A quadratic bottleneck bounds Hausdorff measure
ProvedErdos1041.Counterexample.s3_bottleneck_hausdorffdegree-seven-counterexampleerdos-1041polynomial-lemniscate
Let p be a complex polynomial and cc a nonzero-value critical point in its strict unit lemniscate. Suppose b₁ and b₂ are distinct roots in cc’s component, are the only roots in that component, and cc is its only critical point. Assume a nonzero quadratic coefficient aHat, a positive disk radius h, the stated uniform 1/4 quadratic approximation on that disk, and δ=1−|p(cc)| with δ>0 and δ<|aHat|h²/4. Then every preconnected subset K of the component containing b₁ and b₂ has one-dimensional Hausdorff measure at least the nonnegative-real embedding of |b₁−cc|+|b₂−cc|−(8/3)√(δ/|aHat|).
Preamble
import Definitions.Def_ErdosProblems_Erdos1041_Counterexample_Defs import Definitions.Def_ErdosProblems_Erdos1041_Counterexample_BarrierAlgebra import Definitions.Def_ErdosProblems_Erdos1041_Counterexample_BarrierGraphs import Definitions.Def_ErdosProblems_Erdos1041_Counterexample_BarrierSigns import Definitions.Def_ErdosProblems_Erdos1041_Counterexample_Bottleneck import Definitions.Def_ErdosProblems_Erdos1041_Counterexample_InstanceBarriers import Definitions.Def_ErdosProblems_Erdos1041_Counterexample_InstanceCritical import Definitions.Def_ErdosProblems_Erdos1041_Counterexample_InstanceConnectivity import Definitions.Def_ErdosProblems_Erdos1041_Counterexample_HausdorffLength import Mathlib import Mathlib.Algebra.Polynomial.Derivative import Mathlib.Algebra.Polynomial.Div import Mathlib.Algebra.Polynomial.Roots import Mathlib.Analysis.Calculus.Deriv.Polynomial import Mathlib.Analysis.Real.Pi.Bounds import Mathlib.Analysis.SpecialFunctions.Complex.Log import Mathlib.Analysis.SpecialFunctions.Pow.Real import Mathlib.Analysis.SpecialFunctions.Trigonometric.Basic import Mathlib.Tactic import Mathlib.Tactic.ComputeDegree import Mathlib.Tactic.NormNum import Mathlib.Tactic.Ring import Mathlib.Topology.Connected.LocallyConnected import Mathlib.Topology.Connected.PathConnected import Mathlib.Topology.EMetricSpace.BoundedVariation import Mathlib.Topology.MetricSpace.Contracting import Mathlib.Topology.Order.IntermediateValue /-! External source: ani, erdosproblems.com forum thread 1041, 7 Sept 2026. -/ /-! # Erdős #1041 with length as one-dimensional Hausdorff measure Formal Conjectures states Erdős #1041 with the length of a path defined as the one-dimensional Hausdorff measure `μH[1]` of its image. `erdos1041_counterexample` bounds the total variation of a parametrisation instead. This module proves the Hausdorff form for the same polynomial `f`, and in a stronger shape: every preconnected subset of the strict lemniscate `Ω(f)` that contains two distinct roots has one-dimensional Hausdorff measure greater than two (`erdos1041_counterexample_hausdorff`). The image of any path joining two roots is such a set. The argument reuses the bottleneck geometry of `Bottleneck.lean` and adds no arc-extraction, rectifiability or length-of-arc lemma. Removing the slit preimage from the component of `Ω(f)` through the critical point leaves an open set in which no preconnected subset contains both roots, because the polynomial restricted there is a covering of a simply connected base. So a preconnected set `K` through both roots must, for every radius between the slit preimage and a root, meet the circle of that radius about the critical point inside the connected component of that root (`bottleneck_sheet_crossing`). The two components are disjoint open sets, the distance to the critical point is 1-Lipschitz, and on the real line `μH[1]` is Lebesgue measure, so `μH[1] K` is at least the sum of the two radial lengths (`s3_bottleneck_hausdorff`). That is the same bound `s3_bottleneck_length` gives for total variation, so the numerical margin of `Assembly.lean` applies unchanged. `erdos1041_hausdorff_negation` and `erdos1041_hausdorff_answer_false` state the Formal Conjectures parent `Erdos1041.erdos_1041` in its own vocabulary, with `fcLength` its `length`, and refute it. The mathematics of the counterexample is ani's. The polynomial is the single member `s = 10⁻⁶` of ani's family fixed in `Defs.lean`. -/ noncomputable section open scoped ENNReal open MeasureTheory Polynomial Metric /-! ## The slit domain separates the two roots -/ /-! ## Crossing every circle inside each sheet -/ /-! ## From radial crossings to Hausdorff measure -/ open Erdos1041.Counterexample
Formal statement
theorem Erdos1041.Counterexample.s3_bottleneck_hausdorff
(p : Polynomial ℂ) (cc : ℂ) (hcc : cc ∈ Omega p)
(hcrit : (Polynomial.derivative p).IsRoot cc)
(hv : p.eval cc ≠ 0)
(b₁ b₂ : ℂ) (hne : b₁ ≠ b₂)
(hb₁ : b₁ ∈ connectedComponentIn (Omega p) cc)
(hb₂ : b₂ ∈ connectedComponentIn (Omega p) cc)
(hr₁ : p.IsRoot b₁) (hr₂ : p.IsRoot b₂)
(hzeros : ∀ w ∈ connectedComponentIn (Omega p) cc, p.IsRoot w → w = b₁ ∨ w = b₂)
(huniq : ∀ c' ∈ connectedComponentIn (Omega p) cc,
(Polynomial.derivative p).IsRoot c' → c' = cc)
(aHat : ℂ) (haHat : aHat ≠ 0) (h : ℝ) (hh : 0 < h)
(hdisk : ∀ z : ℂ, ‖z‖ ≤ h → ‖(shiftQuad p cc).eval z / aHat - 1‖ ≤ 1 / 4)
(δ : ℝ) (hδ : δ = 1 - ‖p.eval cc‖) (hδpos : 0 < δ)
(hδsmall : δ < ‖aHat‖ * h ^ 2 / 4)
(K : Set ℂ) (hK : IsPreconnected K)
(hKsub : K ⊆ connectedComponentIn (Omega p) cc)
(hK₁ : b₁ ∈ K) (hK₂ : b₂ ∈ K) :
ENNReal.ofReal (‖b₁ - cc‖ + ‖b₂ - cc‖ - 8 / 3 * Real.sqrt (δ / ‖aHat‖))
≤ μH[1] K := by sorry
Source
Lean source: https://github.com/wcook04/plectis-erdos-lean/blob/cc7e541cf2081c6fef5a5e377d52e365e33b01eb/ErdosProblems/Erdos1041/Counterexample/HausdorffLength.lean#L183-L277
Construction by ani: https://www.erdosproblems.com/forum/thread/1041#post-8861
Related paper and provenance: https://github.com/wcook04/plectis-erdos/blob/551bae6dc6e732cf85172d66323c8d2bc77ba962/paper/1041/erdos-1041-lemniscate-newton-flow.tex#L25-L99
Paper prior-art bibliography: https://github.com/wcook04/plectis-erdos/blob/551bae6dc6e732cf85172d66323c8d2bc77ba962/paper/1041/erdos-1041-lemniscate-newton-flow.tex#L1662-L1772
AI-assisted formalization in Will Cook's project; ani is credited for the degree-seven construction. Independent correspondence of the 1958 Problem 5 wording to this modern formulation is unrecorded.