Id: palomar/PALOMAR-2026-11-21-000005 Url: https://github.com/MColbrook/NLA-formal-verification-20-problems/tree/d46c1b54e85af6790d3ac0df59a9bbe8d2faaf4d Json: https://data.palomar-registry.org/entries/PALOMAR-2026-11-21-000005-v1.json Repo: https://github.com/MColbrook/NLA-formal-verification-20-problems Tags: math.NA, math.FA, math.PR, math.AG, cs.CC, msc:15A18, msc:15A23, msc:15A45, msc:15A60, msc:15A69 Score: 0 Title: Numerical Linear Algebra, Explained: Twenty Problem Resolutions Trust: high Domain: math.NA Source: palomar Authors: Matthew J. Colbrook, George Stepaniants Reviews: 0 Subject: math.NA Category: research Comments: 0 Disputed: false Division: math Mentions: 0 Verified: true Discussed: 0 Published: 2026-11-21 Formalized: true Formalizes: title=Numerical Linear Algebra, Explained: Twenty Problem Resolutions identifier=urn:sha256:b66976c1efa4f625f3b380fef6e945ddf4d7beea6e1b264e2f87d6790548bba5 relationship=formalizes Page: https://delphi.science/r/palomar/PALOMAR-2026-11-21-000005 Lean formalizations accompanying twenty numerical linear algebra problem resolutions. They give uniform matrix-product growth and local 1/d-Hölder continuity of the joint spectral radius; two real generators with any prescribed nonnegative polynomial growth exponent; and sharp losses 2^r and 4^r for randomly pivoted Cholesky and LU. They show that concave matrix functions can destroy spectral approximation guarantees and amplify nuclear-norm excess without bound, and give sharp Frobenius transfer bounds with and without semidefinite order. Further results are the random-row adversarial…