6

I'm asked to calculate the curvature Riemann and Ricci tensors and the curvature scalar of a metric that comes from a conformal transformation from a flat metric, that is, of a metric $g'=\Omega^2g$ where $g$ is flat. Now, I've been able to calculate the Riemann tensor of the $g'$ metric in terms of the other one, getting to this result:

$$R'{_{abc}}^d=R{_{abc}}^d+2C{_{c[a}}^d\nabla_{b]}+2\nabla_{[a}C{_{b]c}}^d+2C{_{c[a}}^eC{_{b]e}}^d$$

Where $R{_{abc}}^d$ are the components of the Riemann tensor in the $g$ metric.

Now for the Ricci, I just contract $b$ and $d$ indices in both sides and the same for the curvature scalar. I don't know though, if the $g$ metric is flat, what to get besides those expressions.

Thank you in advance.

EDIT:

I calculated that by using $R{_{abc}}^d\omega_d=(\nabla_a\nabla_b-\nabla_b\nabla_a)\omega_c$ for $1$-forms $\omega$. And $C{_{abc}}^d$ is the components of the difference tensor of both cov. derivatives. Now, obviously, the terms that go like $C\nabla$ must vanish because it wouldn't be a linear operator. That means that it must be tru for $1$-forms that

$$C{_{c[a}}^d\nabla_{b]}\omega_d$$

must be $0$, but I can't see why.

MyUserIsThis
  • 3,702
  • 1
  • 24
  • 39
  • 1
    If the metric is flat, your curvature tensor should be $0$ in the first place, and the contractions are still $0$ :) – Ted Shifrin Jun 03 '14 at 20:50
  • No, but the curvature tensor $R$ is $0$, and I want to calculate $R'$- – MyUserIsThis Jun 03 '14 at 21:14
  • 2
    Can you, please, explain your formula in the display: it looks wrong to me. What is $C_{a b}{}^c$? Also, please read this answer where I give the calculation in the classical invariant notation. – Yuri Vyatkin Jun 04 '14 at 00:49
  • In this question you can see the correct identities that you are after. In my answer to it you can find further references that you should be aware if you want to progress in this topic. – Yuri Vyatkin Jun 04 '14 at 06:11
  • I have posted my answer that shows that your formula is correct up to a weird second term $2C{{c[a}}^d\nabla{b]}$ in the right hand side, which obviously makes no sense (the curvature is a linear object!). – Yuri Vyatkin Jun 04 '14 at 06:44
  • The terms $C \nabla$ go away because of the symmetries that one can see in an explicit calculation. I have expanded my answer to show this. – Yuri Vyatkin Jun 06 '14 at 03:45

1 Answers1

5

As far as I can see, the OP uses the difference tensor $$ C_{a b}{}^c := - \delta_a{}^c \Upsilon_b - \delta_b{}^c \Upsilon_a + \Upsilon^c g_{a b} $$ that appears in the expression $$ \widehat{\nabla}_a \omega_b = \nabla_a \omega_b + C_{a b}{}^c \omega_c $$ where $\nabla_a$ and $\widehat{\nabla}_a$ are the Levi-Civita connections of the metrics $g_{a b}$ and ${\widehat{g}}_{a b} = \Omega^2 g_{a b}$ respectively. Here $\Upsilon_a := \nabla_a \log \Omega$, that is the same thing as $\mathrm{d} \omega$ in the settings of this answer. The difference formula in the above display is just a restatement of the equation (1) in that answer in the abstract index notation.

The formula for the curvature $\widehat{R}$ of $\widehat{g}_{a b}$ should then look as $$ {\widehat{R}}_{a b c}{}^d = R_{a b c}{}^d + 2 \, \nabla_{[a} C_{b] c}{}^d + 2 \, C_{[a| c}{}^e C_{e| b]}{}^d $$ which is a direct application of the general difference formula:

Theorem (The difference formula for curvatures). Let $\nabla$ and $\nabla'$ be two arbitrary connections in the tangent bundle of manifold $M$. Let $K$ and $K'$ be the curvatures of the connections $\nabla$ and $\nabla'$ respectively, and $A$ be the difference tensor of the pair $(\nabla',\nabla)$, that is $ \nabla' = \nabla + A$. Then $$ K' = K + \nabla \wedge A + A \wedge A $$

Proof. For the sake of simplicity I will do the calculation for the case of torsion-free connections $\nabla$ and $\nabla'$ that is sufficient for our purposes (we deal with the Levi-Civita connections in the main question). In this situation the curvature $K$ of the connection $\nabla$ satisfies the identity $K_{a b c}{}^d \omega_d = 2\,\nabla_{[a} \nabla_{b]} \omega_c$, as we discussed here.

