Summarized Dimensions Revisited

The task of summarizing memory regions becomes increasingly important in the analysis of data structures and the analysis of recursive functions. Calculating summaries of memory regions containing several abstract variables becomes quite subtle when using relational abstract domains. We address this challenge using two principal operations, fold and expand, that calculate such summaries for any numeric relational domain and argue optimality. We thereby generalize an earlier approach in the literature that was geared towards memory cells modelled by a single abstract variable.