It was my understanding that the Hamiltonian formalism was inadequate to describe systems that are invariant under time reparametrization or that have gauge symmetries.

However, I see in Classical Dynamics by Jorge V. José and Eugene J. Saletan, that both a relativistic particle and a particle under an electromagnetic potential are described using the Hamilton-Jacobi formalism, dealing the right equations of motion.

I wonder why does this work: Are there systems that can be treated with by Hamilton-Jacobi formalism but yield false results when treated by Hamilton? Is there a way to adequately treat systems with the mentioned invariances through Hamiltonian mechanics? If so, are their Hamiltonians always of the form $H=T+V$?

  • 1
    $\begingroup$ I am not sure what exactly is meant by "Hamilton formalism" vs "Hamilton-Jacobi formalism", but the Hamilton-Jacobi equation is a first order PDE whose characteristics are given by Hamilton's equations. Solving them is mathematically equivalent. It is true that Hamilton's formalism (unlike Lagrange's) does not have a good generalization to gauge fields, but motion of particles in external fields is far from that. $\endgroup$
    – Conifold
    Dec 22, 2019 at 4:36
  • $\begingroup$ Which page in J&S? $\endgroup$
    – Qmechanic
    Dec 22, 2019 at 5:51

1 Answer 1

  1. Hamiltonian formalism also works for gauge systems, although one has to introduce constraints (and possibly the corresponding Lagrange multipliers), see e.g. Ref. 1. For the relativistic point particle, see e.g. this Phys.SE post.

  2. In fact the Hamilton-Jacobi equation is derived from the Hamiltonian formalism, not the other way around.

  3. The Hamiltonian $H$ is not always of the form $T+U$. See e.g. this Phys.SE post for the corresponding Lagrangian question.


  1. M. Henneaux & C. Teitelboim, Quantization of Gauge Systems, 1994.

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.