Having read through some recent posts regarding the Lagrangian formulation being interpreted into an initial value problem rather than the familiar boundary condition problem we are familiar with, I have a question regarding Hamilton-Jacobi theory (HJT).

My understanding of (HJT) is that we can view canonical transformations as an infinitesimal advancement in time using Hamilton's principle function and therefore regard a problem as an initial value problem much like Newtonian physics. This is supported by the following result from many classical physics texts, where the Hamilton-Jacobi action differs from the Lagrangian action up to some constant $S_0(\boldsymbol{q_0})$ which disappears upon taking variations $\delta S$. \begin{equation} S(\boldsymbol{q,\alpha},t)=\int ^{t}_{t_0}\mathcal L(\boldsymbol{q,\dot q,}t)dt+S_0(\boldsymbol {q_0}) \end{equation} Where $\alpha$ is any constant of the motion, for instance the initial momentum conditions $\alpha _i =p_i(t_0)$. This suggests that the HJ action is the lagrangian action + some other term, sort of like an initial value problem advancing the action step by step.

Firstly is the above true? Secondly if so, how come Hamiltonian mechanics can be thought of as an initial value problem but Lagrangian cannot?


1 Answer 1


At least three different quantities in physics are customary called an action and denoted with the letter $S$:

  1. The off-shell action functional $S[q;t_i,t_f]$,

  2. The (Dirichlet) on-shell action function $S(q_f,t_f;q_i,t_i)$, and

  3. Hamilton's principal function$^1$ $S(q,\alpha, t).$

For their definitions and how they are related, see e.g. my Phys.SE answer here.

  1. On one hand, the (Dirichlet) on-shell action function (2) and Hamilton's principal function (3) are closely related, cf. this Phys.SE post. Explicit solutions to (2) and (3) are only known sufficiently simple cases.

  2. On the other hand, the off-shell action functional (1) is the one which is used in the stationary action principle with suitable boundary conditions imposed. The other two (2) and (3) cannot be used in a variational principle.

For a discussion of boundary value vs. initial value problems, see e.g. this Phys.SE post.


$^1$ Following e.g. H. Goldstein, CM, the Hamilton's principal function (3) is a type 2 generating function of canonical transformations. The integration constants $\alpha_i$ are identified with the new momenta $P_i$.

  • $\begingroup$ Thank you for your answer! I know understand that the on--shell action is the action of the classical trajectory. This action is a function of the initial/final-position/time of that ONE CORRECT trajectory. Hamilton's principal function (HPF) is the same as the on shell action apart from an undetermined constant which disappears upon taking variations. Can this be viewed therefore as an initial value problem? An initial action + a trajectory integral? $\endgroup$ Mar 11, 2015 at 0:08
  • $\begingroup$ Secondly with regard to the on shell vs off shell actions, the first is a function the second is a functional. Do they both take on the same meaning for the same physical paths? Does the off shell functional "reduce" to the function for the on shell solution? $\endgroup$ Mar 11, 2015 at 0:11
  • $\begingroup$ 1. HPJ does not depend on initial values but rather on new momenta $P_i$. 2. Yes. $\endgroup$
    – Qmechanic
    Mar 12, 2015 at 16:02

Your Answer

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

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