29 Lamé FunctionsLamé Polynomials

§29.15 Fourier Series and Chebyshev Series

Contents
  1. §29.15(i) Fourier Coefficients
  2. §29.15(ii) Chebyshev Series

§29.15(i) Fourier Coefficients

Polynomial 𝑢𝐸2⁢nm⁡(z,k2)

When ν=2⁢n, m=0,1,…,n, the Fourier series (29.6.1) terminates:

A convenient way of constructing the coefficients, together with the eigenvalues, is as follows. Equations (29.6.4), with p=1,2,…,n, (29.6.3), and A2⁢n+2=0 can be cast as an algebraic eigenvalue problem in the following way. Let

29.15.2 𝐌=[β0α00⋯0γ1β1α1⋱⋮0⋱⋱⋱0⋮⋱γn−1βn−1αn−10⋯0γnβn]

be the tridiagonal matrix with αp, βp, γp as in (29.3.11), (29.3.12). Let the eigenvalues of 𝐌 be Hp with

29.15.3 H0<H1<⋯<Hn,

and also let

29.15.4 [A0,A2,…,A2⁢n]T

be the eigenvector corresponding to Hm and normalized so that

29.15.5 12⁢A02+∑p=1nA2⁢p2=1

and

29.15.6 12⁢A0+∑p=1nA2⁢p>0.

Then

29.15.7 aν2⁢m⁡(k2)=12⁢(Hm+ν⁢(ν+1)⁢k2),

and (29.15.1) applies, with ϕ again defined as in (29.2.5).

Polynomial 𝑠𝐸2⁢n+1m⁡(z,k2)

When ν=2⁢n+1, m=0,1,…,n, the Fourier series (29.6.16) terminates:

In (29.15.2) replace αp, βp, and γp as in (29.3.13), (29.3.14). Also, replace (29.15.4), (29.15.5), (29.15.6) by

29.15.9 [A1,A3,…,A2⁢n+1]T,
29.15.10 ∑p=0nA2⁢p+12=1,
29.15.11 ∑p=0nA2⁢p+1>0.

Then

29.15.12 aν2⁢m+1⁡(k2)=12⁢(Hm+ν⁢(ν+1)⁢k2),

and (29.15.8) applies.

Polynomial 𝑐𝐸2⁢n+1m⁡(z,k2)

When ν=2⁢n+1, m=0,1,…,n, the Fourier series (29.6.31) terminates:

In (29.15.2) replace αp, βp, and γp as in (29.3.15), (29.3.16). Also, replace (29.15.4), (29.15.5), (29.15.6) by

29.15.14 [B1,B3,…,B2⁢n+1]T,
29.15.15 ∑p=0nB2⁢p+12=1,
29.15.16 ∑p=0n(2⁢p+1)⁢B2⁢p+1>0.

Then

29.15.17 bν2⁢m+1⁡(k2)=12⁢(Hm+ν⁢(ν+1)⁢k2),

and (29.15.13) applies.

Polynomial 𝑑𝐸2⁢n+1m⁡(z,k2)

When ν=2⁢n+1, m=0,1,…,n, the Fourier series (29.6.8) terminates:

In (29.15.2) replace αp, βp, and γp as in (29.6.11). Also, replace (29.15.4), (29.15.5), (29.15.6) by

29.15.19 [C0,C2,…,C2⁢n]T,
29.15.20 (1−12⁢k2)⁢(12⁢C02+∑p=1nC2⁢p2)−12⁢k2⁢∑p=0n−1C2⁢p⁢C2⁢p+2=1,
29.15.21 12⁢C0+∑p=1nC2⁢p>0.

Then

29.15.22 aν2⁢m⁡(k2)=12⁢(Hm+ν⁢(ν+1)⁢k2),

and (29.15.18) applies.

Polynomial 𝑠𝑐𝐸2⁢n+2m⁡(z,k2)

When ν=2⁢n+2, m=0,1,…,n, the Fourier series (29.6.46) terminates:

In (29.15.2) replace αp, βp, and γp as in (29.3.17). Also replace (29.15.4), (29.15.5), (29.15.6) by

29.15.24 [B2,B4,…,B2⁢n+2]T,
29.15.25 ∑p=0nB2⁢p+22=1,
29.15.26 ∑p=0n(2⁢p+2)⁢B2⁢p+2>0.

Then

29.15.27 bν2⁢m+2⁡(k2)=12⁢(Hm+ν⁢(ν+1)⁢k2),

and (29.15.23) applies.

