TOPICS
Search

Generalized Fourier Series


A generalized Fourier series is a series expansion in a complete orthogonal system, with convergence in the relevant L2-norm. The prototypical example is the Fourier series based on the constant function 1 and the functions cos(nx) and sin(nx), n>=1, on [-pi,pi]. Another example is the Laplace series, a double series expansion in the spherical harmonics Y_l^m(theta,phi). Their inner product uses complex conjugation and the area element sinthetadthetadphi over theta in [0,pi] and phi in [0,2pi] (NIST, §14.30(ii)).

For a complete orthogonal system of nonzero univariate functions {phi_n}_(n=0)^infty on an interval R, take a nonnegative measurable function w as weight. Work in the L2-space H=L^2(R,w(x)dx), identifying functions equal almost everywhere with respect to w(x)dx. The orthogonality relationship is

 int_Rphi_m(x)phi_n(x)^_w(x)dx=c_mdelta_(mn),
(1)

where c_m=||phi_m||^2>0 and delta_(mn) is the Kronecker delta. For f in H, the expansion

 f=sum_(n=0)^inftya_nphi_n
(2)

means that the partial sums S_N=sum_(m=0)^(N)a_mphi_m converge to f in norm (Axler 2026, pp. 242-244). For fixed n and N>=n, finite summation gives

 int_RS_N(x)phi_n(x)^_w(x)dx
=int_Rsum_(m=0)^Na_mphi_m(x)phi_n(x)^_w(x)dx
=sum_(m=0)^Na_mint_Rphi_m(x)phi_n(x)^_w(x)dx
=sum_(m=0)^Na_mc_mdelta_(mn)
=a_nc_n.
(3)

Passing to the limit is justified by Schwarz's inequality, since |<S_N-f,phi_n>|<=||S_N-f||||phi_n||->0. Thus the expansion coefficients are

 a_n=1/(c_n)int_Rf(x)phi_n(x)^_w(x)dx.
(4)

This argument uses finite sums before a norm limit, and does not require interchanging an infinite series and an integral.

A useful special form occurs when the constant function 1 and two families {f_1(n,x)}_(n=1)^infty and {f_2(n,x)}_(n=1)^infty together form a complete orthogonal system in H. Assume 0<int_Rw(x)dx<infty, and write the orthogonality relationships as

int_Rf_1(m,x)f_1(n,x)^_w(x)dx=c_mdelta_(mn)
(5)
int_Rf_2(m,x)f_2(n,x)^_w(x)dx=d_mdelta_(mn)
(6)
int_Rf_1(m,x)f_2(n,x)^_w(x)dx=0
(7)
int_Rf_1(m,x)w(x)dx=0
(8)
int_Rf_2(m,x)w(x)dx=0
(9)

for m,n>=1, where c_m,d_m>0. Define the finite partial sum

 S_N(x)=e+sum_(n=1)^Na_nf_1(n,x)+sum_(n=1)^Nb_nf_2(n,x).
(10)

For N>=n>=1, the corresponding calculations are

 int_RS_N(x)f_1(n,x)^_w(x)dx=eint_Rf_1(n,x)^_w(x)dx+int_Rsum_(m=1)^Na_mf_1(m,x)f_1(n,x)^_w(x)dx+int_Rsum_(m=1)^Nb_mf_2(m,x)f_1(n,x)^_w(x)dx
=e·0+sum_(m=1)^Na_mint_Rf_1(m,x)f_1(n,x)^_w(x)dx+sum_(m=1)^Nb_mint_Rf_2(m,x)f_1(n,x)^_w(x)dx
=sum_(m=1)^Na_mc_mdelta_(mn)+sum_(m=1)^Nb_m·0
=a_nc_n
int_RS_N(x)f_2(n,x)^_w(x)dx=eint_Rf_2(n,x)^_w(x)dx+int_Rsum_(m=1)^Na_mf_1(m,x)f_2(n,x)^_w(x)dx+int_Rsum_(m=1)^Nb_mf_2(m,x)f_2(n,x)^_w(x)dx
=e·0+sum_(m=1)^Na_mint_Rf_1(m,x)f_2(n,x)^_w(x)dx+sum_(m=1)^Nb_mint_Rf_2(m,x)f_2(n,x)^_w(x)dx
=sum_(m=1)^Na_m·0+sum_(m=1)^Nb_md_mdelta_(mn)
=b_nd_n
int_RS_N(x)w(x)dx=eint_Rw(x)dx+int_Rsum_(m=1)^Na_mf_1(m,x)w(x)dx+int_Rsum_(m=1)^Nb_mf_2(m,x)w(x)dx
=eint_Rw(x)dx+sum_(m=1)^Na_mint_Rf_1(m,x)w(x)dx+sum_(m=1)^Nb_mint_Rf_2(m,x)w(x)dx
=eint_Rw(x)dx+sum_(m=1)^Na_m·0+sum_(m=1)^Nb_m·0
=eint_Rw(x)dx.
(11)

Taking limits in the inner product, as above, gives

a_n=1/(c_n)int_Rf(x)f_1(n,x)^_w(x)dx
(12)
b_n=1/(d_n)int_Rf(x)f_2(n,x)^_w(x)dx
(13)
e=(int_Rf(x)w(x)dx)/(int_Rw(x)dx).
(14)

The usual Fourier series is recovered by taking f_1(n,x)=cos(nx) and f_2(n,x)=sin(nx) for n>=1, together with the constant function 1, over [-pi,pi] with weight w(x)=1. For every positive integer m,

c_m=int_(-pi)^picos^2(mx)dx=pi
(15)
d_m=int_(-pi)^pisin^2(mx)dx=pi.
(16)

Therefore, the Fourier series of a function f(x) is given by

 f(x)=e+sum_(n=1)^inftya_ncos(nx)+sum_(n=1)^inftyb_nsin(nx),
(17)

where the coefficients are

a_n=1/piint_(-pi)^pif(x)cos(nx)dx
(18)
b_n=1/piint_(-pi)^pif(x)sin(nx)dx
(19)
e=1/(2pi)int_(-pi)^pif(x)dx.
(20)

A complete biorthogonal system generalizes the coefficient pairing beyond orthogonal functions. Its completeness alone does not assert that the associated series converges in norm to every vector; a Schauder basis requires that additional expansion property (Heil 1998, pp. 50-51).


See also

Bessel Function Neumann Series, Bessel's Inequality, Expansion Coefficient, Fourier-Bessel Series, Fourier-Legendre Series, Fourier Series, Generalized Fourier Integral, Kapteyn Series, Laplace Series, Orthogonal Basis, Orthogonal Polynomials, Orthonormal Basis, Parseval's Theorem

Explore with Wolfram|Alpha

References

Axler, S. Measure, Integration & Real Analysis. Author-hosted revision, June 12, 2026. https://measure.axler.net/MIRA.pdf.Heil, C. A Basis Theory Primer. Electronic manuscript, corrected Jan. 1998. https://heil.math.gatech.edu/papers/bases.pdf.National Institute of Standards and Technology. "Fourier Series" and "Spherical and Spheroidal Harmonics." §§1.8 and 14.30 in NIST Digital Library of Mathematical Functions. https://dlmf.nist.gov/1.8. https://dlmf.nist.gov/14.30.

Referenced on Wolfram|Alpha

Generalized Fourier Series

Cite this as:

Weisstein, Eric W. "Generalized Fourier Series." From MathWorld--A Wolfram Resource. https://mathworld.wolfram.com/GeneralizedFourierSeries.html

Subject classifications