Rigid Distinguishing Family Cardinality Bound
Abstract
Rigid distinguishing continuations are bounded by the output alphabet.
Theorem 1.1 (A rigid distinguishing family injects into the output alphabet).
Proof. Machine-checked in Lean as D5/S0/Automata/DistinguishingFamilyCardinalityBound.card_le_card_output_of_rigid_continuations (✓ std3). ∎
Source. Repository-derived.
Commentary.
The types A, O and I range over arbitrary universes. D is a domain of words, T is a total output function, and C is a distinguishing family indexed by I. Fintype denotes a finite-type instance; append is list concatenation.
Rigidity is required only for the family’s own prefixes and is displayed explicitly. For a nontrivial index type, continuation_constant constructs one continuation shared by all distinct pairs. Evaluating T after that continuation gives an injection into O. The subsingleton case uses the output T of the empty word.
Theorem 1.2 (The certificate is bounded by both outputs and states).
Proof. Machine-checked in Lean as D5/S0/Automata/DistinguishingFamilyCardinalityBound.distinguishing_certificate_bounded_by_output_alphabet (✓ std3). ∎
Source. Repository-derived.
Commentary.
S is an arbitrary state type and M is a DFAO over A with outputs in O. In addition to the finite-type instances, the statement requires CorrectOn(M,D,T) and the same explicit rigidity hypothesis.
The first conjunct uses the cardinality theorem and its live continuation_constant lemma. The second applies the frozen state_lower_bound_of_distinguishing_family theorem. No rigidity property of powers encodings is asserted here.