I'm trying to show that a cross product $C$ of two vectorial operators (let's say $A$ and $B$) it's a vector by it's own, which means, I want to show $$\left [J_i,C_j \right ] = i\hbar \epsilon_{ijk}C_k.$$

Attempt of solution:

$A$ and $B$ are vectorial operators, so $$\left [J_i,A_j \right ] = i\hbar \epsilon_{ijk}A_k,$$ $$\left [J_i,B_j \right ] = i\hbar \epsilon_{ijk}B_k.$$

$C$ is the cross product of $A$ and $B$, so $$C_i = \epsilon_{ijk}A_jB_k$$ $$C_j = \epsilon_{jik}A_iB_k$$ $$C_k = \epsilon_{kij}A_iB_j.$$

Building the commutator of $J_i$ and $C_j$, I found this: $$\left [ J_i,C_j \right ] = \left [ J_i,\epsilon_{ijk}A_iB_k \right ] = \epsilon_{ijk}(\left [ J_i,A_i\right ]B_k + A_i\left [ J_i,B_k\right ]).$$

The term with $\left [ J_i,A_i\right ]$ is zero, so we have $$\left [ J_i,C_j \right ] = \epsilon_{ijk}(A_i\left [ J_i,B_k\right ]) = \epsilon_{ijk} A_i (\epsilon_{ikj}B_j) = i\hbar\epsilon_{ikj}\epsilon_{ijk} A_iB_j = i\hbar \epsilon_{ikj}C_k = -i\hbar \epsilon_{ijk}C_k.$$

I was not supposed to find this minus signal, but I cannot find the first mistake in my calculations.

  • $\begingroup$ You're not writing out the correct expressions for $C_j$ and $C_k$, where are you getting those from? $\endgroup$
    – jacob1729
    Mar 21, 2020 at 18:56
  • $\begingroup$ Theres is a general statement that for any vector operator commutator will be $i\epsilon_{ijk}C_k$. You should just contract one index, $\epsilon_{ijk}\epsilon_{krs}$ and obtain the answer $\endgroup$ Mar 21, 2020 at 19:06
  • $\begingroup$ $C_j=\epsilon_{ijk}A_iB_k$ can be rewritten by changing alphabetical labels $C_i=\epsilon_{jik}A_jB_k = -\epsilon_{ijk}A_jB_k$ which contradicts your expression for $C_i=+\epsilon_{ijk}A_jB_k$. $\endgroup$
    – JEB
    Mar 21, 2020 at 19:22
  • 1
    $\begingroup$ What follows "I found this:" is worse than wrong: you are abusing i as both a free index and a saturated dummy index! Use another letter. $\endgroup$ Mar 21, 2020 at 21:55
  • $\begingroup$ Jacob, Artem and Cosmas, thank very much for your comments, I found out what was wrong. I posted the solution below $\endgroup$ Mar 21, 2020 at 22:56

2 Answers 2


Reading the comments of jacob1729, Artem Alexandrov and Cosmas Zachos, I found out the so many mistakes of my calculations. Here comes the complete solution in case anyone needs this proof in the future:

$C_j$ can be writen as $C_j = \epsilon_{jqr}A_qB_r$, so we have

$$[J_i,C_j] = [J_i,\epsilon_{jqr}A_qB_r] = \epsilon_{jqr} [J_i,A_qB_r] = \epsilon_{jqr} ([J_i,A_q,]B_r + A_q[J_i,B_r])$$

Using the comutation relations for A and B $$\left [J_i,A_q \right ] = i\hbar \epsilon_{iqp}A_p$$ $$\left [J_i,B_r \right ] = i\hbar \epsilon_{irs}B_s$$

we can found

$$[J_i,C_j] = \epsilon_{jqr} (i\hbar \epsilon_{iqp}A_pB_r + A_qi\hbar \epsilon_{irs}B_s) = i\hbar\epsilon_{qjr}\epsilon_{qip}A_pB_r - i\hbar\epsilon_{rjq}\epsilon_{ris}A_qB_s$$

Remembering the levi-civita "multiplication" identity, we obtain

$$[J_i,C_j] = i\hbar(\delta_{ji}\delta_{rp}-\delta_{jp}\delta_{ri})A_pB_r - i\hbar(\delta_{ji}\delta_{qs}-\delta_{js}\delta_{qi})A_qB_s = i\hbar(A_rB_r - A_jB_i-A_sB_s+A_iB_j)$$

Using the Einstein Sum Notation we can see the equality between $A_rB_r$ and $A_sB_s$, and because of this, these terms annihilate each other.

In the end of the day, we have this:

$$[J_i,C_j] = i\hbar(A_iB_j- A_jB_i) = i\hbar\epsilon_{ijk}\epsilon_{klm}A_lB_m = i\hbar\epsilon_{ijk}C_k$$


There is the general statement about commutator between momenta operators, $J_i$ with scalar, vector and tensor operators. In your case the answer is $$[J_i,C_j]\sim\epsilon_{ijk}C_k,$$ where you substitute $C_k=\epsilon_{kab}A_aB_b$, so $$\epsilon_{ijk}\epsilon_{kab}A_aB_b=\epsilon_{ijk}\epsilon_{abk}A_aB_b=(\delta_{ia}\delta_{jb}-\delta_{ib}\delta_{ja})A_aB_b=A_iB_j-A_jB_i.$$

I am not sure what do you want to show, but let me scetch the derivation of general fact, $[J_i,C_j]=\epsilon_{ijk}C_K$, where $C_j$ is an arbitrary vector operator. Consider coordinate system $r'$ and coordinate system $r$, which differs in rotation, $R_{\phi}$. Two coordinate systems relates to each other by the relation $$r'=R_{\phi}r,$$ where $R_{\phi}$ is rotation matrix (and it will be written explicitly below). Then consider an arbitrary vector operator $\hat{C}$. Its components should transform by the similar relation, $$\hat{C'}=e^{-i\phi(n\cdot J)}\hat{C}e^{+i\phi(n\cdot J)},$$ which comes from the fact that operator $J$ describes rotations. For small angle $\delta\phi$, we can write down $$\hat{C'}\approx (1-i\delta\phi(J\cdot n))\hat{C}(1+i\delta\phi(J\cdot n))\approx\hat{C}-i\delta \phi [(n\cdot J),\hat{C}]\approx\hat{C}-\delta\phi[n\times\hat{C}],$$ where the last term comes from the Euler relation for rotation by small angle and this term can be rewritten in tensor denotations as $$n_i[J_i,\hat{C}_j]=-ie_{kil}n_i\hat{C}_l=i\epsilon_{ikl}n_i\hat{С}_l,$$ where you should understand unit vector $n_i$ as a direction of rotation. The last expression gives exactly the desired general statement for an arbitrary vector operator. Hope that this will help.

  • $\begingroup$ I want to show exactly the first statement. I tried to write $C_j = \epsilon_{jab}A_aB_b$ and plug this in the statament $\endgroup$ Mar 21, 2020 at 19:19
  • $\begingroup$ As I understand, you want to show that for any vector operator $C_j$ commutator will be $[J_i,C_j]=\epsilon_{ijk}C_k$? $\endgroup$ Mar 21, 2020 at 21:17
  • $\begingroup$ Exactly! I want to show that $C_j$ obeys this statement $\endgroup$ Mar 21, 2020 at 22:25
  • $\begingroup$ @CauêEvangelista I have updated the answer and hope that it will help. $\endgroup$ Mar 21, 2020 at 23:13

Your Answer

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

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