Skip to main content

omnibias.sos

0.1.0a2AlphaPython >=3.10AGPL / commercial

Sum-of-squares constructions, polynomial positivity checks, and rigorous validation of proposed Gram matrices.

import omnibias.sos

Use sos for supported polynomial positivity or moment-based certificate constructions. The monomial basis, coefficient ordering, degree, and domain constraints determine the represented problem. A floating-point optimization routine proposes a candidate; that candidate is not itself a rigorous certificate.

The validation path checks the required matrix and polynomial identities under its stated arithmetic assumptions. Positive definiteness is stronger than semidefiniteness, and near-singular candidates can fail a strict enclosure check even when a floating-point solver reports success. Keep solver output and verified status separate.

A certificate applies to the polynomial and domain encoded in its obligations. It does not certify arbitrary sampled functions or establish an unrestricted analytic theorem. Preserve the complete evidence artifact and inspect optional formal-check availability. This alpha package uses AGPL-or-commercial licensing.

Module contract & mathematical conventions

omnibias-sos: certified universal positivity by optimization.

An SOS decomposition p = z(x)^T Q z(x) with positive-semidefinite Q certifies p(x) >= 0 for every x. Some nonnegative polynomials are not sums of squares. A floating-point semidefinite program *proposes* the Gram matrix Q; the *proof* is a rigorous, outward-rounded interval LDL^T positive-definiteness certificate from omnibias.core.verified -- the same finite obligation the Mathlib-free Lean kernel re-checks, so a sealed certificate can earn theorem_prover_verified.

The optimizer only proposes; the interval algebra and the Lean kernel prove. Everything is *sound by construction*: a failed rounding or positive-definite margin yields an **inconclusive** verdict, never a false positivity claim.

Shared formulas do not imply universal floating-point identity. Verify dtype, shape, derivative-order support, and the stated numerical assumptions for the path you use.

Imports & aliases​

These names share a definition with another public module. Follow a name for its full contract, parameters, and examples.

Import from this moduleCanonical reference
DEFAULT_DENOMINATORSomnibias.sos.DEFAULT_DENOMINATORS
Exponentomnibias.sos.Exponent
FINITE_DIM_SYSTEMomnibias.sos.FINITE_DIM_SYSTEM
GALERKIN_TRUNCATIONomnibias.sos.GALERKIN_TRUNCATION
GLOBAL_POLYNOMIALomnibias.sos.GLOBAL_POLYNOMIAL
MonomialBasisomnibias.sos.MonomialBasis
Polynomialomnibias.sos.Polynomial
PolynomialInequalityBackendomnibias.sos.PolynomialInequalityBackend
PositivstellensatzCertificateomnibias.sos.PositivstellensatzCertificate
RationalPolynomialomnibias.sos.RationalPolynomial
SOSCertificateomnibias.sos.SOSCertificate
SOSMultiplieromnibias.sos.SOSMultiplier
SOSProblemomnibias.sos.SOSProblem
SOSScopeomnibias.sos.SOSScope
arrangement_adapted_basisomnibias.sos.arrangement_adapted_basis
certify_nonneg_on_setomnibias.sos.certify_nonneg_on_set
certify_sosomnibias.sos.certify_sos
certify_sos_rationalomnibias.sos.certify_sos_rational
degree_reduction_reportomnibias.sos.degree_reduction_report
gram_productsomnibias.sos.gram_products
gram_to_polyomnibias.sos.gram_to_poly
honesty_labelsomnibias.sos.honesty_labels
is_nonneg_on_setomnibias.sos.is_nonneg_on_set
is_sosomnibias.sos.is_sos
is_theorem_prover_verifiedomnibias.sos.is_theorem_prover_verified
lean_availableomnibias.sos.lean_available
lean_check_sosomnibias.sos.lean_check_sos
monomial_basisomnibias.sos.monomial_basis
named_adapted_problemsomnibias.sos.named_adapted_problems
rational_gramomnibias.sos.rational_gram
seal_positivstellensatz_certificateomnibias.sos.seal_positivstellensatz_certificate
seal_sos_certificateomnibias.sos.seal_sos_certificate