How statement and proof provenance work
The first chip identifies the source of the statement or construction; the second identifies the source of its local proof or verification.
- Literature-sourced: the exact statement appears in a cited source; only wording and notation differ.
- AI-adapted: a semantically identical restatement of literature-sourced material, modulo indexing, notation, and boundary cases adopted by the library.
- AI-generated: a genuinely novel statement formulated by AI, with no source for the claim itself.
These labels describe origin, not correctness: citations and verification chips remain separate evidence.
Finite-Automata Transition and Determinisation Interfaces: Examples
1 · Prerequisites
- Construction of the Natural Numbers
- Construction of the Real Numbers via Cauchy Sequences
- Construction of the Real Numbers via Dedekind Cuts
- Countability and Uncountability
- Deterministic Finite Automata and Regular Languages
- Finite Counting, Factorials and Binomial Coefficients
- Finite-Automata Transition and Determinisation Interfaces
- Formal Languages, Encodings, and Decision Problems
- Foundations of the Real Numbers for Analysis
- Linear Recurrences and Rational Generating Functions
- Relations, Functions, and Quotients
- Roots, Rational Powers, and Classical Inequalities
- The ZFC Axioms and the Basic Set Constructions
2 · Summary
The first example evaluates closure and six word transitions in a fully specified three-state automaton. The next computes its reachable subset DFA and proves the table exhaustive for all words. A two-state counterexample then separates the failures caused by omitting initial closure and by omitting closure after a letter.
3 · Logical flowchart
4 · Definitions, theorems and proofs
None yet.
5 · Examples, counterexamples and false statements
Epsilon closure and word transitions on three states
Example
Let , , start , and , where the displayed states and letters are distinct. Specify the entire transition table by
| State | |||
|---|---|---|---|
Then and . The state sets from on are, respectively, . Thus are accepted and are rejected.
Facts & Assumptions
Given: The nine table entries, start , and final set above; write .
Epsilon closure is finite saturation; the empty-word value is the initial closure and each next letter takes a successor union followed by closure. Acceptance tests nonempty intersection with the final set (Epsilon-NFAs, finite epsilon closure, and extended word transition).
Verification
The table gives a subset of at each of its nine state/label pairs, so the transition is total on the required domain. The finite sets, start and finals meet the epsilon-NFA requirements. Starting from , saturation adds in one stage and adds nothing thereafter, since neither nor has an epsilon successor. Thus ; more generally if , and otherwise. In particular , and .
The empty-word value is . For , the successor union from this set is , already closed; hence . For , the corresponding union is , so .
After the set is . Its -successor union is , and its -successor union is . Both are closed, giving and . From the -successor union is empty, so .
The sets for each contain , whereas the sets for do not. Intersecting them with gives in the first three cases and in the other three, proving every asserted acceptance and rejection.
Source
The witness is original to this example. The calculation uses the closure and word-transition rules of Gallier, Definitions 3.6–3.7, pp. 30–32, via the local interface.
A complete determinisation table for a three-state epsilon-NFA
Example
Take , , start , and . Define , , , and all other state/label entries to be empty. All state names and letter names within their respective sets are distinct. Put The reachable part of the full subset DFA has start , accepting states , and complete table
| State | ||
|---|---|---|
Exactly these four states are reachable. The other four states of are unreachable.
Facts & Assumptions
Given: The entire NFA prescription above; denote the specified subsets, and .
Epsilon closure is the least epsilon-closed superset, computed by finite saturation (Epsilon-NFAs, finite epsilon closure, and extended word transition).
The subset DFA has , start , and final subsets meeting . Its extended transition equals the NFA word transition from that start and its full state set has elements (Subset construction and its exact reachability invariant).
Induction from zero and successor proves a property for every natural length (The principle of mathematical induction).
Verification
There are nine domain pairs; the three prescribed nonempty entries and the six empty entries all lie in . This makes a total epsilon-NFA with the stated start and finals. Its only epsilon edge is from to , so saturation adds precisely when is present and absent. Hence , and are each closed.
From , the -union is and the -union is . From , the -union is and the -union is . Closing these unions leaves them unchanged. This computes the first two rows.
From , the -union is and the -union is . From , both unions are empty, so both transitions return . Again each result is closed. This computes the last two rows.
The start is , and the words reach , respectively: the middle two follow the transitions , and . Thus each of the four states is reachable.
Every word reaches one of these four states. Indeed at length zero the state is . If a prefix reaches one of the four, then appending or reaches one of the four by the computed rows. Induction on word length therefore gives the assertion for every word; the induction applies to the property that every word of a given length reaches the four-state set.
The eight subsets are . By the two reachability implications, precisely are unreachable. Of the reachable subsets, and , giving exactly the displayed accepting states. The full DFA also designates its unreachable subsets containing as final; the table states the finals of its reachable part.
Source
This independently specified witness instantiates Aho, Lecture 3 §2. The table is for the reachable part of the full-power-set construction; its exhaustiveness follows from the induction above.
Dropping epsilon closure changes the accepted language
Statement refuted
“For every epsilon-NFA, determinisation preserves the accepted language if the start closure is omitted, or if post-letter closure is omitted.” In the first variant only the start changes to ; in the second the start remains and the letter map changes to . Both variants retain the full power set and the final-subset intersection test.
Facts & Assumptions
Given: We test the two stated variants separately, writing .
An epsilon-NFA starts its extended transition at , closes every post-letter successor union, and accepts exactly when the resulting subset meets (Epsilon-NFAs, finite epsilon closure, and extended word transition).
The correct subset DFA has that closed start, those closed successor unions, and final subsets meeting , with exactly the same accepted language (Subset construction and its exact reachability invariant).
Counterexample
Let , , start , and , with . Put , , , and . All four domain entries are specified subsets of the finite state set, so this is an epsilon-NFA. Saturation from gives and then stays there; thus .
In the correct construction the empty word reaches , whose intersection with is , so it is accepted. In the first defective construction the empty word leaves the start at , whose intersection with is empty, so it is rejected. This mismatch occurs before any letter transition, even though all post-letter closures in that variant are retained.
In the second defective construction the start is correctly , but on the single letter its successor union is . Omitting closure leaves that set, disjoint from , so the variant rejects . The correct construction instead closes it to , which meets ; equivalently the original NFA has the accepting path .
Each defective rule still gives a total DFA on the finite power set: its letter outputs are subsets and each of its starts is a subset. The failures are therefore language failures, not missing transitions. The empty word distinguishes the first variant and distinguishes the second, refuting both universal preservation claims.
Source
The two-state witness is original. It tests the two separate closures in Aho, Lecture 3 §2; neither is dispensable in general.