The study of graph structure has advanced in recent years with great strides: finite graphs can be described algebraically, enabling them to be constructed out of more basic elements. Separately the properties of graphs can be studied in a logical language called monadic second-order logic. In this book, these two features of graph structure are brought together for the first time in a presentation that unifies and synthesizes research over the last 25 years. The authors not only provide a thorough description of the theory, but also detail its applications, on the one hand to the construction of graph algorithms, and, on the other to the extension of formal language theory to finite graphs. Consequently the book will be of interest to graduate students and researchers in graph theory, finite model theory, formal language theory, and complexity theory.

The top-down and bottom-up tree transducer are incomparable with respect to their transformation power. The difference between them is mainly caused by the different order in which they use the facilities of copying and nondeterminism. One can however define certain simple tree transformations, independent of the top-down/bottom-up distinction, such that each tree transformation, top-down or bottom-up, can be decomposed into a number of these simple transformations. This decomposition result is used to give simple proofs of composition results concerning bottom-up tree transformations. A new tree transformation model is introduced which generalizes both the topdown and the bottom-up tree transducer. Introduction. Finite state transformations (fst) of trees into trees were introduced and studied by Rounds [6] and Thatcher [8]. Finite state transformations are meant to be a model of the kind of tree transformations which are investigated, for instance, in the study of transformational grammars in linguistics and in the study of syntax-directed translations of context-free languages in compiler theory.

Macro tree transducers are a combination of top-down tree transducers and macro grammars. They serve as a model for syntax-directed semantics in which context information can be handled. In this paper the formal model of macro tree transducers is studied by investigating typical automata theoretical topics like composition, decomposition, domains, and ranges of the induced translation classes. The extension with regular look-ahead is considered.

Summary. The notion of a branching process is introduced, as a formalization of an initial part of a run of a Petri net, including nondeterministic choices. This generalizes the notion of a process in a natural way. It is shown that the set of branching processes of a Petri net is a complete lattice, with respect to the natural notion of partial order. The largest element of this lattice is the unfolding of the Petri net.

No abstract