1

I am following Pavel Grinfield's Tensor Calculus book, in page-137, he shows this remarkable equation:

$$ \epsilon_{rst} \cdot \epsilon^{ijk}=d_{rst}^{ijk} = \begin{vmatrix} \delta_r^i & \delta_s^i & \delta_t^i \\ \delta_r^j & \delta_s^j& \delta_t^j\\ \delta_r^k &\delta_s^k & \delta_t^k \end{vmatrix}$$

This is the definition of determinant given in his book:

$$ |A| = \delta_{rst}^{ijk} \frac{a_i^r a_j^s a_k^t}{3!}$$

Note: $\epsilon_{ijk}$ and $\epsilon_{rst}$ are permutation symbols.

Here is my attempt at proving the fact:

Call: $$B=\begin{bmatrix} \delta_r^i & \delta_s^i & \delta_t^i \\ \delta_r^j & \delta_s^j& \delta_t^j\\ \delta_r^k &\delta_s^k & \delta_t^k \end{bmatrix}$$

Then,

$$|B|= \frac{1}{3!}\delta_{rst}^{ijk} (\delta_{p(i)}^{p(r)}\delta_{p(j)}^{p(s)}\delta_{p(k)}^{p(t)})$$

Where $p$ is a function which maps the index set in the following way : $p(1)=i, p(2)=j,p(3)=k$.. now I don't get how to simplfy..

Edit: Ok, everyother exercise in the proceeding section is based on this..

Related

  • Wow, that equation is a mess (the author's, not yours). Seriously that is the ugliest definition of determinant I've ever seen. What is the range of values allowed for the subscripts and superscripts? – RobertTheTutor May 05 '21 at 21:49
  • @RobertTheTutor $i,j,k$ can take values in ${1,2, 3 }$. Though I have to say, even if it is a mess, every other delta system identity can be made from it (as shown be the exercise right after), so it's crazy powerful as well – tryst with freedom May 05 '21 at 21:50
  • 729 terms to sift through, then. Most of which are zero, but still. You can whittle it down to "only" 36 terms on the left side, but on the right? I don't see how you could even begin to simplify. – RobertTheTutor May 05 '21 at 21:54
  • Yeah brute forcing doesn't seem like a good idea :o – tryst with freedom May 05 '21 at 21:56

1 Answers1

1

An overview of how we can avoid brute force:

We can write $B^a_b=\delta^{g(a)}_{h(b)}$ with $g(1)=i$ etc., so$$\det B=\epsilon_{ace}\epsilon^{bdf}B^a_bB^c_dB^e_f=\epsilon_{ace}\epsilon^{bdf}\delta^{g(a)}_{h(b)}B^{g(c)}_{h(d)}B^{g(e)}_{h(f)}=\epsilon_{ace}\epsilon^{j(a)j(c)j(e)},\,j(k):=h^{-1}(g(k)).$$The antisymmetries of this expression, together with its behaviour in the case $a=1,\,c=2,\,e=3$, finishes the job.

J.G.
  • 115,835
  • Could you explain the last idea with the $h^{-1}(g(k) )$ thing? – tryst with freedom May 06 '21 at 10:57
  • @Buraian Since $h$ is invertible, in terms of Iverson brackets$$\delta^{g(a)}_{h(b)}=[g(a)=h(b)]=[j(a)=b]=\delta^{j(a)}_b$$etc. Finally, $\epsilon^{bdf}\delta^{j(a)}_b=\epsilon^{j(a)df}$ etc. – J.G. May 06 '21 at 11:25
  • Wow, the Iverson bracket thing is powerful. I couldn't pick up on what it means to use a bracket on an equality like[g(a)=h(b)] , could you explain that in another way? – tryst with freedom May 06 '21 at 13:47
  • @Buraian The statement $g(a)=h(b)$ is equivalent to the statement $j(a)=b$. Therefore, these statements are either both true or both false. In the former case, they each have Iverson bracket $1$; in the latter case, they each have Iverson bracket $0$. In general, $[u=v]=\delta_{uv}$ allows us to define the Kronecker delta in terms of the Iverson bracket. – J.G. May 06 '21 at 13:53
  • Oh so the [g(a) = h(b)] is an object in itself but why did you bring in the 'h'? wouldn't everything be doable in terms of g and j? – tryst with freedom May 06 '21 at 13:57
  • 1
    Ok I think I got it, you did it so oyu could get the contraction directly – tryst with freedom May 06 '21 at 14:04