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

Easy Variation Question?

  1. Dec 7, 2012 #1
    I am considering the variation of

    [itex] \delta ( \sqrt{g} R_{abcd} R^{abcd} ) [/itex]

    and I know the answer is

    [itex] - \frac12 \sqrt{g} g_{\mu\nu}R_{abcd} R^{abcd} +\sqrt{g} R_{( \mu}{}^{bcd} R_{\nu ) bcd} + \ldots [/itex]

    what i do not understand is the coefficient of the last term. For example, when we evaluate the Maxwell Action

    [itex] \sqrt{g} F_{ab} F^{ab} [/itex]

    what we do is to write down as

    [itex] \sqrt{g} g^{\mu\nu} g^{\alpha\beta} F_{\mu\alpha} F_{\nu\beta}[/itex]

    so when we vary the action, we get

    [itex] -\frac12 \sqrt{g} g_{\mu\nu} F^2 + 2 \sqrt{g} F_{(\mu}{}^{\sigma} F_{\nu ) \sigma}[/itex]

    why is it not working with Riemann Tensor? How come there is no factor of 4 on the front of the last term in the variation of Riemann squared action?

    Thanks in advance
  2. jcsd
  3. Dec 7, 2012 #2


    User Avatar
    Science Advisor

    My reference (DeWitt's lectures) does have a factor of 4. Plus there are several terms involving the Ricci tensor.
  4. Dec 7, 2012 #3
    do you have a link for these lecture notes?
  5. Dec 7, 2012 #4


    User Avatar
    Science Advisor

    Sorry, my mistake, the additional factor is 2 not 4. The reference is "Dynamical Theory of Groups and Fields", which is apparently not available online, although many libraries have it. Although DeWitt claims the calculation is easy, it is not! So let me quote his result in full (He must be using the opposite sign convention):

    L3 ≡ √g RμνστRμνστ
    δL3/δgμν = √g (4Rμνσ -2R;μν -2RμστρRνστρ +½gμνRστρλRστρλ -4RμσντRστ +4RμσRνσ)

  6. Dec 7, 2012 #5
    You are absolutely right about the factor of 2! i am sorry for the typo. but i still do not understand why the factor is not 4, but 2?
  7. Dec 7, 2012 #6


    User Avatar
    Science Advisor

    Without plowing through the algebra, only a vague comment. When you write FμνFμν = gμαgνβFμνFαβ you know you've factored out all the g-dependence, and it's easy to see there are two g's. But when you write RμνστRμνστ = gμαgνβgσγgτδRμνστRαβγδ, the R's still contain g-dependence, so it's not going to be just a factor of four.
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook