Context
The original source report described a finite check but its program and certificate were unavailable. This replacement audit publishes both and does not claim to reproduce the original expression corpus.
Definitions
- The construction oracle evaluates the quotienting expression as a relation on the normalized derivative carrier of the target expression.
- The comparison oracle explores the synchronized derivative product and tests whether some reachable prefix state accepts the quotienting expression while the target state accepts the supplied suffix.
Hypotheses and scope
- The alphabet is .
- The finite domain contains 27 quotienting expressions, 15 target expressions, and all suffixes of length at most four.
Proof or evidence
The first saved checker reports 12,555 comparisons and zero ordinary mismatches while detecting reversed concatenation, one-step star truncation, reversed letter edges, and loss of the distinguished start state. A second checker independently parses the corpus, uses Antimirov partial derivatives for construction and Thompson epsilon-NFAs for its oracle, and reproduces the exact semantic transcript. The review checker uses an epsilon-free Glushkov position-automaton oracle, reproduces the same transcript, and confirms the unbounded-unfolding witness for k=0 through 64.
Verification notes
The saved checks use Brzozowski derivatives, Antimirov partial derivatives, Thompson epsilon-NFAs, and Glushkov position automata in distinct roles. The fixed expression strings, alphabet, and case ordering remain shared by the original pair; the review checker additionally shares its parser with its two sides. All certificates were reproduced before publication.
Limitations
- Finite evidence does not replace the structural proof for arbitrary regular expressions.
- The replacement audit deliberately matches the earlier comparison count but does not claim the same test corpus.
- The four mutation controls are targeted and do not establish general fault-detection completeness.
- The audits share the fixed expression corpus and case ordering, and the review audit shares one parser between its two automata constructions.