Both $\hat{a}^\dagger\hat{a}$ and $\hat{a}\hat{a}^\dagger$ are Hermitian, how do we know which one represents the particle number?

up vote 9 down vote accepted

We require that the number operator have the following property:

$$\hat n |0\rangle = 0.$$

We know that

$$\hat a |0\rangle = 0$$

and we know that

$$\hat a |1\rangle = |0\rangle$$

and we know that

$$\hat a^{\dagger} |0\rangle = |1\rangle. $$

Thus, it follows that

$$\hat a \hat a^{\dagger} \ne \hat n$$


$$\hat a \hat a^{\dagger} |0\rangle = \hat a|1\rangle \ne 0.$$

Now, it remains to be shown that

$$\hat a^{\dagger}\hat a = \hat n. $$

Can you take it from here?

  • Thank you! You clearly solved my problem, but now I have another question: what's the difference between 0 and $|0\rangle$? (should I open a new question?) – LePtC Oct 11 '14 at 4:05
  • 1
    @LePtC, $|0\rangle$ is a state - the state with 0 energy quanta present while 0 is a number. The product of the number 0 and any state is the number 0. Since $|0\rangle$ is an eigenstate of the number operator with eigenvalue 0, it must be the case that $\hat n |0\rangle = 0 |0\rangle = 0$. – Alfred Centauri Oct 11 '14 at 16:56
  • 1
    @LePtC yes, that should be a new question, except that it's already been asked. – David Z Oct 12 '14 at 7:30
  • @AlfredCentauri you mean we can leave the ket when we have a 0 eigenvalue? when we get a zero spin for s=1 particle or zero energy for the Hamiltonian, why don't we leave the ket as well? – LePtC Oct 12 '14 at 16:13
  • @LePtC, I don't understand what you're trying to ask. A ket that is an eigenstate of some operator with eigenvalue zero has non-zero length, i.e., $\langle 0 | 0 \rangle \ne 0$. Only the $\varnothing$ (null) ket has zero length $\langle \varnothing | \varnothing\rangle = 0$ and it is not a state. – Alfred Centauri Oct 12 '14 at 16:35

Since you define e.g. in the bosonic case

$c_j^\dagger: H_N^S \rightarrow H_{N+1}^S,\quad c_j^\dagger | \ldots n_j \ldots \rangle := \sqrt{n_j+1} |\ldots n_j+1 \ldots \rangle$ $c_j: H_N^S \rightarrow H_{N-1}^S,\quad c_j | \ldots n_j \ldots \rangle := \sqrt{n_j} |\ldots n_j-1 \ldots \rangle$

it makes more sence to use $a^\dagger a$ which will give you $n_j$ (instead of $n_j+1$) as prefactor when acting on a state $| \ldots n_j \ldots \rangle $.

  • But you are using the eigen state of $\hat{a}^\dagger\hat{a}$, maybe we can redefine the eigen state $|n'_i\rangle=|n_i+1\rangle$ to avoid that problem? – LePtC Oct 10 '14 at 9:17
  • 2
    @LePtC: This would only "shift" your problem... – pawel_winzig Oct 10 '14 at 10:50

The vacuum should have particle number $0$. In some detail: we would like $\hat{N}\ |0\rangle=0$ and $\hat{N}=a^\dagger a$ is the only ordering that does that. It follows from the usual commutation relations that $\hat{N} |n\rangle =n\ |n\rangle$ which is in sync with interpretation of $|n\rangle $ as an $n$-particle state in second-quantisation.

  • 2
    You want to elaborate on this a bit? – Pranav Hosangadi Oct 10 '14 at 8:05
  • 1
    I think he means that $a^\dagger a$ kills the vacuum while $a \, a^\dagger$ doesn't. – Jold Oct 10 '14 at 23:50
  • @PranavHosangadi Explanation added. I hope that suffices. – suresh Oct 12 '14 at 7:15

Your Answer


By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

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