The spherical harmonics of degree $k$ in $n$ dimensions are the restriction to the sphere $\mathbb S^{n-1}$ of harmonic polynomials homogeneous of degree $k$ in $n$ variables. It is a classical fact of analysis that an Hilbertian basis of $L^2(\mathbb S^{n-1})$ can be made with spherical harmonics and that the Laplace operator on $L^2(\mathbb S^{n-1})$ can be written as $$ -\Delta_{\mathbb S^{n-1}}=\sum_{k\ge 0}k(k+n-2)\mathbb P_k, \qquad \text{Id}_{L^2(\mathbb S^{n-1})}=\sum_{k\ge 0}\mathbb P_k, $$ where $\mathbb P_k$ is the orthogonal projection onto the space of spherical harmonics of degree $k$.

My Question : The proofs that I know of the second identity above are not so elementary and I would like to know if there is a simple proof of the fact that the orthogonal (in $L^2(\mathbb S^{n-1})$) of the space of all spherical harmonics is reduced to $\{0\}$.

  • 5
    $\begingroup$ To fix a threshold for "elementarity", can you mention an elementary proof that the trigonometric system is complete on $L^2(\mathbb S^1)$? (By this I mean that its $L^2$ orthogonal is ${0}$) $\endgroup$ – Giuseppe Negro Oct 11 '17 at 13:11

In other words, you want to prove that spherical harmonics are dense in $L^2(S)$. By a standard argument it is enough to show that they are dense in $C(S)$ (continuous functions). Or even dense in the space of smooth functions. For every continuous function on the sphere, we can solve Dirichlet problem (this is elementary: there is an explicit formula due to Poisson). So we obtain a harmonic function in the ball. Expand it to a series of homogeneous harmonic polynomials (again this is an explicit series: just expand the kernel of the Poisson formula and integrate term-by-term). Taking a partial sum of this series we obtain a harmonic polynomial approximating our $L^2(S)$ function.

  • 1
    $\begingroup$ The space $P$ of all polynomials on ${\mathbb R} ^n$ can be written as a direct sum $$P=\sum _{m=0}^{\infty} Q^m H$$ where $H$ is the space of harmonic polynomials and $Q=\sum x_i^2$ is the standard quadratic form on $\mathbb R ^n$. It follows, upon restriction to the unit sphere, that all polynomials on $S^{n-1}$ are sums of harmonic polynomials. Then by Stone Weierstrass, Harmonic polynomials are dense in the space of continuous functions on the sphere. $\endgroup$ – Venkataramana Oct 11 '17 at 14:34
  • $\begingroup$ Thanks for your comment: How do you get that identity? $\endgroup$ – Bazin Oct 13 '17 at 11:59
  • $\begingroup$ @Venkataramana Thanks for your comment: How do you get that identity? $\endgroup$ – Bazin Oct 29 '17 at 15:55
  • $\begingroup$ @Bazin, this identity, while simple and algebraic, is a little involved to explain. BY Zariski density, we may deal with $V=\mathbb C ^n$ instead of ${\mathbb R }^n$. Consider the quadric $Z$ defined by $Q=0$ in the projective space ${\mathbb P}({\mathbb C}^n)$. The $m$-th symmetric power $sym ^m (V^*)$ is irreducible for $GL_n(\mathbb C)$, and restricts non-trivially to the quadric $Z$; the quadric $Z=G/P$ where $G=SO(n)$ and $P$ a suitable parabolic. Then (by the Borel-Weil theorem), the restriction of $sym ^m (V^*)$ to $Z$ is irreducible. This is the space of Harmonic degree $m$ polys . $\endgroup$ – Venkataramana Oct 30 '17 at 14:38
  • $\begingroup$ For a reference to this, you may look at Goodman and Wallach book. $\endgroup$ – Venkataramana Oct 30 '17 at 14:40

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.