0
$\begingroup$

Notation: In the following $*$ is the hodge operator from $\Lambda^1(\mathbb R^{1\times 3})\cong \mathbb R^{1\times 3}$ to $\Lambda^2(\mathbb R^{1\times 3})\cong A\subset\mathbb R^{3\times 3}$ (or its inverse), where $A$ is the subspace of antisymmetric matrices.

I would like to check whether the assumption that the electromagentic tensor $F$ transforms like a tensor implies that the magnetic field $B$ transforms like a pseudo vector. $\newcommand{\hodge}{{*}}$ The electromagnetic tensor w.r.t. to a chart $\phi$ is given by an antisymmetric matrix \begin{equation} F_\phi=\left[ \begin{array}{c|c} 0 & -E^t \\ \hline E & \mathscr B \end{array} \right]\in\mathbb R^{4\times 4} \end{equation} with $\mathscr B=*(B^t)\in\mathbb R^{3\times 3}$ and we assume that $F_\psi=M^tF_\phi M$ where $M\in\mathbb R^{4\times 4}$ is the matrix identified wtih $D(\phi\circ\psi^{-1})\in L(\mathbb R^4,\mathbb R^4)$. For the case $$M=\left[ \begin{array}{c|c} 1 & 0 \\ \hline 0 & O^t \end{array} \right]$$ with $O\in\mathbb R^{3\times 3}$ orthogonal this reduces to \begin{equation} F'=\left[\begin{array}{c|c} 0 & -E^tO^t \\ \hline OE & O\mathscr BO^t \end{array} \right] \end{equation} We immediately see that $E$ transforms like a vector, but it is not obvious that $B$ transforms like a pseudo vector. Thus, we want to prove the following implication: \begin{equation} \mathscr B'=O\mathscr BO^t\Rightarrow B'=(\det O)OB \end{equation} This implication is an immediate consequence of the following equation: \begin{equation} *(O\mathscr BO^t)=(\det O)B^tO^t \end{equation} Can someone give me a hint regarding how to prove the last equation?

$\endgroup$

2 Answers 2

5
$\begingroup$

You use rather a lot of complicated notation to express the simple fact that $$ B_i = \frac 12 \epsilon_{ijk}F^{jk}. $$ Now $F$ is a tensor, the Levi-Civita symbol is a pseudo-tensor (changes sign under orthogonal transformations $O$ that have ${\rm det} O=-1$), so their contracted product $B$ inherits the sign-change properties, making it a psudovector.

$\endgroup$
3
  • $\begingroup$ The sums over $j$ and $k$ are from $1$ to $3$ and not from $0$ to $3$, right? $\endgroup$
    – Filippo
    Commented Sep 8, 2022 at 15:07
  • $\begingroup$ Yes. You can enforce this by using $\epsilon_{0ijk}$. The result remains true $\endgroup$
    – mike stone
    Commented Sep 8, 2022 at 15:34
  • $\begingroup$ "Now $F$ is a tensor..." - By assumption the $4\times 4$-matrix $F$ transforms like a tensor, but it is not obvious that the $3\times 3$-matrix $F=∗(B^t)=:\mathscr B$ transforms like a tensor as well, is it? But this is precisely what I have proven in my question, so maybe the "complicated notation" turned out to be useful :) Also note that I did not make up this notation, it is introduced in Theodore Frankel's "The geometry of physics" and turns out to be useful for the reformulation of the Maxwell equations in terms of $F$ (see section 7.2b in the book). $\endgroup$
    – Filippo
    Commented Sep 9, 2022 at 9:52
0
$\begingroup$

As @mikestone suggested, we can complete the proof by using the fact that the Levi-Civita-symbol transforms like a pseudo-tensor, i.e. $\epsilon_{klj}\det O=\epsilon_{pqr}O^p{}_kO^q{}_lO^r{}_j$: \begin{align} &\forall i:\det O(B^tO^t)_i=\det OB_jO^i{}_j=\det O\epsilon_{klj}\mathscr{B}_{kl}O^i{}_j&\\ &=\epsilon_{pqr}O^p{}_kO^q{}_lO^r{}_j\mathscr{B}_{kl}O^i{}_j=\underbrace{O^i{}_jO^r{}_j}_{=\delta_{ir}}\epsilon_{pqr}O^p{}_kO^q{}_l\mathscr{B}_{kl}&\\ &=\epsilon_{pqi}O^p{}_k\mathscr{B}_{kl}O^q{}_l=\epsilon_{pqi}(O\mathscr{B}O)^p{}_q=*(O\mathscr{B}O^t)_i& \end{align}

$\endgroup$

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.