1. PF Contest - Win "Conquering the Physics GRE" book! Click Here to Enter
    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!

General relativity question - geodesics.

  1. Apr 27, 2010 #1
    I'm doing some revision for a General relativity exam, and came across this question:

    A Flat Earth space-time has co-ordinates (t, x, y, z), where z > 0, and a metric

    ds2 = ((1 + gz)2)dt2 − dx2 − dy2 − dz2

    where g is a positive constant.
    Write down the geodesic equations in this space-time.

    Hence, or otherwise, show that A Flat Earth physicist, stationary at a point with z = h, will measure a ‘gravitational acceleration’ of magnitude g(1 + gh)−1


    I've solved the geodesics, but can't get the 'show that' at the end:

    Using the Euler-Lagrange equation:

    [tex] \stackrel{d}{ds} \stackrel{\partial L}{\partial \dot{x^{\mu}}} - \stackrel{\partial L}{\partial x^{\mu}} =0[/tex]

    I get

    [tex]\ddot{z} = 2g(1+gz)\dot{t}^2[/tex]


    [tex]\dot{t} = E/(1+gz)^2[/tex] ( where E is a constant )


    [tex]\ddot{z} = g{E^2}/(1+gz)^3[/tex]

    which isn't right. Alternatively,

    [tex]\ddot{z}= \dot{t}^2 z'' + \ddot{t}z'[/tex] ( where ' is (d/dt) )

    it's stationary, so z' = 0 thus [tex]z''=\ddot{z}/\dot{t}^2=2g(1+gz)[/tex]

    which still isn't right. Any takers?
    Last edited: Apr 27, 2010
  2. jcsd
  3. Apr 27, 2010 #2


    User Avatar
    Homework Helper

    I think the two in your geodesic equation shouldn't be there. You probably missed the 1/2 in the Christoffel symbol. Moreover you made a mistake in trying to find an expression for [itex](dt/ d\tau)^2[/itex]. Can you show us how you obtained your expression?
    Last edited: Apr 27, 2010
  4. Apr 27, 2010 #3
    Yes the 2 is wrong. My apologies.

    For the mu = 0 equation I have:

    [tex](d/ds)(2\dot{t}(1+gz)^2) =0[/tex]


    [tex] \dot{t} (1+gz)^2 = constant [/tex]


    [tex]\ddot{z} = \dot{t}d\dot{z}/dt = \dot{t}(d/dt)(\dot{t}z') = \dot{t}(z'(d\dot{t})/dt +\dot{t}z'')=\dot{t}^2z''+\ddot{t}z'[/tex]
    Last edited: Apr 27, 2010
  5. Apr 27, 2010 #4
    Update: I've now solved it using the Christoffel symbols, but am still confused as to why the Euler-Lagrange method hasn't worked.

    This is wrong:
    [tex]\ddot{z} = -g{\dot{t}^2}(1+gz)[/tex] (1)

    It should read:

    [tex]\ddot{z} =-g{\dot{t}^2}/(1+gz)[/tex] (2)

    I got (1) direct from the EL equation

    [tex]-2\ddot{z} - {\dot{t}^2}d_{z}(1+gz)^2 = 0[/tex] (3)
    Last edited: Apr 27, 2010
  6. Apr 27, 2010 #5


    User Avatar
    Homework Helper

    You're probably losing terms in places where you should be summing over the indices. Using the EL-equations I end up with [itex]\ddot{z}-g(1+gz)\dot{t}^2 =0 [/itex].

    Your second equation is wrong as well. Did you derive that one using the geodesic equation? Either way I just calculated it both ways and I get the same result, as it should be. I can't be of further assistance unless you show me your full derivation. Also I don't think you have gotten your [itex]\dot{t}[/itex] correct yet.
    Last edited: Apr 27, 2010
  7. Apr 27, 2010 #6
    Thanks. Given that's the case then [tex]z''=\ddot{z}/\dot{t}^2=g(1+gz)[/tex] can't be true (since this is the wrong answer).

    I can't see any thing wrong for [tex]\dot{t}[/tex]:

    whats wrong with:
    [tex]L = (1+gz)^2dt^2-dx^2-dy^2-dz^2)[/tex]

    therefore by EL:

    [tex](d/ds)(2\dot{t}(1+gz)^2) =0[/tex]


    [tex] \dot{t} (1+gz)^2 = constant [/tex]
  8. Apr 27, 2010 #7


    User Avatar
    Homework Helper

    Your Lagrangian is wrong, recall [itex]L=g_{\alpha \beta}\dot{x}^\alpha\dot{x}^\beta=(1+gz)^2 \dot{t}^2-\dot{x}^2-\dot{y}^2-\dot{z}^2[/itex].
    Last edited: Apr 27, 2010
Know someone interested in this topic? Share this thread via Reddit, Google+, Twitter, or Facebook