I was dealing with a problem, which said that,

Supposedly Hamiltonian of a conservative system in classical mechanics is $\omega xp$, where $\omega$ is a constant, and $x$ and $p$ are the position and momentum operator respectively. The corresponding Hamiltonian in quantum mechanics is________.

To solve the problem, an initial step was taken

$$\omega xp= \frac{\omega}{2}(xp+px).$$

stating that the quantum operator should be Hermitian in nature.

What I don't understand is, how the above proceeded step was an important part/property of Hermitian operator, i.e., why the above step was taken?


This is an interesting question because in general there is no unique answer, although there is a "natural" answer for the simplest observables, which include $xp$. The lack of uniqueness is linked to the ordering problem.

The simplest quantization is due to Dirac; he tried to implement the idea that the commutator of two operators representing observables should be $i\hbar$ times the operator corresponding to the Poisson bracket of the two respective classical quantities.

This is based on the observations that \begin{align} [\hat x,\hat p]=i\hbar \widehat{\{x,p\}} \end{align} where $\{\cdot,\cdot\}$ is the classical Poisson bracket. Dirac basically proposed to generalize this to \begin{align} [\hat f,\hat g]=i\hbar\widehat{\{f,g\}} \, . \tag{1} \end{align} subject to the additional assumptions that \begin{align} \widehat{f(x)}=f(\hat x)\, ,\qquad \widehat{g(p)}=g(\hat p) \end{align} The problem is that (1) eventually fails although it does work for $xp$. To get the result you want you need to ask what is $\widehat{xp}$. To get there start with \begin{align} \frac{1}{i\hbar}[x^2,p^2]&=2\left(\hat x\hat p+\hat p\hat x\right) =\widehat{\{x^2,p^2\}}= 4 \widehat{xp} \end{align} from which you get \begin{align} \widehat{xp}=\frac{1}{2}\left(\hat x\hat p+\hat p\hat x\right) \end{align} which is the usual symmetrization rule.

Gronewold in

Groenewold, H.J., 1946. On the principles of elementary quantum mechanics, Physica 12 (1946) 405-460,

showed that (1) eventually fails, i.e. there is no rule consistent with (1) that will allow you to find, for instance, the operator corresponding to $x^2p^2$. This is excellently reviewed in

Chernoff, P.R., 1981. Mathematical obstructions to quantization. Hadronic Journal, 4(3), pp.879-898.

The remarkable thing is that of course the rule (1) works for the ``most common'' observables like angular momentum. However, it is possible to devise classical Hamiltonian for which the quantization is not unique, as per the example of this post.

| cite | improve this answer | |
  • $\begingroup$ Thanks! (for the reference of the papers too) $\endgroup$ – kamer_kane Jul 5 at 16:08

The hamiltonian of a system has to be an hermitian operator since it's associated to a measurable quantity, the energy. By symply taking the classical hamiltonian$$H=\omega xp$$ and converting it directly to an operatore $$\hat H = \omega\hat x\hat p$$ you can easily see that this operator is not hermitian since, given $\hat x$ and $\hat p$ hermitian, $$\hat H^\dagger = \omega\hat{p}^\dagger\hat{x}^\dagger = \omega \hat p\hat x \neq H$$ since the two operators are now in reverse order. Then you have to make $\hat H$ hermitian.

Classically speaking $$H = \omega xp = \frac{\omega}{2}(xp+px)$$ won't change the hamiltonian in any way since $x$ and $p$ being purely variables, and not operators, they clearly commute. This form of the classical hamiltonian is now more suited for converting it into a quantum hamiltonian by using operators instead of simple functions $$H = \frac{\omega}{2}(\hat x\hat p + \hat p \hat x)$$ and now you can easily convince yourself that this form of the hamiltonian is clearly hermitian.

Such a procedure to take a classical hamiltonian to a suitable quantum hamiltonian is called Weyl ordering.

| cite | improve this answer | |

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.