logo

Complete Orthonormal Basis and Complete Orthonormal Set 📂Lebesgue Spaces

Complete Orthonormal Basis and Complete Orthonormal Set

Theorem: Equivalence Conditions of an Orthonormal Set

Let {ϕn}1\left\{ \phi_{n} \right\}_{1}^\infty be an orthonormal set of L2(a,b)L^2(a,b) and denote fL2(a,b)f \in L^2(a,b). Then, the following conditions are equivalent.

  • (a)(a) For all nn, if f,ϕn=0\left\langle f, \phi_{n} \right\rangle=0 then f=0f=0.

  • (b)(b) For all fL2(a,b)f\in L^2(a,b), the series 1f,ϕnϕn\sum_{1}^\infty \left\langle f,\phi_{n}\right\rangle\phi_{n} converges to ff in the norm sense. That is, the following equation holds:

    f=1f,ϕnϕn f=\sum_{1}^\infty \left\langle f,\phi_{n}\right\rangle\phi_{n}

  • (c)(c) For all fL2(a,b)f \in L^2(a,b), it satisfies the following equation known as Parseval’s equation:

    f2=n=1f,ϕn2 \| f \|^2 = \sum \limits_{n=1}^{\infty} \left| \left\langle f,\phi_{n} \right\rangle \right|^{2}

Explanation

The orthonormal set satisfying (a)(c)(a) - (c) is called an orthonormal basis or a complete orthonormal set.

Observing these three conditions reveals that the orthonormal basis serves a role equivalent to a basis in finite-dimensional vector spaces.

  • When {ϕn}\left\{ \phi_{n} \right\} is an orthonormal basis, the constants f,ϕn\left\langle f, \phi_{n}\right\rangle are called (generalized) Fourier coefficients.

  • The series f,ϕnϕn\sum \left\langle f, \phi_{n}\right\rangle\phi_{n} is referred to as a (generalized) Fourier series.

Lemma

Assume fL2(a,b)f \in L^2(a,b) and that {ϕn}\left\{ \phi_{n} \right\} is an orthonormal set in L2(a,b)L^2(a,b). Then the series f,ϕnϕn\sum \left\langle f,\phi_{n} \right\rangle\phi_{n} converges in the norm sense. And it satisfies the following inequality:

f,ϕnϕnf \left\| \sum \left\langle f,\phi_{n}\right\rangle \phi_{n} \right\| \le | f|

Proof

  • (a)    (b)(a) \implies (b)

    Assume (a)(a). Then, by the lemma, f,ϕnϕn\sum \left\langle f, \phi_{n} \right\rangle\phi_{n} converges in the norm sense. Let’s define the difference of the series as gg.

    g=fn=1f,ϕnϕn g=f-\sum \limits_{n=1}^{\infty} \left\langle f, \phi_{n} \right\rangle\phi_{n}

    Then, it can be shown that g=0g=0.

    g,ϕm= f,ϕmn=1f,ϕnϕn,ϕm= f,ϕmf,ϕm= 0 \begin{align*} \left\langle g,\phi_{m} \right\rangle &=\ \left\langle f,\phi_{m}\right\rangle - \sum \limits_{n=1}^{\infty}\left\langle f,\phi_{n} \right\rangle \left\langle \phi_{n}, \phi_{m} \right\rangle \\ &=\ \left\langle f,\phi_{m}\right\rangle - \left\langle f,\phi_{m}\right\rangle \\ &=\ 0 \end{align*}

    Therefore, by assumption, g=0g=0. Thus, f=n=1f,ϕnϕnf= \sum_{n=1}^\infty \left\langle f, \phi_{n} \right\rangle\phi_{n}

  • (b)    (c)(b) \implies (c)

    Assume (b)(b). Then, since f=1f,ϕnϕnf=\sum_{1}^\infty \left\langle f, \phi_{n}\right\rangle\phi_{n},

    f2= n=1f,ϕnϕn2=limNn=1Nf,ϕnϕn2=limNn=1Nf,ϕnϕn2=limNn=1Nf,ϕn2=n=1f,ϕn2 \begin{align*} \| f \|^2 &=\ \left\| \sum \limits_{n=1}^{\infty} \left\langle f, \phi_{n} \right\rangle \phi_{n} \right\| ^2 \\ &= \left\| \lim \limits_{N \rightarrow \infty} \sum \limits_{n=1} ^{N} \left\langle f, \phi_{n} \right\rangle\phi_{n} \right\| ^2 \\ &= \lim \limits_{N \rightarrow \infty} \left\| \sum \limits_{n=1} ^{N} \left\langle f, \phi_{n} \right\rangle\phi_{n} \right\| ^ 2 \\ &= \lim \limits_{N \rightarrow \infty} \sum _{n=1}^{N} | \left\langle f,\phi_{n} \right\rangle |^2 \\ &= \sum \limits _{n=1} ^{\infty} | \left\langle f, \phi_{n} \right\rangle |^2 \end{align*}

The third equation holds because the series converges in the norm sense by assumption. The fourth equation is valid due to the Pythagorean theorem.

  • (c)    (a)(c) \implies (a)

    Assume (c)(c). Then,

    f2=n=1f,ϕn2 \| f \|^2 =\sum \limits _{n=1} ^{\infty}\left| \left\langle f,\phi_{n} \right\rangle \right|^{2}

    Therefore, for all nn, if f,ϕn=0\left\langle f, \phi_{n} \right\rangle=0, then f=0f=0 is true.