Would Lifting help tilt the balance?
Lifting is one of the many transformations to reduce the size of a SAT encoding
- Uses variablized actions and fluents & unification (instead of ground ones)
- Equivalent failures need be discovered fewer times
- Kautz et. al. [KRR-96] used lifted version of causal encodings to claim that causal encodings are the smallest
BUT, Lifting is an encoding-independent transformation (Can be applied to state-based encodings)
- Lifted causal encodings will still be dominated by best state-based encodings