Zonal spherical harmonics
In the mathematical study of rotational symmetry, the zonal spherical harmonics are special spherical harmonics that are invariant under the rotation through a particular fixed axis. The zonal spherical functions are a broad extension of the notion of zonal spherical harmonics to allow for a more general symmetry group.
On the two-dimensional sphere, the unique zonal spherical harmonic of degree ℓ invariant under rotations fixing the north pole is represented in spherical coordinates by
where Pℓ is a Legendre polynomial of degree ℓ. The general zonal spherical harmonic of degree ℓ is denoted by , where x is a point on the sphere representing the fixed axis, and y is the variable of the function. This can be obtained by rotation of the basic zonal harmonic
In n-dimensional Euclidean space, zonal spherical harmonics are defined as follows. Let x be a point on the (n−1)-sphere. Define to be the dual representation of the linear functional
for all Y ∈ Hℓ. The integral is taken with respect to the invariant probability measure.
Relationship with harmonic potentials
The zonal harmonics appear naturally as coefficients of the Poisson kernel for the unit ball in Rn: for x and y unit vectors,
where is the surface area of the (n-1)-dimensional sphere. They are also related to the Newton kernel via
where x,y ∈ Rn and the constants cn,k are given by
The coefficients of the Taylor series of the Newton kernel (with suitable normalization) are precisely the ultraspherical polynomials. Thus, the zonal spherical harmonics can be expressed as follows. If α = (n−2)/2, then
where cn,ℓ are the constants above and is the ultraspherical polynomial of degree ℓ.
- The zonal spherical harmonics are rotationally invariant, meaning that
- for every orthogonal transformation R. Conversely, any function ƒ(x,y) on Sn−1×Sn−1 that is a spherical harmonic in y for each fixed x, and that satisfies this invariance property, is a constant multiple of the degree ℓ zonal harmonic.
- If Y1,...,Yd is an orthonormal basis of Hℓ, then
- Evaluating at x = y gives