I was following Yakov Shnir his book about magnetic monopoles, there they derive the monopole spherical harmonics. I will sketch the derivation briefly,

The starting equation is the eigenvalue problem of the total angular momentum,

\begin{align} \hat{\textbf{J}}^2 Y_{\mu lm}(\theta, \phi) &= \hbar^2 \Big\{\frac{-1}{\sin^2 \theta} \Big[ \sin \theta \frac{\partial}{\partial \theta} \big( \sin \theta \frac{\partial}{\partial \theta} \big) + \big( \frac{\partial}{\partial \varphi} - i\mu (1-\cos \theta )\big)^2 \Big]+ \mu^2\Big\} Y_{\mu lm}(\theta, \phi)\\&= \hbar^2\lambda Y_{\mu lm}(\theta, \phi). \end{align}

The $\theta$-parameter can be replaced by the relation $x=\cos\theta$ and the $\phi$ dependence can be eliminated by $Y_{\mu lm}(\theta, \phi)= \exp(i(\mu+m)\phi) P(\theta)$,

\begin{equation} \label{eq: diff eq angular part1} \Big\{ -(1-x^2)\frac{d}{dx^2} + 2x \frac{d}{dx} + \frac{(m+\mu x)^2}{1-x^2} +\mu^2 \Big\} P(x) = \lambda P(x). \end{equation}

To solve the problem we want to rewrite it as the Euler's hypergeometric function. To obtain that form $P(x)$ has to be rewritten as,

\begin{equation} P(x) = (1-x)^{-\frac{\mu +m}{2}} (1+x)^{-\frac{\mu -m}{2}} F(x), \end{equation}

equation (\ref{eq: diff eq angular part1}) will then become,

\begin{equation} (1-x^2) \frac{d^2F}{dx^2} + 2 ( m+(\mu-1)x) \frac{dF}{dx} + (\mu - \mu^2 + \lambda)F = 0. \end{equation} If now the transformation $z= (1+x)/2$ is applied, then we have rewritten the differential equation as the Euler's hypergeometric differential equation,

\begin{align} z(1-z) F'' &+ \big\{ m-\mu + 1 + 2z(\mu-1) \big\} F' + (\mu -\mu^2 + \lambda) F \\ & = z(1-z) F'' + \big\{ c-(a+b+1)z \big\}F' -abF = 0, \end{align}

where $c=m-\mu+1$,$~ab = \mu^2 -\mu -\lambda$ and $a+b+1 = 2(1- \mu)$. The solution of this differential equations is the hypergeometric function $_2F_1(a;b;c;z) $ which is defined as, \begin{equation} _2F_1(a;b;c;z) = \sum_{n=0}^{\infty} \frac{(a)_n(b)_n}{(c)_n} \frac{z^n}{n!} \equiv \sum_{n=0}^{\infty} \alpha_n z^n. \end{equation}

Now my question is the following, they consider that $a$ is negative integer such that the sum terminates, my question now is why? Because if the sum not terminates then will the convergence radius be 1, but since $\theta$ is between $0$ and $\pi$ and by definiton of $z$ you still have a bijection from $z$ to the range of $\theta$.

This indeed doesn't say anything about being square integrable but how could one see this from the hypergeometric function/series? In fact I am searching for an analytical argument why the quantum numbers only differ in integers?

Thanks in advance.

  • $\begingroup$ Examining an analytical argument why quantum numbers only differ in integers should probably be posted separately. $\endgroup$ Commented Apr 30, 2017 at 12:13
  • $\begingroup$ First of all you have $z=\cos(\theta)$, so the radius of convergence being one is ok. Second, there exists indeed simple argument why the spherical harmonic is a fixed order polynomial in $\cos(\theta)$, in fact the order is equal to $l$ if i'm not mistaken. You need to work out a couple more formulae, fistly that $\lambda = l(l+1)$, and then it seems to me things will only work out if $\mu$ is an integer which can also be justified by single-valuedness under $\phi -> \phi+2\pi$. $\endgroup$
    – Kphysics
    Commented Feb 17, 2019 at 1:55

1 Answer 1


This is not a direct answer to the question asked, but rather a different approach to the problem. It's much easier to derive the monopole harmonics from a geometric consideration of the Hopf map ${\rm Hopf}: S^3\to S^2$, which is the same thing as the Hopf bundle $\pi: {\rm SU}(2) \to {\rm SU}(2)/{\rm U}(1)$. Then one immediately sees monopole spherical harmonics coincide with the ${\rm SU}(2)$ representation matrices. Recall that the ordinary spherical harmonics are given in terms of these matrices as $$ Y^L_m(\theta, \phi)= \sqrt{\frac{2L+1}{4\pi}}[D^L_{m,0}(\theta, \phi,\psi)]^* $$ where, on the LHS, $\theta$ $\phi$ are the spherical polar angles on $S^2$. On the RHS $\theta$, $\phi$, $\psi$ are the Euler-angle co-ordinates on $S^3$. The Hopf map simply takes $(\theta, \phi, \psi)$ to $(\theta, \phi)$. Because $$ D^L_{m,n}(\theta, \phi,\psi)= <L,m|\exp(-i\phi J_z)\exp(-i\theta J)\exp(-i \psi J_z)|L,n> $$ the RHS does not depend on $\psi$ when $n=0$.

For a monople of strength $\int B d(Area)= 4\pi \Lambda $, where $4\pi \Lambda $ must be an integer multiple of $2\pi$, the monopole harmonics are
$$ {\mathcal Y}^J_{m, \Lambda} (\theta, \phi,\psi)= \sqrt{\frac{2J+1}{4\pi}}[D^J_{m\Lambda }(\theta,\phi,\psi)]^* $$ Now both sides depend on $\psi$, so one must choose a $\psi$ for each pair $(\theta, \phi)$. This is just a choice of gauge. For a derivation and explanantion of all this see page 278 in my online lecture notes https://courses.physics.illinois.edu/phys509/sp2017/bmaster.pdf.

  • $\begingroup$ The OP asked why is the a parameter integer, no? $\endgroup$
    – Kphysics
    Commented Feb 17, 2019 at 1:39
  • $\begingroup$ Hi Mike. Is there an updated home for these notes? $\endgroup$ Commented Nov 2, 2022 at 16:25
  • 1
    $\begingroup$ @Ryan Paul is keeing them alive at goldbart.gatech.edu/PostScript/MS_PG_book/bookmaster.pdf. (although he is now at Stonybrook) $\endgroup$
    – mike stone
    Commented Nov 2, 2022 at 16:59
  • $\begingroup$ Awesome, thank you! $\endgroup$ Commented Nov 4, 2022 at 3:53

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.