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

Definition of Lagrange function

  1. Mar 9, 2004 #1
    I'm learning classical mechanics right now, and I have a question about the "initial definition" of the Lagrange function. In my book, it is only introduced as L=T-V, but in many cases this doesn't help a lot, since it's not obvious what T or V is. For example, how would I come to the Lagrangian of a particle in a el.-magn. field without ever having heard of the vector potential, B=rot(A) ?

    This question came to me when I wanted to find the (time-dependant) Lagrangian of a simple one-dim. movement with friction proportional to velocity:
    m*d2(x) = f(x) - c*x

    thanks for help

  2. jcsd
  3. Mar 9, 2004 #2
    sorry, the equation should be
    m*d2(x) = f(x) - c*d(x)

    d( ) and d2( ) are the derivatives, f a force function
  4. Mar 9, 2004 #3


    User Avatar
    Science Advisor

    "Lagrangian of a simple one-dim. movement with friction proportional..."

    Maybe I am talking stupidly (I am going by memory from some reading done about ten years ago), but isn't it the case that extremal methods (such as those involving a Lagrangian) are not useful for solving the dynamics of non-conservative systems (such as those involving friction)?

    If I am wrong about this, I am sure somebody here will correct me.
  5. Mar 9, 2004 #4
    The Lagragian is defined as that which appears under the action integral. It's usually given by L = Kinetic Energy - Generalized Potential. I don't know of a way to derive the Lagrangian for a particle in an EM field - sorry. I only know the Lagrangian itself. For a relativistic particle it's at


    For non-relativistic particle its

    L = T - U


    U = Phi - qv*A

    where Phi = Coulomb Potential, q = charge, v = velocity and A = vector potential. Plug this into Lagrange's equations and you get the Lorentz force
  6. Mar 9, 2004 #5


    User Avatar
    Science Advisor
    Gold Member

    You can tabulate the work done by non-conservative forces in a Lagrangian formulation by including generalized forces.

    [tex] L = T^* - U [/tex]

    [tex] \frac{d}{dt}\left(\frac{\partial L}{\partial \dot q_i}\right) - \frac{\partial L}{\partial q_i} = Q_i [/tex]

    (That's just the notation I remember: qi = ith generalized coordinate, Qi = ith generalized force, T* = kinetic coenergy, U = potential energy, etc.)

    Back to the original question, here's a site where the Lagrangian of a particle is found starting with the Lorentz force law. I don't know if you can get around dealing with the magnetic vector potential. Hope that helps.
  7. Mar 10, 2004 #6


    User Avatar
    Homework Helper

    The electromagnetic potential is necessarily a 4-vector. The only literal answer to your question that I can think of is: "Someone can tell you what the Lagrangian is for a charge in an E&M field."
  8. Mar 11, 2004 #7
    Re: Re: Definition of Lagrange function

    Well, certainly you end up with the same potential, whatever you do. I meant, how would you come to it by only looking at the equation of movement. In my course, the potential U = Phi - qv*A had simply been "claimed", and we checked whether it is of the correct form for the Lagrange function.

    Well, I'm sure you're right. When attacking that little but frustrating problem I mentioned, I thaught it would be an exception to that rule, but now I think it's one of those excercises where you'd have to show that they cannot be solved:smile: .
  9. Mar 11, 2004 #8


    User Avatar
    Homework Helper

    Re: Re: Re: Definition of Lagrange function

    I've never been completely satisfied with the Lagrangian formulation as a fundamental formulation of mechanics. I think that you have to already have a good understanding of E&M in Newtonian mechanics. You would then know of course that the equation of motion can be written in component form as:

    d2xn/dt2 = (e/m) { En + εnml (dxm/dt) Bl }

    and that, in general:

    En = -∂nφ - ∂tAn


    Bl = εlkj∂kAj

    where the Latin indices represent Cartesian coordinate indices, ∂n is the partial derivative with respect to the coordinate with the given index, and εnml/εlkj is the Levi-Civita symbol.

    Putting the potential expressions into the equation of motion gives:

    d2xn/dt2 = (e/m) { ( -∂nφ - ∂tAn ) + εnml (dxm/dt) ( εlkj∂kAj ) }
    = (e/m) { -∂nφ - ∂tAn + εnmlεlkj (dxm/dt) ∂kAj }
    = (e/m) { -∂nφ - ∂tAn + ( δnkδmj - δnjδmk ) (dxm/dt) ∂kAj }
    = (e/m) { -∂nφ - ∂tAn + ( (dxm/dt) ∂nAm - (dxm/dt) ∂mAn ) }
    = (e/m) { -∂nφ + (dxm/dt) ∂nAm - ( ∂tAn + (dxm/dt) ∂mAn ) }
    = (e/m) { -∂nφ + (dxm/dt) ∂nAm - (dAn/dt) }


    d2xn/dt2 + (e/m) (dAn/dt) = (e/m) { -∂nφ + (dxm/dt) ∂nAm }


    (d/dt){ dxn/dt + (e/m) An } + (e/m) { ∂nφ - ∂n[ (dxm/dt) Am ] + ∂n[ (dxm/dt) ] Am } = 0


    (d/dt){ m dxn/dt + e An } + ∂n{ e φ - e (dxm/dt) Am } = - e ∂n[ (dxm/dt) ] Am

    I can't remember off the top of my head how to argue that the R.H.S. vanishes, so this is the one flaw that I leave to you. Anyway, assuming that this has been argued, we have:

    (d/dt){ m dxn/dt + e An } - ∂n{ -e φ + e (dxm/dt) Am } = 0

    which can be identified with the Euler-Lagrange equation:

    (d/dt){ ∂L/∂vn } - ∂n{ L } = 0

    to give

    ∂L/∂vn = m dxn/dt + e An


    L = -e φ + e (dxm/dt) Am + f(vi,t)

    where f(vi,t) is an arbitrary function of velocity and time. Therefore:

    L = f(vi,t) - e ( φ - (dxm/dt) Am )
    = (1/2)mv2 - e ( φ - v.A )

    Clearly then, e ( φ - v.A ) is the potential energy, and φ - v.A is the electromagnetic potential.
    Last edited: Mar 11, 2004
  10. Mar 12, 2004 #9
    It should be noted that the L in that equation may not obey Hamilton's principle.

    turin wrote
    What is is that you're calling "electromagnetic potential"?

    The 4-potential is defined as a 4-vector whose time component is proportional to the Coulomb potential and whose spatial portion is the vector potential .
  11. Mar 12, 2004 #10
    The Lag. is not always derived. It's like finding a closed formula for the integral of a new function. You guess at a definition and try it to see if it works. Maybe in some cases, if you are unlucky, you do too much trial-and-error work.
  12. Mar 12, 2004 #11


    User Avatar
    Homework Helper

    Yes, that's exactly what I meant.
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook