On a Method of Multiprogramming (Monographs in Computer Science)

Among all the interests in parallelism, there is an essential and fundamental one that has remained largely unexplored, namely the question of how to design parallel programs from their specification. And that is what this book is about. It proposes a method for the formal development of parallel programs - multiprograms as we have preferred to call them -, and it does so with a minimum of formal gear, viz. with the predicate calculus and with the meanwhile well-established theory of Owicki and Gries. The fact that one can get away with just this theory will probably not convey anything to the uninitiated, but it may all the more come as a surprise to those who were exposed earlier to correctness of multiprograms. Contrary to common belief, the Owicki/Gries theory can indeed be effectively put to work for the formal development of multiprograms, regardless of whether these algorithms are distributed or not. That is what we intend to exemplify with this book.