Research the different voter realizations, that's an interesting topic

[balodja]

1. Should the voter be "folding" (e.i. a + (b + (c + ...))) or "treeing" (i.e. ((...) + (...)) + ((...) + (...)))?
2. Should the voter use IC-logic (i.e. summators and all) or just comparison values(i.e. naive realization in Sesyrel.FaultTree.Base)?
3. Should the elimination order finding algorithms be randomized? (I think they should, and now they are not)
4. Should the elimination be done with more sophisticated algorithms? Tho simple trees with voting varies VASTLY in complexity depending on the presence of relations between base nodes, on the elimination algorithm. And none of that algorithms is even suboptimal. So... I need more!

[rakhimov]

Some nice thing about static fault trees (i.e., propositional directed acyclic graphs or Boolean formulas)
is the existence of canonical forms,
for example, CNNF (conjunctive negation normal form), DNNF(disjunctive ...), BDD (binary Shannon decomposition), etc.

I wonder if dynamic fault tree systems can have such normalized forms,
so that instead of dealing with various flavors of "gates"/"logic"/"connectives" (e.g., Vote, PAND),
all can be simplified to some sort of normal form (i.e., AND/OR/NOT).
With that, I believe you would have to implement fewer algorithms or "special handling".

On the other hand, this may be a wishful thinking due to complex ordering constraints.