Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Strategies for solving Hamilton's equations

  1. Nov 11, 2013 #1
    When solving virtually any non-trivial system via Hamilton's equations, it seems that I'm ending up with equations of the form

    ## ∂H/∂p = dx/dt = p/m ##
    ## ∂H/∂x = -dp/dt = -F(x) ##
    ## p(t) = ∫(dp/dt)dt ##
    ## d^2x/dt^2 = F(x) ##
    E.g., everything is given as a function of x, but we need to solve for t. I can change the variable of integration to x and get e.g. p(x), but that doesn't really help (especially since you can often directly solve for p(x)). One particular problem that's been irking me is where F(x) is either a delta function or a step function.

    So...are there any tricks for solving Hamilton's equations (or Euler/Lagrange) for systems that are NOT like the ## d^2x/dt^2 = -Kx ## that invariably show up in textbooks?
     
  2. jcsd
  3. Nov 11, 2013 #2
    Well, there's really only a very small number of systems that have exact analytical solutions. Most systems have to be integrated numerically. If you post a specific example someone might be able to help you solve it, but I don't think there's really much general advice one can give.
     
  4. Nov 11, 2013 #3
    Actually, the system that brought me here is one where the equations of motion are well known: elastic collisions in one dimension -- either between two particles or between a particle and a potential barrier (possibly infinite).

    I'm afraid to give too specific / complete a derivation, as it will probably get deleted as "homework-like", but one problem I've been wrestling with has ## H = p^2/2m + Kθ(x) ##, where θ(x) is a step function, and K → ∞...so in the above example, F(x) would be -Kδ(x), and the equations of motion would be the particle moving from -∞ toward zero at constant momentum ##p_0##, then at x = 0, it immediately changes to ##-p_0##, and moves back towards -∞. So, ## dp/dt = -2p(x)δ(x) ##, and ## x(t) = -|p_0t/m| ##. But, for the reasons stated above, I can't get from H to the (known) solution...and I'm having similar problems with "simpler" problems (e.g. a ramp instead of a delta function). If you're really interested, there's another thread on that particular problem here: https://www.physicsforums.com/showthread.php?p=4568733#post4568733...but I think the root of the problem there is I suck at integrals -- so if I get this figured out, that should follow.

    Thanks,
    Justin
     
Know someone interested in this topic? Share this thread via Reddit, Google+, Twitter, or Facebook




Similar Discussions: Strategies for solving Hamilton's equations
  1. Solving the Equations (Replies: 2)

Loading...