Termination of Logic Programs Using Various Dynamic Selection Rules

We study termination of logic programs with dynamic scheduling, as it can be realised using delay declarations. Following previous work, our minimum assumption is that derivations are input-consuming, a notion introduced to define dynamic scheduling in an abstract way. Since this minimum assumption is sometimes insufficient to ensure termination, we consider here various additional assumptions on the permissible derivations. In one dimension, we consider derivations parametrised by any property that the selected atoms must have, e.g. being ground in the input positions. In another dimension, we consider both local and non-local derivations. In all cases, we give sufficient criteria for termination. The dimensions can be combined, yielding the most comprehensive approach so far to termination of logic programs with dynamic scheduling. For non-local derivations, the termination criterion is even necessary.