Assume we have the following Lagrangian field density where $x, x'$ both three dimensional real vectors are coordinates and $t$ represents time, field is given by $\phi$. Assume for the sake of concreteness that $\phi$ is a function that returns three-dimensional real vector.

$$ \mathscr{L} = f(x) (\partial_t\phi)^2 + \int g(x, x', \phi(x), \phi(x')) dx'. $$

Define canonical momentum to be:

$$ \pi = \frac{\partial \mathscr{L}}{\partial(\partial_t\phi)} = 2f(x)\partial_t\phi. $$


  1. Is it true that we can write equations of motion as: $$ \partial_t \pi_i = - \sum \limits_j \frac{\partial}{\partial x_j} P_{ij}? $$ Here, $P_{ij}$ is some function.

  2. Is there an equivalent law of conservation of energy? How does it look like?

  1. To keep the notation simple let us consider bi-local point-mechanics $$ S[q,v]~=~\int \! dt~dt^{\prime}~{\cal L}(t,t^{\prime}), \tag{A} $$ where we use the shorthand notation $$ {\cal L}(t,t^{\prime})~=~{\cal L}(q(t),q(t^{\prime}),v(t),v(t^{\prime}),t,t^{\prime}).\tag{B}$$ (It is in principle straightforward to generalize to multi-local field theory.)

  2. Define momentum $$p(t)~:=~\frac{\delta S[q,v]}{\delta v(t)} ~=~\int \! dt^{\prime}~\frac{\partial[{\cal L}(t,t^{\prime})+{\cal L}(t^{\prime},t)]}{\partial v(t)},\tag{C}$$ force $$F(t)~:=~\frac{\delta S[q,v]}{\delta q(t)}~=~\int \! dt^{\prime}~\frac{\partial[{\cal L}(t,t^{\prime})+{\cal L}(t^{\prime},t)]}{\partial q(t)},\tag{D}$$ Lagrangian $$ L(t)~:=~\int \! dt^{\prime}~[{\cal L}(t,t^{\prime})+{\cal L}(t^{\prime},t)], \tag{E}$$ and energy function $$ H(t)~:=~p(t)v(t)-L(t).\tag{F}$$

  3. Next we identify $$\dot{q}(t)~\approx~v(t),\tag{G}$$ and assume the stationary action principle (SAP). The Lagrangian EOM becomes $$\dot{p}(t)~\approx~F(t).\tag{H}$$

  4. One may show that if ${\cal L}(t,t^{\prime})$ has no explicit time dependence, then the energy function $H(t)$ is conserved on-shell.

  5. For more on non-local action functionals, see e.g. this Phys.SE post.

  • $\begingroup$ Thanks! 1. Why do you define Lagrangian (and respective quantities) in such a symmetric way? 2. Why is Lagrangian defined not in a way that action is integral of Lagrangian (which I think is the usual convention)? 3. When you write in point 4 that $\mathcal{L}$ has no explicit time dependence do you mean that both partial derivatives are zero? $\endgroup$ – HydrodynamicsPlease Aug 6 '19 at 1:00
  • 1
    $\begingroup$ 1. The symmetric form comes from the SAP. 2. What you want to call the Lagrangian for non-local theories is not set in stone. The definition (E) is here chosen to have eq. (F). 3. Yes. $\endgroup$ – Qmechanic Aug 6 '19 at 12:56

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.