Physics Stack Exchange is a question and answer site for active researchers, academics and students of physics. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I know this is an already answered question, but I couldn't make head or tail of it, and it's bugging me. I know I'm probably asking a silly question, but please bear with me as I'm 14 and this is my first post.

I saw this perturbation while studying up a bit of linearised gravity:

$$g_{\mu\nu} = \eta_{\mu\nu} + h_{\mu\nu} \tag{1}$$

and then the next line says: " you raise the indices to get":

$$g^{\mu\nu} = \eta^{\mu\nu} - h^{\mu\nu}. \tag{2}$$

I have done a bit of tensor calculus, but I still couldn't understand how the obvious fact $g_{\mu\nu}\, g^{\mu\rho} = \delta^{\rho}_{\nu}$ leads to the second equation. I tried everything I could, but all I get is the first equation with raised indices different from the given ones, and a persistent + sign. Please explain and derive the second equation step by step, as if to an idiot.

share|cite|improve this question
Where did you see this question answered on the site? – JamalS May 19 '14 at 10:23
up vote 9 down vote accepted

Note that: $$ h^{\mu \nu} = \eta^{\mu \rho}\eta^{\nu \lambda} h_{\rho \lambda} $$ Therefore, up to first order, we have: \begin{equation} \begin{aligned} g^{\mu \nu}g_{\nu \sigma} & = (\eta^{\mu \nu} - h^{\mu \nu})(\eta_{\nu \sigma} + h_{\nu \sigma}) \\& =\eta^{\mu \nu}\eta_{\nu \sigma} + \eta^{\mu \nu}h_{\nu \sigma} - \eta_{\nu \sigma} h^{\mu \nu} + \mathcal{O}(h^2) \\& = \delta^\mu_\sigma + \eta^{\mu \nu}h_{\nu \sigma} - \eta_{\nu \sigma} \eta^{\mu \rho}\eta^{\nu \lambda} h_{\rho \lambda} + \mathcal{O}(h^2) \\& = \delta^\mu_\sigma + \eta^{\mu \nu}h_{\nu \sigma} - \delta_\sigma^\lambda \eta^{\mu \rho} h_{\rho \lambda} + \mathcal{O}(h^2) \\& = \delta^\mu_\sigma + \eta^{\mu \nu}h_{\nu \sigma} - \eta^{\mu \rho} h_{\rho \sigma} + \mathcal{O}(h^2) \\& = \delta^\mu_\sigma + \eta^{\mu \nu}h_{\nu \sigma} - \eta^{\mu \nu} h_{\nu \sigma} + \mathcal{O}(h^2) \\& = \delta^\mu_\sigma + \mathcal{O}(h^2) \end{aligned} \end{equation} which is what we require.

Edit in response to comments. The inverse of $g_{\nu \sigma}$, which is denoted by $g^{\mu \nu}$, is defined to satisfy the following equation: $$ g^{\mu \nu}g_{\nu \sigma} = \delta^\mu_\sigma $$ In other words, we need to find an expression for $g^{\mu \nu}$ such that the following equations is satisfied: $$ g^{\mu \nu}(\eta_{\nu \sigma} + h_{\nu \sigma}) = \delta^\mu_\sigma $$ As I have shown above, the function that obeys the above equation up to first order in $h_{\mu \nu}$ is: \begin{equation} g^{\mu \nu} = \eta^{\mu \nu} - h^{\mu \nu} \end{equation} That is really all that is going on.

share|cite|improve this answer
Hunter, I do understand what you mean, but that's not what I asked. I asked how do we get the second equation from the first used equation 2 in the first step itself. I only wanted to know how HOW do you get that....But thanks anyway!! – GRrocks May 19 '14 at 10:44
@GRrocks that makes no sense to me. What source are you studying from? – Hunter May 19 '14 at 10:45
@GRrocks It is appears to be the exact same situation as equations (4.13) and (4.14). – Hunter May 19 '14 at 10:57
Ahh...I get it now..Thanks a lot @Hunter..I guess I was stupid...but you showed earlier that $\eta^{\mu\nu}$ -$h^{\mu\nu}$ is the required function, so its more like trial and error. Can we not take an arbitrary tensor and show that to get equation 2 of my question ( i just edited the numbering of equations), that arbitrary tensor has to be $\eta^{\mu\nu}$ -$h^{\mu\nu}$ ???apologies if im being foolish again.. – GRrocks May 19 '14 at 11:17
@GRrocks: well, the only 100% rigorous way to do what you ask is to write down $h_{ab}$ in components, invert the matrix, and drop the second order terms. Then, you can extract out the equation you want. But that's obviously a ton of work. – Jerry Schirmer May 19 '14 at 16:32

If you have $g_{\mu\nu} = \eta_{\mu\nu} + h_{\mu\nu}$, with $|h_{\mu\nu}|\ll 1$ a perturbation, $\eta_{\mu\nu} = \text{diag}(-1,1,1,1)$, then you can simply perform a Taylor expansion to obtain the inverse, $$g^{\mu\nu} = \eta^{\mu\nu} - h^{\mu\nu}+ \mathcal{O}(h^2),$$ where the indices of $h^{\mu\nu}$ are raised by $\eta^{\mu\nu}$. This is easy to see if you take the diagonal elements first, the off-diagonal ones are of second order, negligible. You don't need to solve a matrix equation for this, because the inverse is linearly dependent on $\eta_{\mu\nu}=\eta^{\mu\nu}$, $h_{\mu\nu}$ and their products, hence it must be of the form $g^{\mu\nu} = a\eta^{\mu\nu} + b h^{\mu\nu} + \ldots$, where the ellipses imply linear dependencies of higher order and the coefficients are constants. Also, note that $h_{\mu\nu} \eta^{\mu\mu} \eta^{\nu\nu} = h^{\mu\nu}$, no summation implied, namely $h^{\mu\nu} \propto h_{\mu\nu}$.

Also, a straightforward way to calculate $\delta g^{\mu\nu}$ is by use of $\delta g^{\mu\nu} = - g^{\mu\rho} g^{\nu\sigma} \delta g_{\rho\sigma}$ with $\delta g_{\mu\nu} = h_{\mu\nu}$.

share|cite|improve this answer
In general relativity, the demand $|h_{\mu\nu}| \ll 1$ is not enough to guarantee that the perturbation is small. Normally you treat case by case to determine what 'small' means for your particular background. – JamalS May 19 '14 at 16:27
@JamalS If we have a background described by $\eta_{\mu\nu}$, doesn't $|h_{\mu\nu}|\ll 1$ define smallness unambiguously? – auxsvr May 19 '14 at 16:30
For the Minkowski metric, of course, but not for a general background. I'm sorry if it seemed I was challenging your answer; the comment was meant as an addendum for future viewers. – JamalS May 19 '14 at 16:34
I assume that the OP is attempting to derive the newtonian limit, otherwise none of the answers makes sense. – auxsvr May 19 '14 at 17:03

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.