ORIGINS IN SINGLE OPERATOR THEORY 7 spectrum (namely the unit circle), whereas both A and B are Fredholm operators whose Fredholm indices are, respectively, - 1 , and +1. It follows that A and B cannot be equivalent. Let us return now to quasitriangular operators. In 1968, Halmos showed [37] that there exist operators which are not quasitriangular (the shift is one such). Douglas and Pearcy later clarified the situation somewhat by proving [28]. THEOREM 1.16. If A is quasitriangular, then index(/4 - XI) 0 for every A G C for which A - XI is semi-Fredholm, This implies Halmos' earlier result, because the shift is a Fredholm operator whose index is - 1 . In a series of papers which contain a deep analysis of the spectral properties of operators, Apostol, Foias, and Voiculescu provided a remarkable converse to Theorem 1.16 [3, Corollary 5.5]. THEOREM 1.17. If A is an operator such that index(^ - XI) ^ 0 for every A G C for which A — XI is semi-Fredholm, then A is quasitriangular. Thus, the only obstruction to membership in the class 2,3~ is an index obstruction. Theorem 1.17 also has implications about invariant subspaces it implies that every nonquasitriangular operator is intransitive. For by Theorem 1.17, such an operator A would have a scalar translate A - XI of negative index, hence A* - XI would have an eigenvector, hence A would have an invariant subspace of codimension one. So if there exists a transitive operator on a Hilbert space (and I personally believe that such operators do exist), then it must be quasitriangular.
Purchased from American Mathematical Society for the exclusive use of nofirst nolast (email unknown) Copyright 1984 American Mathematical Society. Duplication prohibited. Please report unauthorized use to cust-serv@ams.org. Thank You! Your purchase supports the AMS' mission, programs, and services for the mathematical community.