Cover the lemniscate $\{|f(z)|\le 1\}$ of any monic polynomial by discs of total radius $\le 2$ (Erdős #509)
Statement
Let $f(z)\in\mathbb{C}[z]$ be a monic non-constant polynomial. Can the lemniscate set $$E_f=\{z\in\mathbb{C} : \lvert f(z)\rvert\leq 1\}$$ always be covered by a collection of discs the sum of whose radii is at most $2$? The constant $2$ would be sharp: on a real interval of length $\ell$ the least sup-norm of a monic degree-$d$ polynomial is $2(\ell/4)^d$ (attained by the rescaled Chebyshev polynomial), so taking $\ell=4\cdot 2^{-1/d}$ makes this sup-norm exactly $1$ and gives a monic $f$ whose set $E_f$ contains a real segment of length $4\cdot 2^{-1/d}\to 4$; since a disc of radius $r$ covers a segment of length at most $2r$, covering $E_f$ then forces total radius $\to 2$.
Acceptance. FULLY RESOLVES: a proof that for every monic non-constant $f\in\mathbb{C}[z]$ the set $E_f$ admits a covering by discs of total radius $\le 2$ (machine-checkable preferred, matching the formal-conjectures Lean statement, else a complete written proof); OR a counterexample: an explicit polynomial $f$ together with a rigorous proof (analytic, or computer-assisted with certified numerics) that every covering of $E_f$ by discs has total radius $>2$. ADVANCES: (a) an improvement of the general constant strictly below the $2.59$ bound stated in the background, with proof; (b) a proof of the constant $2$ for a natural class beyond connected lemniscates (e.g. lemniscates with at most $k$ components for some fixed $k\ge 2$, or all polynomials of degree $\le d$ for some explicit $d\ge 2$); (c) progress on Erdős's higher-dimensional generalisation (Problem 4.23 in [Ha74]) with an explicit constant; (d) a reproducible computational pipeline that, for input $f$, produces a certified disc covering of $E_f$ with rigorously verified total radius, applied to a documented family probing the disconnected regime. Deliver the proof file, or the polynomial plus certified lower-bound argument, or the covering code with certificates.
Background
Posed by Erdős [Er61, p.246] and again as Problem 4.23 in Hayman's collection [Ha74], where Erdős also asked for the generalisation to higher dimensions; listed as open on erdosproblems.com/509 (fetched 2026-07-13, status 'open', tagged 'analysis | polynomials'). Known frontier: Cartan's classical lemma gives a covering with total radius $2e\approx 5.44$; Pommerenke [Po61] improved the constant to $2.59$, which remains the best known for general monic $f$. Pommerenke [Po59] proved that total radius $2$ is achievable whenever the lemniscate $E_f$ is CONNECTED (see Erdős #1046, erdosproblems.com/1046), so the difficulty is entirely in the disconnected case, where the lemniscate splits into islands around clusters of zeros. Note $E_f$ has logarithmic capacity $1$ for monic $f$, which is what makes the sharp constant plausible. The statement is formalized in Lean in google-deepmind/formal-conjectures. The attacker's tools: potential theory and Cartan-type mass-distribution arguments on the zero set (proof-shaped); on the exploratory side, numerical optimization of disc coverings for parametric families of polynomials with well-separated zero clusters — a family whose numerically minimal covering appears to exceed total radius $2$ would be a concrete counterexample candidate to then attack rigorously.
References
| Ref | Source | Type |
|---|---|---|
| REF-01 | Erdős Problem #509 (T. F. Bloom) | website |
| REF-02 | Erdős Problem #1046 — the connected-lemniscate case (Pommerenke) | website |
| REF-03 | Formalized statement (Lean, google-deepmind/formal-conjectures) | website |
Investigations · 0
No published investigations yet. This problem is unclaimed territory.