Skip to content

Erdős Problem #1129

Editorial depth: Authored explanatory DRAFT — NOT editorially accepted.
Snapshot classification: proved (not automatically the live result).
Subject areas: analysis, polynomials.

Read this status correctly

The status above is reproduced from GCL's immutable external-catalogue snapshot. It is not a new claim of current mathematical resolution, a source-semantic review, an independent proof check, or GCL certification. A published proof and a GCL replay are separate events.

UNACCEPTED DRAFT — mathematical editorial review pending

This material is a provisional, source-qualified manuscript published for line-by-line scrutiny and real HTML/LaTeX testing. It has NOT received separate GCL editorial acceptance or independent mathematical certification. Its presence on this website is not evidence of a solution, reviewed coverage, or final publication.

Review the underlying draft on GitHub.

The question in plain language

When we interpolate values at \(n\) chosen points with a polynomial, errors in the data can be amplified. How should we choose the interpolation points to keep the worst possible amplification as small as possible?

Erdős Problem #1129 asks for the points minimizing the Lebesgue constant. With distinct ordered nodes \(x_1<\cdots<x_n\) in \([-1,1]\), define the Lagrange basis

\[ \ell_k(x)=\prod_{i\ne k}\frac{x-x_i}{x_k-x_i} \]

and then

\[ L(x)=\sum_{k=1}^n|\ell_k(x)|,\qquad \Lambda=\max_{x\in[-1,1]}L(x). \]

This is an extremal optimization problem, not merely the choice of some convenient interpolation nodes.

A small example

For two nodes \(-1,1\), the Lagrange polynomials are \((1-x)/2\) and \((1+x)/2\). Both are nonnegative on the interval, so \(L(x)=1\) and

\[ \Lambda(-1,1)=1. \]

For the canonical three-node choice \(-1,0,1\),

\[ \ell_{-1}(x)=\frac{x(x-1)}2,\quad \ell_0(x)=1-x^2,\quad \ell_1(x)=\frac{x(x+1)}2. \]

The maximum of the sum of their absolute values is exactly \(5/4\), attained at \(x=\pm1/2\):

\[ \Lambda(-1,0,1)=\frac54. \]

These examples are exact, but they do not independently establish optimality or uniqueness over all possible three-node configurations.

What is known

Faber, Bernstein and Erdős established increasingly sharp lower bounds, culminating in the optimal leading coefficient

\[ \Lambda\ge\frac2\pi\log n-O(1). \]

Chebyshev-root interpolation nodes achieve the same coefficient up to an additive constant.

Erdős suggested that optimal nodes should have the same peak \(L(x)\) value in each interval between successive nodes. Kilgore and Cheney (1976) established existence of a configuration with this property, Kilgore (1977) established its necessity for optimality, and de Boor and Pinkus (1978) established a uniqueness result. This resolved the main structural characterization.

The canonical problem additionally fixes the endpoints \(x_1=-1\) and \(x_n=1\). Known explicit canonical optimizers include \(-1,1\) for \(n=2\), \(-1,0,1\) for \(n=3\), and an algebraic symmetric four-node configuration described by Rack and Vajda (2015).

GCL's current state

Authenticated source audit #2400 verifies the \(n=2\) and \(n=3\) Lebesgue constants and separates canonical and unrestricted formulations.

The older variational existence, necessity and uniqueness proofs have not been independently reconstructed.

What remains

Review Kilgore's and de Boor–Pinkus's original theorem statements, the precise equioscillation definition, and endpoint conventions, and independently verify the known algebraic \(n=4\) canonical optimizer.

Sources and claim boundaries

Erdős #1129; catalogue history; Kilgore–Cheney (1976), Kilgore (1977), de Boor–Pinkus (1978), Rack–Vajda (2015); GCL return. Externally proved, finite formulas independently verified. Unaccepted draft.

Catalogue and evidence provenance

  • Pinned catalogue source: teorth/erdosproblems at 536408716724.
  • GCL issue: #2400 — protected issue/assignment identity; its labels and results are live, not reproduced automatically here.
  • GCL triage: OUTSIDE_STRICT_OPEN_TRIAGE.
  • Formal-statement coverage: No statement was confirmed in GCL's pinned formal-provider snapshot. Presence does not mean the statement or proof has been checked.
  • External mathematical status: consult current problem page and its dated bibliography.

Back to catalogue · Compendium introduction