12.07.2015 Views

Normed versus topological groups: Dichotomy and duality

Normed versus topological groups: Dichotomy and duality

Normed versus topological groups: Dichotomy and duality

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

<strong>Normed</strong> <strong>groups</strong> 1019. The Semigroup TheoremThis section, just as the preceding one, is focussed on metrizable locally compact <strong>topological</strong><strong>groups</strong>. Since a locally compact normed group possesses an invariant Haar-measure,much of the theory developed there <strong>and</strong> here goes over to locally compact normed <strong>groups</strong>– for details see [Ost-LB3]. In this section G is again a normed locally compact <strong>topological</strong>group. The aim here is to prove a generalization to the normed group setting of thefollowing classical result due to Hille <strong>and</strong> Phillips [H-P, Th. 7.3.2] (cf. Beck et al. [BCS,Th. 2], [Be]) in the measurable case, <strong>and</strong> to Bingham <strong>and</strong> Goldie [BG] in the Baire case;see [BGT, Cor. 1.1.5].Theorem 9.1 (Category (Measure) Semigroup Theorem). For an additive Baire (resp.measurable) subsemigroup S of R + , the following are equivalent:(i) S contains an interval,(ii) S ⊇ (s, ∞), for some s,(iii) S is non-meagre (resp. non-null).We will need a strengthening of the Kestelman-Borwein-Ditor Theorem, Th. 1.1.involving two sets. First we capture a key similarity (their <strong>topological</strong> ‘common basis’,adapting a term from logic) between the Baire <strong>and</strong> measure cases. Recall ([Rog2, p. 460])the usage in logic, whereby a set B is a basis for a class C of sets whenever any memberof C contains a point in B.Theorem 9.2 (Common Basis Theorem). For V, W Baire non-meagre in a group Gequipped with either the norm or the density topology, there is a ∈ G such that V ∩ (aW )contains a non-empty open set modulo meagre sets common to both, up to translation.In fact, in both cases, up to translation, the two sets share a norm G δ subset which isnon-meagre in the norm case <strong>and</strong> non-null in the density case.Proof. In the norm topology case if V, W are Baire non-meagre, we may suppose thatV = I\M 0 ∪ N 0 <strong>and</strong> W = J\M 1 ∪ N 1 , where I, J are open sets. Take V 0 = I\M 0 <strong>and</strong>W 0 = J\M 1 . If v <strong>and</strong> w are points of V 0 <strong>and</strong> W 0 , put a := vw −1 . Thus v ∈ I ∩ (aJ). SoI ∩ (aJ) differs from V ∩ (aW ) by a meagre set. Since M 0 ∪ N 0 may be exp<strong>and</strong>ed to ameagre F σ set M, we deduce that I\M <strong>and</strong> J\M are non-meagre G δ -sets.In the density topology case, if V, W are measurable non-null let V 0 <strong>and</strong> W 0 be the setsof density points of V <strong>and</strong> W. If v <strong>and</strong> w are points of V 0 <strong>and</strong> W 0 , put a := vw −1 . Thenv ∈ T := V 0 ∩ (aW 0 ) <strong>and</strong> so T is non-null <strong>and</strong> v is a density point of T. Hence if T 0comprises the density points of T, then T \T 0 is null, <strong>and</strong> so T 0 differs from V ∩ (aW ) bya null set. Evidently T 0 contains a non-null closed, hence G δ -subset (as T 0 is measurablenon-null, by regularity of Haar measure).

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!