I'm struggling with a following task: Let $A = (Q, Σ, δ, q_{0}, F)$ be a DFA, in which every state is attainable (attainable state meaning that there exists a word that after reading that word, the automata will end up in said state). Is it possible to ALWAYS modify F in such way that the automata will be minimal?
So basically we have to think if it is always possible to modify F in an automata such that the automata will be minimal after that modification.
I know, that if we have a DFA with N states, then we have $2^N$ options to choose F from.
We have to find a counterexample DFA in which for every F picked there will be an option to further minimize the DFA.
I can't find a counterexample for that, could you help me with this task or maybe somehow narrow it down to some examples?
THANKS!
Possible Method: Let's assume there are no un-reachable nodes. Then DFA minimization works on the assumption that there are equivalent, non-distinguishable nodes. Lets say there are two equivalent nodes, then one can always set one of the equivalent nodes to be in F and the other to not be in F, and thus distinguish them.
But what if there are more than 2 equivalent nodes? In such cases, even after changing F, some 2 nodes may remain non-distinguishable. Concretely consider the following example:
In this case, the DFA accepts all words of size 2 or more. Can we find some subset of accepting states for which this is the minimal DFA? $C_0,C_1$ must be distinguished, so one must be accepting and the other not. Similarly, $D_0,D_1$ must be distinguished, so one must be accepting and the other not. But then, the accepting states, say $C_0,D_1$ can be merged and so can $C_1,D_0$ as they are non-distinguishable. So we cannot find any subsets of accepting states for which this is minimal.