}~u(x^{\prime})~\delta(x^{\prime}\!-\!x)$$ includes a delta function, which removes the integral on the rhs. of eq. (6). The next eq. (7) is similar.

  2. OP asks the following in a comment.

    I can't take $\delta u/\delta u$ as 1?

    OP is apparently thinking of a 'same-$x$' functional derivative $\frac{\delta u(x)}{\delta u(x)}=1$, which is also discussed in e.g. this and this Phys.SE posts. However, the functional derivatives in eq. (5) are not 'same-$x$' functional derivatives, so the answer to the comment is No.

$\endgroup$
  • $\begingroup$ I can't take $\delta u/\delta u$ as 1? $\endgroup$ – Sukan Apr 9 '16 at 23:54
  • $\begingroup$ $\uparrow$ @Sukan: No, not in the above formulation. I updated the answer. $\endgroup$ – Qmechanic Apr 9 '16 at 23:56

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.