Statement

Let GG be an with μ\mu, and give its G^\widehat G the μ^\widehat\mu. The Fourier inversion theorem states that if fL1(G,μ)f\in L^1(G,\mu) and its f^\widehat f belongs to L1(G^,μ^)L^1(\widehat G,\widehat\mu), then

f0(x)=G^f^(γ)γ(x)dμ^(γ)f_0(x)=\int_{\widehat G}\widehat f(\gamma)\gamma(x)\,d\widehat\mu(\gamma)

defines a continuous function vanishing at infinity, and f0=ff_0=f . The inverse integral converges absolutely for every xx and uses the fixed dual normalization. Hence f0(x)=f(x)f_0(x)=f(x) at every point when ff itself is continuous.

Why the hypotheses matter

The two L1L^1 assumptions make both transforms ordinary absolutely convergent integrals. Without integrability of f^\widehat f, inversion may still hold in L2L^2, by summability, or in the sense of distributions, but the displayed integral need not converge pointwise. The general theorem and its normalization are developed in Rudin, Chapter 1.

Relation to duality

Applying Fourier transformation on G^\widehat G and using the evaluation isomorphism GG^^G\cong\widehat{\widehat G} from the gives

f^^(x)=f(x)\widehat{\widehat f}(x)=f(-x)

in additive notation under the convention f^(γ)=Gf(x)γ(x)dμ(x)\widehat f(\gamma)=\int_Gf(x)\overline{\gamma(x)}\,d\mu(x). The reflection changes when the transform convention changes.

Standard examples

For G=RnG=\mathbb R^n, the theorem is Euclidean Fourier inversion with the constants determined by the exponential convention. For G=ZG=\mathbb Z, it reconstructs an absolutely summable sequence from its Fourier transform on the circle. On a compact abelian group, the dual is discrete and the inverse integral becomes a sum over characters.

References
  1. Walter Rudin, Fourier Analysis on Groups, Wiley-Interscience, 1962. Wiley DOI record. Relevant: Chapter 1, inversion and duality.
  2. Gerald B. Folland, A Course in Abstract Harmonic Analysis, 2nd ed., CRC Press, 2016. DOI record. Relevant: Chapter 4, Fourier inversion on locally compact abelian groups.