← All papers
First page of Halving the original Kalton–Roberts upper bound for nearly additive set functions

Halving the original Kalton–Roberts upper bound for nearly additive set functions

Boon Suan Ho, Tomasz Kania

math.CO Jun 5, 2026 · v2
The improved upper bound on the Kalton–Roberts constant is formalized in about 6.2k lines of Lean 4 with Mathlib, with AI assistance.
Let $K_{KR}$ be the least constant governing uniform approximation of approximately additive real-valued set functions by finitely additive signed measures. We prove that $$K_{KR}\le\frac{374167}{20000}=18.70835.$$ Starting from a norming functional for best uniform approximation, we form two families of almost extremal sets with complementary point frequencies. Intersections and biregular expanders then give two recombination schemes. The sharper estimate comes from using the schemes on different ranges and from cancelling the positive and negative errors with unequal weights. This improves the previously published bound $38.8$.

The Kalton–Roberts constant K_KR is the least constant such that every approximately additive set function is uniformly close to a finitely additive signed measure. The best published upper bound was 38.8.

A Hahn–Banach dual certificate for best ℓ∞-approximation yields two families of near-extremal sets with complementary point frequencies. Rational mixtures of neighbouring intersection levels lower these frequencies, and Pippenger-style random biregular expanders drive two recombination schemes, used on different case ranges. The argument is formalized in Lean 4 with Mathlib, where K_KR is defined as an infimum and the exact rational estimate is proved; a Python script performs independent arithmetic checks.

The proof gives K_KR ≤ 694198146664396294486127753/34994834677886019996000000 ≈ 19.837, hence K_KR < 19.838, more than halving the original bound of 89/2. The abstract states a sharper bound of 374167/20000 = 18.70835, which differs from the constant derived in the body and formalized in Lean.

SourceUpper bound
Kalton–Roberts89/2 = 44.5
Bondarenko–Prymak–Radchenko38.8
Present paper, body and Lean formalization≈ 19.837 (< 9919/500)
Present paper, abstract374167/20000 = 18.70835
Upper bounds on K_KR