Operator Theory and Operator Algebras

点击翻译按钮获取中文摘要

Algebras exist as an abstract object in mathematics, and one tries to understand their underlying structure and behaviour. A tool that has often proven to be great value in studying abstract algebras is to study their representations as "concrete" algebras of linear transformations on a particular class of vector spaces known as Hilbert spaces. The goal of the proposed research is to explore diverse open questions in operator theory and operator algebras.  We mention three explicit objectives below. First, we seek to determine whether every continuous representation of a total reduction algebra is completely bounded.  An operator algebra is said to be a total reduction algebra if,  whenever it is represented as an algebra of operators on a Hilbert space, each invariant subspace admits a topological complement which is also invariant for the algebra. This generalises the important notions of amenability and nuclearity of certain algebras of operators.  In previous joint work, we have demonstrated that each commutative total reduction algebra is similar to a nuclear C*-algebra, resolving (in the commutative setting) a problem which was open for over thirty years. Our first problem is a logical next step in the study of non-commutative total reduction algebras. Second, we propose problems from the theory of single linear operators acting on a Hilbert space, transposed to the setting of elements of C*-subalgebras of operators. I am suggesting two new questions (amongst several such problems that I have) of this kind. One of these is the problem of extending a theorem of Specht which describes unitary equivalence of matrices in terms of a comparison of their traces on all words in two non-commuting variables to the setting of C*-algebras. This is joint work with Y. Zhang. We also seek to extend recent results by myself and other co-authors that try to determine the structure of an operator by considering all of its "off-diagonal corners" to the C*-algebra setting. Third, we explore whether every quasidiagonal operator can be approximated arbitrarily well by algebraic, quasidiagonal operators.  Block-diagonal operators are those whose "building blocks" are finite-dimensional matrices, and quasidiagonal operators are those which can be approximated by block-diagonal operators. Quasidiagonality has played a crucial role in the classification of nuclear C*-algebras, and it has proven to be an interesting but extremely subtle property. We have recently obtained a new characterisation of those operators that can be approximated by algebraic, quasidiagonal operators, and we hope that this may lead to an answer to the above question. We also seek to understand when the direct sum of a contractive operator and a normal operator whose spectrum is the unit disc is quasidiagonal. Each of these questions opens several potential avenues of investigation and training for HQP, introducing them to central areas of research in operator theory.