A Tribute to Dean Rosenzweig Egon B¨ orger Universit`a di Pisa, Dipartimento di Informatica, I-56125 Pisa, Italy boerger@di.unipi.it 1 Dubrovnik Dean Rosenzweig passed away in the second week of January 2007. I met Dean for the first time 24 years ago, during the winter School Foundation of Computation Theory Profs. Rasiowa, Karpinski and Kirin had organized at the Inter-University Centre for Post-graduate studies in Dubrovnik from 16.1. - 29.1.1983. Dean attended my lectures on Complexity of Logical Theories and I noticed him as a particularly bright, sharp and very knowledgeable young colleague with a strong taste for elegance in mathematical matters and a wide range of interests. I vividly remember strolling with Dean and his friends long night hours through the streets of the historic Dubrovnik and along the town wall, discussing issues of logic, complexity, semantics whilst enjoying the splendid view on the town, its harbor, the hills and the sea. We shared the love for Dubrovnik and the town became our fate. 2 ASM Models for Logic Programming Concepts In fact we met again in Dubrovnik seven years later, during two invited lectures I delivered at the Logic and Computer Science Conference ( LIRA, 6.9. - 9.9.1990) and the International Summer Seminar on Artificial Intelligence (CAS, 3.9. - 7.9.1990). Dean was again in the audience when I explained The Dynamic Algebra Approach to Semantics of Prolog and Prolog III and the use of ASMs to construct a practically useful Formal Model for Semantics of Constraint Logic Program- ming Systems. In those lectures I presented the use of ASMs to build ground models [10] for logic programming systems and explained among others the problem Prolog experts had challenged me to solve as test for the usefulness of such models, namely to explain through them the intricate optimization features of the Warren Abstract Machine implementation of Prolog and to verify the correctness of the WAM, which represented the state-of-the-art of logic programming practice and at the time was understood only by a small number of experts worldwide. Dean enthousiastically joined the project to which he brought his detailed knowledge of logic programming implementation issues. 1 Through intensive work both of us enjoyed enormously, started right after my talks in Dubrovnik and going through the fall and winter of 1990/91, when Dean visited me repeatedly at IBM in Heidelberg where I was spending a sabbatical and in Pisa, we built a series of 12 stepwise refined ASM models linking my ASM model of Prolog [4,5] to an ASM model for the Prolog implementation by WAM code, proving the correctness of each refinement step and thus verifying the WAM implementation of Prolog. Less than a year later I could report the successful outcome of this work in an invited talk on the Correctness proof for a class of Prolog Compilers on Warren’s Abstract Machine, delivered to the 13th International Conference on Information Technology Interface (ITI’91) in Dubrovnik-Cavtat (10.6. - 14.6.1991), with Dean again present in the audience, and published in [25,26,30]. The Prolog-to-WAM work represented the first full-fledged ASM method case study that proved the potential of using ASMs for a practical and effectively verifiable design of a complex system, from its high-level specification to an optimized implementation. For three reasons the work became a landmark in the short history of ASMs. 1 On Dean’s web page I found a reference to an apparently unpublished paper [44] he had never mentioned to me that seems to document the source of this knowledge.