Wiki
Wiki

Research notes on every problem, and a library of the papers behind them. Built from the open erdos repository.

Updated

Ball-arithmetic bounds for the cubic family lengths

../

verify/: A distinct PASS grade warrants tier 1 for the strict cubic family bound, exact gamma value, every-cubic inequality and equality classification, at the recorded source and arithmetic boundaries.


From the repository root run:

sh
uv run --no-sync python wiki/theory/polynomials/L144_cubic_lemniscate_length/evidence/main.py

The program uses the repository's tools.Checker and python-flint arb and acb balls at 64 bits. It takes about five minutes on one thread and reads no input files. It implements Sections 5 to 7 of the proof: thirteen parameter boxes of width 1/641/64 on which the scaled integral I(a)I(a) is negative, an adaptive cover of 7/512≤arcsin⁡(2a3)≤π/27/512\le\arcsin(2a^3)\le\pi/2 by parameter boxes on which the length is below L0−1/100L_0-1/100, the check that the two regimes overlap, and the check that the boxes of the second regime tile its range. Every bound is a sum of exact piece widths times ends of balls, so it holds uniformly on its box; tolerances affect only how many pieces are used. A control at arcsin⁡(2a3)=1/2\arcsin(2a^3)=1/2 computes upper and lower bounds and checks that they differ by less than 0.050.05. The upper integral ends at a dyadic enclosure of π\pi from above; the lower integral ends at one from below and omits the singular window. Each window is checked to lie inside [0,π][0,\pi] and to contain the critical angle for the whole parameter box.

The arithmetic uses the enclosing endpoints and unions documented for python-flint arb, with root enclosures from python-flint acb. The proof explains why choosing either quadratic root or any cube root preserves the sum. Integration uses real interval bounds, without assuming that a complex integrand is analytic across a branch cut. A finite enclosure at the depth limit is still a valid, possibly coarse bound; a nonfinite one fails. No tolerance is accepted as an error estimate.

The program does not check the written reductions: the length integral, the root identities, the comparison with z3−cz^3-c, the tail constant's derivation and the reduction of all cubics to the family are proved on the proof page. Every failed obligation, including a nonfinite enclosure in the first regime, causes a nonzero exit. This is author evidence, not an independent examination.