1. Limited time only! Sign up for a free 30min personal tutor trial with Chegg Tutors
    Dismiss Notice
Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Fermat principle and Euler-Lagrange question

  1. Jun 30, 2013 #1

    The Fermat principle says that
    (***) Δt = (1/c) ∫ μ(x,y) √1+y'2 dt

    Say, we are studying a GRIN material where the refraction index is μ = μ(x,y) and want to figure out the shape of the ray trajectory y=y(x).

    Here is what I know (this is not a homework question) but am unsure if the approach is correct:

    (1) [itex]\frac{1}{R}[/itex]=[itex]\frac{∂ ln μ}{∂n}[/itex], from Peters' paper . I expect the shape to be a hyperbola say a/x. Can I use this equation (it is derived by seeking stationary time Δt) to calculate μ=μ(x,y)?
    (R is the radius of curvature, [itex]\frac{∂}{∂n}[/itex], is a derivative along the normal to the curve)

    (2) Once I have μ=μ(x,y), can I plug it into (***), also plug in y=a/x and derive Δt? I am hoping this can then be minimized to find the optimal parameter a or any similar curve parameter.
    Does this make sense?

    (3) or should I write the Euler-Lagrange equation and try to find out something from it?

    (4) or should I use a Lagrange multiplier and "constrain" the ray to the hyperbola a/x

    I am unsure which approach is correct; and also, what of the above known/expected expressions should I plug in into the Lagrangian and then do the analysis.
    My goal is to find out the extremal curve and its equation.

    Need help from someone experienced in Fermat/lagrangian formalism...
  2. jcsd
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook

Can you offer guidance or do you also need help?
Draft saved Draft deleted