Reducible representations of abelian groups (Q5944547): Difference between revisions

It is not known if each bounded linear invertible operator in a reflexive Banach space \(X\) has a nontrivial invariant subspace common with its inverse. By the language of representation theory this problem can be formulated as follows: is it true that any representation of the group \(\mathbb Z\) in \(X\) is reducible? In this paper the reducibility (the existence of a nontrivial closed invariant subspace) is proved under additional assumptions, i.e. for representations of arbitrary abelian groups. Let \(T\) be the representation of an abelian group \(G\) (which is not supposed to be topological) in a complex reflexive Banach space \(X\). Assume that there exist a continuous involution \(J: X\to X\) and nonzero elements \(u\in X\), \(v\in X^*\) such that \(JT(g)=T(-g)J\), \(Ju=u\), and the matrix element \(g\to\langle T(g)u,v\rangle\) is a positive definite function on \(G\). Then \(T\) is reducible. This is the main result of the paper. It is applied to the class of translation invariant function spaces on locally compact abelian groups \(G\) possessing the following property: for compact subsets of \(G\) the restriction operator is continuous in the \(L^1\)-norm with respect to the Haar measure. Let \(E\) have this property and contain a nonzero bounded Borel function \(u\) such that \(u(g)=\overline{u(-g)}\), \(g\in G\). Then \(E\) has a nontrivial translation invariant subspace. This implies the reducibility of the regular representation in weighted \(L^p\)-spaces if the weight \(\omega\) is essentially even (i.e. the function \({\omega(-g)}\over{\omega(g)}\) is bounded).