The F. and M. Riesz theorem on certain transformation groups (Q1114111)

Half of the F. and M. Riesz theorem (1916) is that a measure on the circle whose Fourier coefficients vanish for negative indices is absolutely continuous with respect to linear measure on the circle. The other half is that linear measure is absolutely continuous with respect to the measure, provided the measure does not vanish everywhere. So the whole theorem is that the two have the same null sets. Let G be a compact abelian group, \(\hat G\) its character group, and E a subset of \(\hat G.\) Because of the first half of the Riesz theorem, the set E is said to be a Riesz set if each measure on G whose Fourier transform vanishes off E is absolutely continuous with respect to the Haar measure of G. As far as I know there is no term for E if the other half of the Riesz theorem also holds, so let's call such a set a double Riesz set. Suppose our group G acts on a locally compact Hausdorff space X. Then one may translate measures on X by members of G, \((T_ g\mu)(B)=\mu (g^{- 1}(B))\), and one may convolve measures on X with functions on G, \[ f*\mu =\int_{G}T_ g\mu f(g)dg,\quad \mu \quad bounded,\quad f\in L^ 1(G). \] By the spectrum of the bounded measure \(\mu\) one means the set of those characters \(\chi\) whose convolution with \(\mu\), \(\chi\) *\(\mu\), does not vanish. (So if G acts on itself, the spectrum of \(\mu\) is the set where its Fourier transform does not vanish.) The bounded measure \(\mu\) is said to translate continuously if the norm of the difference \(\mu - T_ g\mu\) is continuous in the group element g, while \(\mu\) is said to be quasi-invariant if it and its translates have the same null sets. If \(\mu\) is quasi-invariant, then it translates continuously, but not vice versa. (If G acts on itself, then a measure translates continuously iff it is absolutely continuous with respect to Haar measure, while a nonvanishing measure is quasi-invariant iff it and Haar measure have the same null sets.) Let G act freely on X, i.e., if the group element g fixes a point of X, then g is the identity. With this hypothesis, the author proves: (i) if the spectrum of \(\mu\) is a Riesz set, then \(\mu\) translates continuously; (ii) if the spectrum of \(\mu\) is a double Riesz set, then \(\mu\) is quasi- invariant. There are other results of a similar nature, but on stability of spectra under Lebesgue decompositions. The proofs, which require great attention to detail, are by the method of disintegration of measures. The title of the paper is somewhat misleading since the theorems provide us with neither Riesz sets nor double Riesz sets. In other words, to obtain the conclusion of (ii) say, we need to know at the outset some double Riesz sets in \(\hat G.\) The prototype of the paper is old work of \textit{K. de Leeuw} and \textit{I. Glicksberg} [Acta Math. 109, 179-205 (1963; Zbl 0126.121)].