Polynomial 𝑠𝑑𝐸2⁢n+2m⁡(z,k2)

When ν=2⁢n+2, m=0,1,…,n, the Fourier series (29.6.23) terminates:

In (29.15.2) replace αp, βp, and γp as in (29.6.26). Also replace (29.15.4), (29.15.5), (29.15.6) by

29.15.29 [C1,C3,…,C2⁢n+1]T,
29.15.30 (1−12⁢k2)⁢∑p=0nC2⁢p+12−12⁢k2⁢(12⁢C12+∑p=0n−1C2⁢p+1⁢C2⁢p+3)=1,
29.15.31 ∑p=0nC2⁢p+1>0.

Then

29.15.32 aν2⁢m+1⁡(k2)=12⁢(Hm+ν⁢(ν+1)⁢k2),

and (29.15.28) applies.

Polynomial 𝑐𝑑𝐸2⁢n+2m⁡(z,k2)

When ν=2⁢n+2, m=0,1,…,n, the Fourier series (29.6.38) terminates:

In (29.15.2) replace αp, βp, and γp as in (29.6.41). Also replace (29.15.4), (29.15.5), (29.15.6) by

29.15.34 [D1,D3,…,D2⁢n+1]T,
29.15.35 (1−12⁢k2)⁢∑p=0nD2⁢p+12+12⁢k2⁢(12⁢D12−∑p=0n−1D2⁢p+1⁢D2⁢p+3)=1,
29.15.36 ∑p=0n(2⁢p+1)⁢D2⁢p+1>0.

Then

29.15.37 bν2⁢m+1⁡(k2)=12⁢(Hm+ν⁢(ν+1)⁢k2),

and (29.15.33) applies.

Polynomial 𝑠𝑐𝑑𝐸2⁢n+3m⁡(z,k2)

When ν=2⁢n+3, m=0,1,…,n, the Fourier series (29.6.53) terminates:

In (29.15.2) replace αp, βp, and γp as in (29.6.56). Also replace (29.15.4), (29.15.5), (29.15.6) by

29.15.39 [D2,D4,…,D2⁢n+2]T,
29.15.40 (1−12⁢k2)⁢∑p=0nD2⁢p+22−12⁢k2⁢∑p=1nD2⁢p⁢D2⁢p+2=1,
29.15.41 ∑p=0n(2⁢p+2)⁢D2⁢p+2>0.

Then

29.15.42 bν2⁢m+2⁡(k2)=12⁢(Hm+ν⁢(ν+1)⁢k2),

and (29.15.38) applies.

§29.15(ii) Chebyshev Series

The Chebyshev polynomial T of the first kind (§18.3) satisfies cos⁡(p⁢ϕ)=Tp⁡(cos⁡ϕ). Since (29.2.5) implies that cos⁡ϕ=sn⁡(z,k), (29.15.1) can be rewritten in the form

This determines the polynomial P of degree n for which 𝑢𝐸2⁢nm⁡(z,k2)=P⁡(sn2⁡(z,k)); compare Table 29.12.1. The set of coefficients of this polynomial (without normalization) can also be found directly as an eigenvector of an (n+1)×(n+1) tridiagonal matrix; see Arscott and Khabaza (1962).

Using also sin⁡((p+1)⁢ϕ)=(sin⁡ϕ)⁢Up⁡(cos⁡ϕ), with U denoting the Chebyshev polynomial of the second kind (§18.3), we obtain

29.15.44 𝑠𝐸2⁢n+1m⁡(z,k2) =∑p=0nA2⁢p+1⁢T2⁢p+1⁡(sn⁡(z,k)),
29.15.45 𝑐𝐸2⁢n+1m⁡(z,k2) =cn⁡(z,k)⁢∑p=0nB2⁢p+1⁢U2⁢p⁡(sn⁡(z,k)),
29.15.46 𝑑𝐸2⁢n+1m⁡(z,k2) =dn⁡(z,k)⁢(12⁢C0+∑p=1nC2⁢p⁢T2⁢p⁡(sn⁡(z,k))),
29.15.47 𝑠𝑐𝐸2⁢n+2m⁡(z,k2) =cn⁡(z,k)⁢∑p=0nB2⁢p+2⁢U2⁢p+1⁡(sn⁡(z,k)),
29.15.48 𝑠𝑑𝐸2⁢n+2m⁡(z,k2) =dn⁡(z,k)⁢∑p=0nC2⁢p+1⁢T2⁢p+1⁡(sn⁡(z,k)),
29.15.49 𝑐𝑑𝐸2⁢n+2m⁡(z,k2) =cn⁡(z,k)⁢dn⁡(z,k)⁢∑p=0nD2⁢p+1⁢U2⁢p⁡(sn⁡(z,k)),
29.15.50 𝑠𝑐𝑑𝐸2⁢n+3m⁡(z,k2) =cn⁡(z,k)⁢dn⁡(z,k)⁢∑p=0nD2⁢p+2⁢U2⁢p+1⁡(sn⁡(z,k)).

For explicit formulas for Lamé polynomials of low degree, see Arscott (1964b, p. 205).