-2
$\begingroup$

$$[X^{\mu\nu}] = [X] = \begin{pmatrix} 2 & 3 & 1 & 0 \\ 1 & 1 & 2 & 0 \\ 0 & 1 & 1 & 3 \\ 1 & 2 & 3 & 0 \end{pmatrix}$$

How to compute the trace of $X^{\mu \nu}$? I mean, by eye I can see it is $4$, but I don't know how to do it by writing with the metric tensor.

I calculated for example

$$X^{\mu}_{\phantom{a}\nu}= X^{\mu\sigma}\eta_{\sigma\nu} $$

And the result is

$$X^{\mu}_{\phantom{a}\nu}= \left( \begin{array}{cccc} 2 & -3 & -1 & 0 \\ 1 & -1 & -2 & 0 \\ 0 & -1 & -1 & -3 \\ 1 & -2 & -3 & 0 \\ \end{array} \right)$$

Its trace is zero. How can I obtain instead the trace of $X^{\mu\nu}$, which is $4$?

Also I understood that $X_{\mu\nu}$ would be, in terms of matrices, the inverse of $X^{\mu\nu}$. But when I compute it I have

$$X_{\mu\nu} = X^{\mu\nu} g_{\mu\alpha} g_{\nu\beta}$$

But $g_{\mu\alpha} g_{\nu\beta}$ is the identity matrix.

Yet the inverse of $X^{\mu\nu}$ is not itself.

$\endgroup$

1 Answer 1

3
$\begingroup$

I think you are confusing contravariant, covariant, and mixed tensors. The trace of $X^{\mu \nu}$ as is written in matrix form above is $\sum_{\mu=1}^4 X^{\mu \mu} = 4$. On the other hand the trace of $X^\mu_{\;\nu}$ is when you have $\mu = \nu$ which is $X^\mu_{\;\mu} = 0$.

I think the confusion stems from interpreting tensor indices as like the matrix indices and then also interpreting contraction in the same sense. The trace of a tensor $X$ is defined as $X^\mu_{\;\mu}$, not the sum of the diagonal elements of $X^{\mu \nu}$.

As for the inverse, it is the same confusion as I have stated. It just so happened that the Minkowski metric $\eta_{\mu \nu}$ is inverse to $\eta^{\mu \nu}$. Nobody said that if you contract a tensor with the metric it will give you the inverse. Again, contraction is a tensor operation NOT necessarily some matrix operation that will give you the inverse. In fact, lowering the index of a contravariant tensor $X^{\mu \nu}$ gives you the covariant tensor $X_{\mu \nu}$. I'm not sure where you got the idea that the covariant tensor is the inverse of the contravariant tensor. These are in a sense different representations of the tensor $X$. If you want the inverse, better calculate it in the standard approach.

$\endgroup$
5
  • $\begingroup$ Oh, thank you for this. Now it's really clear. Can you also help me with the inverse? Maybe youcan enlighten me in the same waY :D $\endgroup$
    – Heidegger
    Commented Dec 30, 2023 at 17:50
  • $\begingroup$ Thank you so much! Just a final question: I understood that to get $X_{\mu\nu}$ I use the metric tensor twice. Now the question is: since the metric tensor applied twice is just like the identity, wouldn't every $X^{\mu\nu}$ be the same as $X_{\mu\nu}$? $\endgroup$
    – Heidegger
    Commented Dec 30, 2023 at 18:03
  • $\begingroup$ @Numb3rs Again, you are confusing tensor operations and the matrix representation of $g_{\mu \nu}$. Lowering and raising indices are tensor operations. Besides $g_{\mu \alpha} g_{\nu \beta}$ is not the identity. What you are saying about the identity is $g_{\mu \alpha} g^{\alpha \nu} = \delta^\nu_\mu$. $\endgroup$
    – mathemania
    Commented Dec 30, 2023 at 18:07
  • $\begingroup$ Then I am a bit confused: using the matrix to represent those tensors, I would have that $X_{\mu\nu} = X^{\mu\nu} g_{\mu\alpha} g_{\nu\beta}$ reads $$[X] = [X][g][g]$$ but in the matrix representation $[g][g]$ gives the identity $\endgroup$
    – Heidegger
    Commented Dec 30, 2023 at 18:10
  • 1
    $\begingroup$ @Numb3rs I see, you forgot to look at the indices of $g_{\mu \nu}$ when interpreting it as matrices! Remember when multiplying matrices you have to align the indices properly so that what you should have is $[X] = [g^T] [X] [g]$ where T is transpose. Try comparing this with $[X] = [X] [g] [g]$, you'll get different results since matrices don't necessarily commute! $\endgroup$
    – mathemania
    Commented Dec 30, 2023 at 18:14

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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