A paper on supersymmetry in 3-dimensions uses results on the spectra of elliptic operators on $S^3$:

The eigenvalues of the vector Laplacian on divergenceless vector fields is $(\ell + 1)^2$ with degeneracy $2\ell(\ell+2)$ with $\ell \in \mathbb{ Z}$.

They came up with a similar statement for spinor fields on the 3-sphere

The eigenvalues of the Dirac operator $D$ is $\pm (\ell + \frac{1}{2})$ with degeneracy $\ell(\ell+1)$ with $\ell \in \mathbb{ Z}$.

On Math.StackExange it was suggested we can restrict the eigenspace decomposition the Laplacian on $SO(4)$ to $S^3$: $$ L^2(SO(4)) = \bigoplus_{\pi \in \mathrm{Irr}[SO(4)]} \pi \otimes \overline{\pi}$$ In our case, how Peter-Weyl Theorem play out when $SO(4)$ acts on bundles of divergenceless vector fields and spinor-fields $S^3$?


Let $M=G/K$ with $G$ compact and let $(\tau,W_\tau)$ be an irreducible representation of $K$, let $E_\tau$ be the associated $G$-homogeneous vector bundle of $M$. Then, the space of $L^2$-sections of $E_\tau$ decomposes as $$ L^2(E_\tau) = \sum_{\pi\in\widehat G} V_\pi\otimes \operatorname{Hom}_K(V_\pi,W_\tau). $$

The proof follows by Peter-Weyl: $\Gamma^\infty(E_\tau)\simeq C^\infty(G;\tau):=\{f:G\to W_\tau:f(gk)=\tau(k)^{-1}f(g)\}$. For each $\pi\in\widehat G$, we have $V_\pi\otimes \operatorname{Hom}_K(V_\pi,W_\tau) \to C^\infty(G;\tau)$ given by $v\otimes L \mapsto (x\mapsto L(\pi(x^{-1}).v))$. This map is $G$-equivariant.

An other way, $C^\infty(G;\tau) = C^\infty(G,W_\tau)^K \simeq (C^\infty(G)\otimes W_\tau)^K$. Peter-Weyl theorem implies that $$ L^2(E_\tau) = \sum_{\pi\in\widehat G} (V_\pi\otimes V_\pi^*\otimes W_\tau)^K = \sum_{\pi\in\widehat G} V_\pi\otimes (V_\pi^*\otimes W_\tau)^K = \sum_{\pi\in\widehat G} V_\pi\otimes\operatorname{Hom}_K(V_\pi,W_\tau). $$

When $M$ is symmetric (e.g. $M=S^3$) and $G$ semisimple, the Laplace operator acts as the Casimir element of $\mathfrak g$ (the Lie algebra of $G$). Furthermore, the square of the Dirac operator $D^2$ coincides with the Casimir element plus $1/8$ the scalar curvature. See Section 3.5 of T. Friedrich, "Dirac operators in Riemannian geometry".

For example, let $S^{3}=\operatorname{Spin}(4)/\operatorname{Spin}(3)$, let $\tau$ be the spin representation, then $E_\tau$ is the spinor bundle. $\tau$ has highest weight $\frac12 \varepsilon_1$, thus $\operatorname{Hom}_K(V_\pi,W_\tau)\neq0$ if and only if $\pi$ has highest weight $\Lambda = \frac12((2k+1)\varepsilon_1\pm\varepsilon_2)$. The casimir element acts on $\pi_\Lambda$ as $\|\Lambda+\varepsilon_1\|-\|\varepsilon_1\|$ (see Wallach's book, Harmonic analysis on homogeneous vector bundles). Check that is equal to $$ \frac{(2k+1)(2k+5)+1}{4} $$ The sectional curvature is 6, then $D^2$ acts by $(k+3/2)^2$. Finally, the eigenvalues of $D$ are $\pm(k+3/2)$.

  • $\begingroup$ Still working through it. It looks like you are proving the $L^2$ decomposition using smooth sections of $\Gamma^\infty$ and $C^\infty$. This discussion has almost a functorial nature - at least you can get the spectra for $\nabla^2$ and $D$ in a similar way. $\endgroup$ – john mangual Aug 14 '13 at 12:34

The spectrum of the Dirac operator on spheres was computed by Christian Bär:


See also the following preprint by Fabian Meier for a different approach:


  • $\begingroup$ Sir, do you know any paper about the eigenvector of Dirac operator on $S^3$? $\endgroup$ – DLIN Dec 30 '19 at 13:02
  • $\begingroup$ Unfortunately not, sorry... $\endgroup$ – Stephan Mescher Dec 30 '19 at 18:33

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.