Writing everything out using the definitions, we get $$ \begin{align} K'_{a b c}{}^d \omega_d & = \nabla'_a \nabla'_b \omega_c - (a \leftrightarrow b) \\ & = \nabla'_a (\nabla_b \omega_c + A_{b c}{}^d \omega_d ) - (a \leftrightarrow b) \\ & = \nabla'_a \nabla_b \omega_c + (\nabla'_a A_{b c}{}^d) \omega_d + A_{b c}{}^d \nabla'_a \omega_d - (a \leftrightarrow b) \\ & = \nabla_a \nabla_b \omega_c + A_{a b}{}^d \nabla_d \omega_c + A_{a c}{}^d \nabla_b \omega_d \\ & \quad + (\nabla_a A_{b c}{}^d + A_{a b}{}^e A_{e c}{}^d + A_{a c}{}^e A_{b e}{}^d - A_{a e}{}^d A_{b c}{}^e) \omega_d \\ & \quad + A_{b c}{}^d ( \nabla_a \omega_d + A_{a d}{}^e \omega_e ) - (a \leftrightarrow b) \end{align} $$

Collecting the terms, we rewrite the above equation as $$ \begin{align} K'_{a b c}{}^d \omega_d & = \nabla_a \nabla_b \omega_c + \underbrace{A_{a b}{}^d \nabla_d \omega_c}_{\text{sym. in a, b}} + \underbrace{A_{a c}{}^d \nabla_b \omega_d + A_{b c}{}^d \nabla_a \omega_d}_{\text{sym. in a, b}} + \underbrace{A_{b c}{}^d A_{a d}{}^e \omega_e}_{\text{cancel}} \\ & \quad + (\nabla_a A_{b c}{}^d) \omega_d + \underbrace{A_{a b}{}^e A_{e c}{}^d \omega_d}_{\text{sym. in a, b}} + A_{a c}{}^e A_{b e}{}^d \omega_d - \underbrace{A_{a e}{}^d A_{b c}{}^e \omega_d}_{\text{cancel}} - (a \leftrightarrow b) \end{align} $$

The underbraced terms vanish for the reasons indicated in the subscripts, so we obtain the equality $$ K'_{a b c}{}^d \omega_d = 2 \, \nabla_{[a} \nabla_{b]} \omega_c + 2 \, (\nabla_{[a} A_{b] c}{}^d) \omega_d + A_{[a| c}{}^e A_{|b] e}{}^d \omega_d $$ which is the explicit form of the claim. QED.

A detailed account of this topic can be found in J.Slovak's dissertation here.

Yuri Vyatkin
  • 11,279
  • Hi, thanks for you help. I edited my question with further information. – MyUserIsThis Jun 05 '14 at 22:27
  • @MyUserIsThis What you've written in your edit to the question does not add any new information. It just shows that you are insisting on your formula, which is wrong in the term "$C \nabla$" that must not appear. I gave to you the precise reason in my answer, so you either address what I said or, please, post new questions, if necessary. Posting lecture notes here is beyond my current ability, sorry :-) – Yuri Vyatkin Jun 06 '14 at 01:54
  • That calculation was exactly what I needed. It's clear now, thank you. – MyUserIsThis Jun 06 '14 at 22:23
  • @YuriVyatkin Thanks so much. Your computation also sort of proves the general formula you mentioned, right? Thanks again. – Matha Mota Sep 30 '22 at 06:17
  • 1
    @BokaPeer Yes, it proves the general fact. – Yuri Vyatkin Oct 01 '22 at 06:51