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 of finite erfcx enclosures
Statement
If Finite certified evaluation of the scaled complementary error function returns success for , the
returned interval contains and passes the requested
absolute-radius test. A relative certificate is valid only if zero is outside
the interval and the required relative-radius test passes. Every execution
with finite caps terminates. The theorem makes no success claim for every
or finite cap.
Facts & Assumptions
Given: A successful return of Finite certified evaluation of the scaled complementary error function using a sound outward interval backend.
Certified range evaluation for elementary real and complex functions supplies the finite square-root range call for ; soundness of that real output is given by Certified real elementary-function evaluations enclose their ranges.
Outward-rounded interval, rectangle and ball operations and Outward-rounded operations are inclusion-isotone and enclose DAG evaluations: every finite arithmetic interval operation encloses its exact result.
Finite continued-fraction enclosure for erfcx: each finite-depth continued- fraction interval contains the exact value.
Signed finite remainder for the erfcx asymptotic expansion: the signed remainder interval formed from and contains the exact value for every .
Certified special-function enclosure result: success requires a containing interval with the stated radius, and relative certification requires zero exclusion.
Finite certified evaluation of the scaled complementary error function: the procedure returns only a continued-fraction or signed-asymptotic candidate, or a non-success status.
Proof
A continued-fraction candidate contains by [F1]. A signed-asymptotic candidate contains it by [F2]. The interval evaluation of each candidate is sound by [A1] and [A2]. If the procedure intersects both candidates, their common exact value lies in the intersection; otherwise the selected candidate remains an enclosure.
Thus every interval eligible for a success return contains the exact value. The finite arithmetic and intersection preserve inclusion by [A2] and the algorithm definition.
The procedure reports success only after the absolute-radius test in [F3] and sets a relative flag only when the interval excludes zero and the relative-radius test passes. No width decrease or universal success is asserted.
The stage loop is finite; every continued-fraction recurrence, asymptotic sum, and elementary-range call has finite depth, terms, format, and truncation index. Every branch has an explicit return or proceeds to the next bounded stage. Therefore every capped execution terminates.
Depends on
- Finite certified evaluation of the scaled complementary error function
- Certified range evaluation for elementary real and complex functions
- Certified special-function enclosure result
- Outward-rounded interval, rectangle and ball operations
- Finite continued-fraction enclosure for erfcx
- Signed finite remainder for the erfcx asymptotic expansion
- Outward-rounded operations are inclusion-isotone and enclose DAG evaluations
- Certified real elementary-function evaluations enclose their ranges
Used by
- Least-term truncation of the erfcx asymptotic expansion Example
- Subtracting erf from one amplifies absolute input errors in the tail Example
Cited to discharge well-definedness by Finite certified evaluation of the scaled complementary error function.
Dependency tree · two levels
32 results within two dependency steps of this one, each drawn at its shortest distance from it. An arrow runs from a result to what uses it, so the chart reads left to right and ends at this result, which carries a heavier outline. Every node is a link to that result. Click elsewhere on the chart to enlarge it.
Sources
- DLMF, §7.9, continued fractions (standard reference, not scraped)
- DLMF, §7.12(i), equation 7.12.1 (standard reference, not scraped)
- N. M. Temme, “Numerical aspects of special functions,” Acta Numerica 17 (2008), §§3 and 5 (standard reference, not scraped)