Home Developers Evidence math.special

Public evidence

math.special

Machine-checked verification for calculation version 1.10.6 · tier candidate

CVP status

CVP VERIFIED · CVP protocol 1.0.0-proposed · Core assurance · all published modes

Real erf/erfc, Gamma, lgamma, integer-order J/Y/I/K, Lambert W_0/W_{-1}, Beta, incomplete beta, sinc. Discovery pages lock mode. Not factorial. Not CAS.

Declared partition coverage
PASS · 15/15 declared partitions (erf, erfc, gamma, lgamma, bessel-j, bessel-y, bessel-i, bessel-k, lambert-w, lambert-wm1, beta, incomplete-beta, sinc, normalized-sinc, invalid-domain)
Versions
Calculation 1.10.6 · CVP protocol 1.0.0-proposed · Evidence 2026-10-19.special-full-review-v8
Verification revision
2026-10-19.special-full-review-v8 · 1/1 property · digest a3f6d148d541

Verification

Legacy regression suite: 57/58 calculation tests · Production surface contract 6/6 · 1 failed

Three distinct suites: educational examples on the calculator page, this legacy regression count (57/58 golden + boundary), and the CVP verification suite below. They are not the same number. Property, metamorphic, round-trip, O3, and interface statuses/checks are recorded in the CVP Manifest . SKIP (0/0) means that suite has no published vectors.

    This calculator CURRENT · Public schema 1.10.6 matches · Semantic contract ✓ · Production attested · Public/cache ✓ · Origin ✓

CVP verification suite

Protocol checks from the Evidence Manifest. Distinct from the legacy regression suite above. Verification revision 2026-10-19.special-full-review-v8. Calculation version 1.10.6. CVP 1.0.0-proposed. 1000/1000 golden · 8/8 CVP boundary · 24/24 invalid · 1/1 property · 3/3 metamorphic · 24/24 cross-interface · 1/1 cross-calculator · 6/6 CVP contract. digest a3f6d148d541.

  • Golden (oracle-backed) 1000 / 1000
  • CVP boundary 8 / 8
  • Invalid input 24 / 24
  • Property 1 / 1
  • Metamorphic 3 / 3
  • Cross-interface 24 / 24
  • Cross-calculator 1 / 1
  • CVP contract 6 / 6
  • Interfaces PASS · UI (SSR) / REST / MCP
  • Interface claim ui-ssr is query-result HTML, not a live browser session. Error-path engine·REST·MCP 1/1 (status, code, calculation_version). SSR compared on URL-canonical requested calculations; empty query is idle (not an error) and JSON-typed object/array inputs are REST/MCP-only. Special response contract also checks all 14 mode formulas and unrounded numeric summaries, including erfc/K tails and a subnormal erf input; UI is shared presentation/SSR, not a live browser session.
  • Verified scope all published modes
  • Assurance Core
  • release artifacts aligned PASS
  • schema output required declared PASS
  • schema valid input executes PASS
  • engine success output validates schema PASS
  • schema rejects incomplete input PASS
  • rest success payload validates schema PASS

Fingerprint sha256:5fbfcc9d018c864dfb99283b1b2d9d11a5e1174a08368648093fec31af92a715

Input schema sha256:edc3a9ac82fb12bade39bb5baa124a9e31d9b5289f56b19ca7c02500463344dd · Output schema sha256:5864981f912b9805e21ea9ddf0882a27e408999a948a3b5b94bbe1de8b76fe9c · Build-time SHA-256 digests prove published artifacts agree with each other (release_integrity.scope=workspace). Production attestation GETs canonical URLs twice (public/cached view and origin), including Capability HTML, via npm run attest:production --write (published at /.well-known/calculatorx-production-attestation.json) and fails if the public Capability page advertises a different calculation_version. This is artifact consistency, not an independent immutable release log.

Build & scope

Engine identity and declared limitations.

Tool
special-functions
Engine
special-functions
Engine build
2026-10-19
Evidence core
2026-10-19T04:04:19.258Z
Rendered
2026-10-23T01:27:11.644Z
Last reviewed
—

Assumptions

  • Relative accuracy weakens near zeros of Bessel J/Y. Small positive K tails are retained within Float64 range.
  • Real argument
  • Gamma poles at non-positive integers
  • lgamma is real for x > 0
  • Bessel J integer n=0…8, |x|≤80
  • Bessel Y integer n=0…8, x>0
  • Bessel I integer n=0…8, |x|≤80
  • Bessel K integer n=0…8, x>0
  • Lambert W_0 for x ≥ −1/e
  • Lambert W_{-1} for −1/e ≤ x < 0
  • Complete Euler B(x,y) for x>0, y>0
  • Incomplete B_x(a,b) for a>0, b>0, 0≤x≤1
  • Unnormalized sinc(x)=sin(x)/x, sinc(0)=1
  • Normalized sinc(x)=sin(πx)/(πx), sinc(0)=1
  • The Result Card includes an engine-linked Live graph for erf, erfc, Γ, ln Γ, J_n / Y_n / I_n / K_n, W₀ / W₋₁, B(x, y) vs x (y held), incomplete B_x on [0, 1], sinc, and normalized sinc. Same special-functions engine as REST/MCP. Integer n! stays on /factorial. Not CAS.

Limitations

  • Real, dimensionless binary64 inputs; publishes12 significant digits, no arbitrary-precision/exact claim.
  • Default mode erf and Bessel order0 apply only when absent; explicit empty/null rejects. All supplied numeric aliases must agree; known irrelevant numeric parameters are parsed then ignored.
  • Gamma poles reject. Nonzero Gamma that underflows rejects NUMERICAL_RANGE_ERROR; positive overflow suggests lgamma. lgamma is computed directly for x>0 with a stable Taylor expansion near1 and2.
  • Bessel integer n0–8, |x|≤80, Y/K require x>0; subnormal error and zeros limit relative accuracy.
  • Lambert W0 and W-1 explicit real branches; binary64 -1/e designated endpoint, no tolerance band absorbs adjacent inputs.
  • Beta shapes positive and at most1e6, except exact B(x,1)=1/x or B(1,y)=1/y. Incomplete beta shapes at most1e6. CF capped10000 steps; failure NO_CONVERGENCE.
  • erf/erfc and beta tails can round to zero only on binary64 underflow, not an absolute display cutoff. Regularized incomplete beta evaluated separately in logs.
  • Normalized sinc uses sinpi reduction and returns zero at representable nonzero integer x. Other tiny nonzero results retain significance.
  • Numeric references have relative1e-10, with J/Y near zeros absolute2e-14,4 minimum subnormals otherwise. No global error proof for all real inputs.
  • Browser interaction computes locally; parameterized share pages, REST and MCP send inputs to server. Logs/retention/region are deployment concerns; no zero-log promise.

Sources

Standards and references supporting this tool specification. Methods are listed separately — they are algorithms, not bibliographic sources.

Verification disclosure

Aggregate outcomes, scope and artifact commitments are public. Complete test inputs, expected values, execution records and oracle infrastructure are retained privately. This summary does not independently reproduce the full verification run.