> Hi, > > On Tue, Dec 10 2019, Jan Hubicka wrote: > > Hi, > > I think the updating should treat self recursive edges as loops: that is > > calculate SUM of counts incomming edges which are not self recursive, > > calculate probability of self recursing and then set count as > > SUM * (1/(1-probability_of_recursion)) > > we do same thing when computing counts withing loops. > > > > Martin, I woner how updating works for indirectly self recursive edges? > > It does not special-case them in any way, if that was the question.
Well, it should, summing only frequencies of edges entering the SCC is not going to give a good answer. If SCC is having single entry and forms a loop the generalization of algorithm above will work, but for general graphs this is hard problem. I suppose we could special case self recurisve nodes (since they are most common) and apply some fixed scale for other SCCs? Also i wonder if the updating is done in RPO so we do not account edges not updated yet? Honza > > Martin
