General linear method: A survey (Q1089749)

Of the various methods devised as generalizations of the classical method of Euler, two extreme approaches are typically followed. One is to generalize the Euler method through the use of multistep methods; the other is to increase the complexity of one-step methods as in the Runge- Kutta method. In the discussion of Section 2, general linear methods are introduced as a middle ground between these types of generalization, and examples of these methods are given. In Section 3, some of the elementary theoretical properties concerned with convergence of the method are considered and a discussion of truncation error follows in Section 4. Some of the difficulties of implementation that these methods share with Runge-Kutta methods are discussed in Section 5 and, in Section 6, properties concerned with A-stability and its non-autonomous and non- linear counterparts are defined and interrelated. Finally, in Section 7, we indulge in some speculation about the possible construction of efficient software using a suite of general linear methods.