Abstract. We prove some new characterisations of honesty of the perturbed semigroup in Kato's Perturbation Theorem on abstract state spaces via three approaches, namely mean ergodicity of operators, adjoint operators and uniqueness of the perturbed semigroup. We then apply Kato's Theorem on abstract state spaces and the honesty theory linked to it to the study of quantum dynamical semigroups. We show that honesty is the natural generalisation of the notion of conservativity.
IntroductionThis paper originates from a perturbation theorem for substochastic semigroups (positive semigroups which are contractive on the positive cone) which is known as Kato's Perturbation Theorem [14,2]. The main idea in Kato's original work in [14] tells us that if A is the generator of a substochastic semigroup on L 1 and B is a positive operator satisfying certain conditions, then there is an extension G of A + B that generates a perturbed substochastic semigroup. Although this theorem is useful as a generation result, with applications in various problems such as birth and death problems, fragmentation problems [14,4] and transport equations [3,24], (see [5, Chapters 7-10] for a survey of the results), our interest in this theorem lies mainly in the honesty theory derived from it.Honesty is a property of the perturbed semigroup in Kato's Theorem. We will give the precise technical definition of honesty in Section 2; for now, it suffices to think of honesty theory as the study of the consistency between the perturbed semigroup and the system it describes in the following sense: A substochastic semigroup on L 1 is often used to model the time evolution of some quantity. The nature of the modelled process often requires that the described quantity should be preserved, i.e. the semigroup describing the evolution is conservative (stochastic). However, in some cases, the semigroup turns out not to be conservative even though the modelled system should have this property.This phenomenon is what we will call dishonesty. For a system modelled by a strictly substochastic semigroup, we have a loss term representing the loss due to the system. Dishonesty in this case would mean that the described quantity is lost from the system faster than predicted by the loss term.Apart from consistency with the system, honesty of the semigroup is also interesting from a purely mathematical point of view. It is a well-known result [5, Theorem 6.13