s-arash
Use of magic set optimization?
From a look at the implementation and accompanying scientific papers it seems as if ascent does not make use of the popular magic sets optimization for semi-naïve bottom-up implementations. At least I couldn't find any code that looks like it, nor find any mention of magic or adorn in the actual source code or the accompanying scientific papers.
The lack of explicit input/output relation declarations makes me further assume it doesn't use magic sets (as these would define the wells/sinks of the program implicit dependency graph and could thus guide the optimization)?
@s-arash @kmicinski Is this optimization something that's planned/feasible for ascent with its lattices?
Either way it would probably make sense to discoverably document whether or not it currently uses said optimization.
