enlage_buchi_acceptance_set ideas
-
enlarge_buchi_acceptance_set_here()
can probably be sped up by doing a first pass to detect SCC without rejecting cycles. (Just set the acceptance to co-Büchi and look for rejecting SCCs.) Those SCCs can be marked as fully accepting without having to check each edge individually. -
I believe propagate_marks_here()
is a cheap approximation ofenlarge_buchi_acceptance_set_here()
; maybe the later should start by calling the former. -
enlarge_buchi_acceptance_set_here()
seems to have a positive effect on the output size oftgba_determinize()
(see a comment in #570 (closed)), but I we should also consider the runtime cost (hence the previous items).