In quantum mechanics we have \begin{equation*} \langle x|p\rangle=C\exp\left(\frac{ipx}{\hbar}\right) \end{equation*} where $C$ is a normalization constant.

It follows that \begin{equation*} -i\hbar\frac{\partial}{\partial x}\bigl(\langle x|p\rangle\bigr) =-i\hbar C\frac{\partial}{\partial x}\exp\left(\frac{ipx}{\hbar}\right) =p\langle x|p\rangle \tag{1} \end{equation*}

However by the product rule we can also write \begin{equation*} -i\hbar\frac{\partial}{\partial x}\bigl(\langle x|p\rangle\bigr) =-i\hbar \left[ \left(\frac{\partial}{\partial x}\langle x|\right)|p\rangle +\langle x|\left(\frac{\partial}{\partial x}|p\rangle\right) \right] \tag{1.1} \end{equation*}

Rewrite as \begin{equation*} -i\hbar\frac{\partial}{\partial x}\bigl(\langle x|p\rangle\bigr) =-i\hbar\left(\frac{\partial}{\partial x}\langle x|\right) |p\rangle +p\langle x|p\rangle \tag{2} \end{equation*}

Then by equivalence of (1) and (2) we have \begin{equation*} -i\hbar\left(\frac{\partial}{\partial x}\langle x|\right) |p\rangle=0 \tag{3} \end{equation*}

Is this correct?

EDIT: No, it is not correct. The mistake is in equation (2). As pointed out in the accepted answer, the correct form of (2) is \begin{equation*} -i\hbar\frac{\partial}{\partial x}\bigl(\langle x|p\rangle\bigr) =-i\hbar\left(\frac{\partial}{\partial x}\langle x|\right) |p\rangle \tag{2'} \end{equation*} because \begin{equation*} \langle x|\left(\frac{\partial}{\partial x}|p\rangle\right)=0 \end{equation*}

It follows from (2') that \begin{equation*} -i\hbar\frac{\partial}{\partial x}\bigl(\langle x|p\rangle\bigr) =p\langle x|p\rangle \end{equation*} which is equivalent to equation (1) as required.

  • 1
    $\begingroup$ Your notation is very confusing (which may be the cause of your trouble). $\endgroup$ Sep 9, 2023 at 17:01
  • $\begingroup$ Careful when taking the derivative of a ket as you wrote it: this is like writing $\frac{\partial}{\partial x}\hat x$, which in itself makes no sense. After all, $\vert x\rangle$ is a unit vector. $\endgroup$ Sep 9, 2023 at 17:13

2 Answers 2


$i\hbar\,\partial/\partial x$ is the momentum operator in the position representation. It is not the momentum operator that acts in the Hilbert space of abstract kets like $|p\rangle$, so your eq. $(2)$ doesn't follow. There is simply no way to make sense of such an expression.

Rather, starting from $\hat{p}|p\rangle=p|p\rangle$, we have that

\begin{align*} \hat{p}|p\rangle=p\int{\rm d}x\,e^{ipx/\hbar}|x\rangle=\int{\rm d}x\,|x\rangle\left(-i\hbar\frac{\partial}{\partial x}\right)e^{ipx/\hbar}, \end{align*} so, from $e^{ipx/\hbar}=\langle x|p\rangle$, it follows that $$\hat{p}=\int{\rm d}x\,|x\rangle\left(-i\hbar\frac{\partial}{\partial x}\right)\langle x|$$

  • $\begingroup$ @TheLittleDogLaughed my suggestion would be to go back to the definition of the derivative: ${\rm d}f/{\rm d}x=\lim_{h\to 0}(f(x+h)-f(x))/h$. Can you recover the product rule when the "product" is $f(x)=\langle x|p\rangle$? (spoiler: no, you cannot, because it is not a product of functions of $x$). $\endgroup$ Sep 9, 2023 at 17:54
  • $\begingroup$ I believe you are technically out by a factor of $1/\sqrt{2\pi \hbar}$ in converting $\langle x\vert p\rangle$ to the exponential. $\endgroup$ Sep 9, 2023 at 19:05
  • $\begingroup$ @ZeroTheHero I always forget which factors go where so you could very well be correct. I think in this case if you choose $\langle p|p^\prime\rangle=2\pi\hbar\,\delta(p-p^\prime)$ then it comes out as I wrote, while yours corresponds to the choice $\langle p|p^\prime\rangle=\delta(p-p^\prime)$. $\endgroup$ Sep 11, 2023 at 2:42
  • $\begingroup$ The natural choice is $\langle x\vert p\rangle=e^{ipx/\hbar}/\sqrt{2\pi \hbar}$ so that $\langle p\vert x\rangle=\langle x\vert p\rangle^*$ so you never have to worry about $2\pi$ factors as you do. $\endgroup$ Sep 11, 2023 at 13:05

Your notations are confusing and not at all consistent with the conventional Dirac notation :

  • an operator $\hat {\mathcal O}$ should be applied to a ket $|\psi\rangle$, forming an expression like $\hat{\mathcal O}|\psi\rangle$. For example, the fact that $|p\rangle$ is an eigenvector of the operator $\hat p$ is written : \begin{equation} \hat p |p\rangle = p|p\rangle\tag{1}\end{equation}
  • $|x\rangle$ and $|p\rangle$ are ket-valued functions of $x$ and $p$ respectively. In other words for each value of $x$ there is a ket $|x\rangle$ representing a particle at the position $x$. Consequently, by taking the hermitian product $\langle x|p\rangle$, we obtain a complex-valued function of $x$ and $p$, and specifically : $$\langle x|p\rangle = C\exp\left(\frac{ipx}{\hbar}\right) \tag 2$$
  • the operator $\hat p$ is $-i\hbar \partial_x$ in the position representation. This can be written in several different ways, one of which is that for any ket $|\psi\rangle$, we have : $$\langle x |\hat p|\psi\rangle = -i\hbar \partial_x (\langle x|\psi\rangle) \tag 3$$ Everything makes sense in the RHS, as $\langle x|\psi\rangle$ is a complex-valued function which we can differentiate easily.

Now, it might seem that equations $(1),(2)$ and $(3)$ allow us to compute $\langle x |\hat p|p\rangle$ in two different ways, so we might wan to check that everything is consistent. On one hand we have : $$\langle x |\hat p|p\rangle = \langle x |\big( p|p\rangle\big) = p\langle x |p\rangle$$ while on the other hand we have : \begin{align} \langle x |\hat p|p\rangle &= -i\hbar \partial_x ( \langle x |p\rangle) \\ &= -i\hbar \Big[ \partial_x \big(\langle x|\big) |p\rangle + \langle x| \partial_x \big(|p\rangle\big)\Big] \end{align} Here, we have applied the product rule to the two ket-valued functions $|x\rangle $ and $|p\rangle$. However, $|p\rangle$ does not depend on $x$, so its derivative with respect to $x$ vanishes and we are left with : $$ \langle x |\hat p|p\rangle = -i\hbar \partial_x \big(\langle x|\big) |p\rangle \tag 4$$

This equation holds for any value of $p$. As the kets $|p\rangle$ form a basis of the Hilbert space, equation $(4)$ is equivalent to : $$\langle x |\hat p = -i\hbar \partial_x \big(\langle x|\big) $$ or, taking the hermitian conjugate, to : $$\hat p |x\rangle = i\hbar \partial_x |x\rangle \tag{4'}$$ We see that what we are left is just an equivalent way of writing down $\hat p $ in the position representation.


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.