Out of Control: Reducing Probabilistic Models by Control-State Elimination
summary
Presents a program-level reduction technique for PRISM-style probabilistic models that eliminates control states by computing summary behavior of adjacent control-flow locations. It treats finite-domain program variables as alternative program counters, performs property-directed, fully automatic reductions that preserve the operational semantics, and supports unspecified parameters to shrink the resulting Markov chain before model checking.
AI-written summary · about