Completeness of the Fourier sine system on an interval #
On an interval [a, b] with a < b, the functions x ↦ sin (k π (x - a) / (b - a)),
k = 1, 2, …, are the eigenfunctions of -d²/dx² with zero boundary values at both ends. This
file proves that they form a complete system in L²(a, b): a square integrable function
orthogonal to all of them vanishes almost everywhere
(TauCeti.ae_eq_zero_of_forall_integral_mul_sin_eq_zero). The theorem does not assume a < b:
when b ≤ a the interval (a, b] is empty, so the restricted measure is zero and the conclusion
holds trivially.
For a < b, the proof reduces to Mathlib's completeness of the exponentials on a circle. Extend g
oddly about a to [2a - b, b]. On this interval of length 2(b - a), the n-th Fourier
coefficient of the odd extension is a nonzero multiple of ∫ g(x) sin (n π (x - a) / (b - a)) dx,
so all of them vanish, and Parseval's identity (hasSum_sq_fourierCoeffOn) forces the extension,
hence g, to vanish almost everywhere. The first step is recorded separately: a square integrable
function on (a, b] whose Fourier coefficients on [a, b] all vanish is zero almost everywhere
(TauCeti.ae_eq_zero_of_forall_fourierCoeffOn_eq_zero).
Main results #
TauCeti.ae_eq_zero_of_forall_fourierCoeffOn_eq_zero: uniqueness of Fourier coefficients of square integrable functions on an interval.TauCeti.ae_eq_zero_of_forall_integral_mul_sin_eq_zero: completeness of the sine systemsin (k π (x - a) / (b - a)),k ≥ 1, inL²(a, b).
References #
- G. B. Folland, Fourier Analysis and its Applications, Section 2.4 (odd extensions and Fourier sine series) and Section 3.4 (completeness of the trigonometric system).
A square integrable function with vanishing Fourier coefficients is zero. If f is square
integrable on (a, b] and every Fourier coefficient of f on [a, b] vanishes, then f = 0
almost everywhere on (a, b]. This is Parseval's identity hasSum_sq_fourierCoeffOn.
Completeness of the sine system #
Completeness of the Fourier sine system. A function g square integrable on (a, b] with
∫ x in a..b, g x * sin (k π (x - a) / (b - a)) = 0 for every k ≥ 1
vanishes almost everywhere on (a, b]. For a < b these sines are the Dirichlet eigenfunctions
of -d²/dx² on (a, b), so this is the completeness half of the Fourier sine expansion. For
b ≤ a the interval (a, b] is empty, the restricted measure is zero, and the conclusion holds
trivially.