I'm to calculate the commutator of the following operators : $\mathbf{\widehat{r}}=\mathbf{e}_{x}x+\mathbf{e}_{y}y+\mathbf{e}_{z}z$ and $\mathbf{\widehat{p}}=-i\hbar\left ( \mathbf{e}_{x}\frac{\partial }{\partial x}+\mathbf{e}_{y}\frac{\partial }{\partial y}+\mathbf{e}_{z}\frac{\partial }{\partial z} \right )$. This is what I have attempted : $$\left [ \mathbf{\widehat{r}},\mathbf{\widehat{p}} \right ]\varphi=-i\hbar\left ( \mathbf{e}_{x}x +\mathbf{e}_{y}y+\mathbf{e}_{z}z\right )\left ( \mathbf{e}_{x}\frac{\partial \varphi }{\partial x}+\mathbf{e}_{y}\frac{\partial \varphi }{\partial y}+\mathbf{e}_{z}\frac{\partial\varphi }{\partial z} \right )+i\hbar \left ( \mathbf{e}_{x}\frac{\partial }{\partial x}+\mathbf{e}_{y}\frac{\partial }{\partial y}+\mathbf{e}_{z}\frac{\partial }{\partial z} \right )\left ( \mathbf{e}_{x}x \varphi +\mathbf{e}_{y}y\varphi +\mathbf{e}_{z}z\varphi \right )=-i\hbar\left ( x\frac{\partial \varphi }{\partial x}+y\frac{\partial\varphi }{\partial y}+z\frac{\partial \varphi }{\partial z} \right )+i\hbar\left (x\frac{\partial \varphi }{\partial x}+y\frac{\partial\varphi }{\partial y}+z\frac{\partial \varphi }{\partial z}+3\varphi \right )=3i\hbar \varphi \Rightarrow \left [\mathbf{\widehat{r}},\mathbf{\widehat{p}} \right ]=3i\hbar$$ What confuses me is that $\mathbf{\widehat{r}}\varphi$ is a vector valued function, so I cannot calculate the gradient $\nabla\left ( \mathbf{\widehat{r}\varphi} \right )$, so instead I took the divergence $\nabla\cdot \left ( \mathbf{\widehat{r}\varphi} \right )$ to obtain the result above. Is commutator a scalar? Judging by all the examples I've seen it is, but nowhere has this point been made explicit.

  • 1
    $\begingroup$ You shouldn't take the divergence. Instead you should study the components $[\hat r_i, \hat p_j]$ where $i,j=x,y,z.$ That is, $[\hat{\mathbf{r}}, \hat{\mathbf{p}}]$ could be thought of as a matrix. The result should be $[\hat r_i, \hat p_j] = i\hbar \, \delta_{ij}.$ $\endgroup$ – md2perpe Sep 15 '18 at 10:13

Obvious it is much easier to just write down the CCR in components$^1$

$$ [\hat{x}^j,\hat{p}_k]~=~i\hbar \delta^j_k \hat{\bf 1} ,\tag{1}$$

but OP seems to ponder the following interesting question.

Question: How do we take the commutator of two vector-valued operators?

Answer: Well, here is one possible construction. Let us generalize a bit and consider two symmetric tensor-valued operators

$$ \hat{\bf T}~=~ \sum_{j_1,\ldots, j_r}{\bf e}_{j_1} \odot \ldots \odot{\bf e}_{j_r} \hat{T}^{j_1\ldots j_r} \tag{2}$$


$$ \hat{\bf S}~=~ \sum_{k_1,\ldots, k_s}{\bf e}_{k_1} \odot \ldots \odot{\bf e}_{k_s} \hat{S}^{k_1\ldots k_s}. \tag{3}$$ Define composition as $$\hat{\bf T}\circ \hat{\bf S} ~:=~ \sum_{j_1,\ldots, j_r,k_1,\ldots, k_s} {\bf e}_{j_1} \odot \ldots \odot{\bf e}_{j_r}\odot {\bf e}_{k_1} \odot \ldots \odot{\bf e}_{k_s} \hat{T}^{j_1\ldots j_r}\circ \hat{S}^{k_1\ldots k_s}.\tag{4}$$ Define the commutator as $$[\hat{\bf T}, \hat{\bf S}] ~:=~\hat{\bf T}\circ \hat{\bf S}- \hat{\bf S}\circ \hat{\bf T} ~=~\sum_{j_1,\ldots, j_r,k_1,\ldots, k_s} {\bf e}_{j_1} \odot \ldots \odot{\bf e}_{j_r}\odot {\bf e}_{k_1} \odot \ldots \odot{\bf e}_{k_s} [\hat{T}^{j_1\ldots j_r} , \hat{S}^{k_1\ldots k_s}]. \tag{5}$$

Now we can make sense to OP's question. The commutator (6) of two vector-valued operators is a 2nd-rank tensor-valued operator:

$$[\hat{\bf x}, \hat{\bf p}]~=~[\sum_j{\bf e}_j\hat{x}^j, \sum_k{\bf e}^k\hat{p}_k]~\stackrel{(1)+(5)}{=}~ \sum_j{\bf e}_j\odot {\bf e}^j~i\hbar \hat{\bf 1}.\tag{6}$$

Notice that the corresponding tensor components are just the CCR (1), so eqs. (1) & (6) are equivalent formulations. A textbook will usually only work in components to avoid having to introduce the notion of tensor operators (2).


$^1$ For the purpose of this answer, let us raise and lower 3D indices with a metric $g_{jk}$.

  • $\begingroup$ It took me a while to digest this answer but it turned out to be very helpful. Thank you. $\endgroup$ – Toshino Kyoko Sep 15 '18 at 10:52

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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