Consider a particle on sphere. Its Hamiltonian in spherical polar coordinates is given by -
$-\frac{\hbar^2}{2mr^2}\Big(\frac{1}{\sin\theta}\frac{\partial}{\partial\theta}\sin\theta\frac{\partial}{\partial\theta}+\frac{1}{\sin^2\theta}\frac{\partial^2}{\partial\phi^2}\Big)=\frac{\hat L^2}{2mr^2}$

$\hat L_z=-i\hbar\frac{\partial}{\partial\phi}$
Commutator of both the operators is 0. So they have simultaneous eigen states.
$\hat L_z\psi_{l,m}=\hbar m\psi_{l,m}\;m\in\mathbb R\tag{1}$
$\hat L^2\psi_{l,m}=\hbar l(l+1)\psi_{l,m}\;l\in\mathbb R\tag{2}$

Solution to $(1)$ is given as $\psi_{lm}(\theta,\phi)=e^{im\phi}P_{l,m}(\theta)$

We require $\psi_{l,m}(\theta,\phi+2\pi)=\psi_{l,m}(\theta,\phi)$ which implies $m\in\mathbb Z$

But now if we consider $\psi_{l,m}(\theta,\phi+2\pi)=-\psi_{l,m}(\theta,\phi)$ ,then this implies $m=\frac{2n+1}{2}$ where $n\in\mathbb Z \tag{2}$

I feel that there is no harm in the above assumption because the probability density $|e^{im\phi}|^2$ remains same which is 1 (while considering only $\phi$ part) which is important.

But the solution of $\theta$ part is associated Legendre polynomials $\alpha\;\bigg(\frac{d}{dx}\bigg)^{|m|}P_l(x)\tag{3}$
But how can we solve the above equation if we consider the assumption in $(2)$ which yields fraction values of $m$?

In short, my question is that
i) if we consider particle on a ring (constant $\theta$), the $\psi_m=e^{im\phi}$. Can we use the assumption in $(2)$? Is there some violation in doing so?

ii) But if we consider particle on a sphere (varying $\theta$), then how can we solve the derivative in $(3)$ for fraction values of $m$?

In $(3)$, can the derivative $\bigg(\Big(\frac{d}{dx}\Big)^{|m|}\bigg)$ be defined for fractional values of $m$? For integer values of $m$, the derivative will be act on the function $m$ times.
If this is defined for fractional values of $m$ also then I think, assumption in $(2)$ can be valid?

  • $\begingroup$ You need fractional derivatives. Feynman did some work on that. Cannot remember where I saw it. $\endgroup$ Sep 6, 2022 at 2:57
  • $\begingroup$ For fractional associated Legendre polynomials, this answer may be helpful. $\endgroup$
    – Luessiaw
    Sep 6, 2022 at 3:15
  • $\begingroup$ Mathematically you can define these objects, but physically they aren't relevant because the wavefunction needs to be single valued. $\endgroup$
    – Andrew
    Sep 6, 2022 at 3:24
  • $\begingroup$ @Andrew, I think probability density $|\psi|^2$ should be single valued. Because probability density os more physically relevant than the wavefunction? $\endgroup$
    – Manu
    Sep 6, 2022 at 3:48

1 Answer 1


$l$ must be an integer, otherwise $P_{l}(x)$ is unnormalizable. So is $m$ since $m=-l,\cdots,l-1,l$.

Mathematically, the logic is that $m$ should be an integer, otherwise eigenfunctions of $L_z$ are multivalued. (We're talking about the orbital angular momentum. In real space a rotaion by $2\pi$ shouldn't change anything, namely, $\psi(\phi+2\pi)=\psi(\phi)$.) With $m$ an integer, the associated Legendre polynomials are normalizable iff $l$ is an integer no less than $|m|$.

When it comes to spin, things are different. A roration by $2\pi$ (in the spin space) leads to a negative sign for fermions.

  • $\begingroup$ Yes, $l$ must be integer so that we get polynomial solution instead of power series which will diverge. The solution will be associated Legendre polynomials. I have one doubt in $(3)$ the polynomial is proportional to $\bigg(\frac{d}{dx}\bigg)^|m| P_l(x)$. Can this derivative be defined for the fractional m? $\endgroup$
    – Manu
    Sep 6, 2022 at 2:45
  • $\begingroup$ Sorry, I was wrong about the logic. See my updates. If $m$ is a fraction, the derivative can be defined, though a bit comlicated. But I'm not sure whether the polynomials are still solutions or not. $\endgroup$
    – Luessiaw
    Sep 6, 2022 at 3:02
  • $\begingroup$ Thanks for the reply. Can you tell why the rotation by $2\pi$ changes the sign in spin space. $\endgroup$
    – Manu
    Sep 6, 2022 at 10:01
  • 2
    $\begingroup$ Roughly speaking, fermions have half-integer spin, for example $\frac12$, which means half rotation will arouse an extral $\pi$ phase shift to the spin. I'm not confident to explain it simply. You can refer to abundant literature such as Littlejohn's lectures for details. $\endgroup$
    – Luessiaw
    Sep 6, 2022 at 12:40

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.