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.
Soundness and termination of finite partition enclosures
Statement
Let be finite and nonempty, let be an exact finite positive weight, and let be a finite observable. Suppose the finite rational input intervals in the definition of alg-csm-frontier-interval-finite-partition contain every exact and . The algorithm terminates after finitely many operations. If it returns finite intervals and , then they contain and , respectively. If it also returns an expectation interval , then its denominator interval has positive lower endpoint and contains the exact normalized expectation , where . An overflow, non-finite operation, or nonpositive denominator lower endpoint produces an unresolved status and carries no normalized-expectation enclosure claim.
Statement references
Facts & Assumptions
Given: The exact finite positive weights and finite observable on the enumerated nonempty state space, the certified rational input intervals, and the algorithm's finite or unresolved return branch.
Certified finite partition and expectation enclosure: the procedure converts rational input endpoints outward, constructs finite interval sums for weights and weighted observables, and divides only when the partition interval's lower endpoint is positive.
Dyadic floating-point format and directed rounding: for an exact rational intermediate within the finite range, directed rounding returns a greatest representable lower endpoint and a least representable upper endpoint; overflow is reported and has no finite-error claim.
Real intervals, complex rectangles and ball enclosures: a finite expression DAG is a finite directed acyclic graph with input and constant nodes; sharing a node reuses the same exact intermediate value.
Outward-rounded interval, rectangle and ball operations: addition, subtraction, multiplication, division away from zero, and squaring use the stated exact endpoint hulls followed by downward and upward endpoint rounding; division by an interval containing zero returns a split or unresolved status rather than a finite quotient.
Directed dyadic rounding encloses with explicit underflow and overflow bounds: each finite directed endpoint rounding brackets the exact real endpoint; overflow is explicitly not a finite enclosure result.
Finite real interval operations enclose their exact values: for a finite acyclic expression DAG whose input and constant intervals enclose their values, each finite computed node interval encloses its exact value, including reused shared nodes.
Proof
Given: [F1]–[F6] and the certified finite rational input intervals in the Statement.
The finite enumeration supplies one input interval for each exact weight and observable. By [F2] and [F5], when endpoint conversion is finite, rounding the lower endpoint down and upper endpoint up preserves containment of each input interval; if conversion overflows, the algorithm returns unresolved and the theorem makes no finite-enclosure claim. The partition sum and weighted sum are finite acyclic expression DAGs by [F1] and [F3].
Order the nodes of each finite DAG topologically. Each input node encloses its exact value by step 1.1. At every operation node with a finite result, [F4] gives the interval extension prescribed for that operation and [F6] proves that its value encloses the exact operation on the predecessor values. Induction over the finite node order therefore places in and in whenever those intervals are returned.
The exact partition sum satisfies because is nonempty and every . If the returned lower endpoint of is positive, then and [F4] defines outward interval division; applying [F6] to this final operation gives . If the lower endpoint is nonpositive, or any operation is non-finite, [F1] returns an unresolved status and no quotient enclosure is claimed. Finally, the procedure performs finitely many conversions and finitely many operations because is finite, and every operation returns a finite interval or a status; hence it terminates.
Depends on
Certified finite partition and expectation enclosure
Dyadic floating-point format and directed rounding
Real intervals, complex rectangles and ball enclosures
Outward-rounded interval, rectangle and ball operations
Directed dyadic rounding encloses with explicit underflow and overflow bounds
Finite real interval operations enclose their exact values