Alec Kriebel

A telescoping proof of the focal antipedal sum invariant

An elliptic billiard orbit gives two antipedal polygons, one about each focus. Reznik, Garcia and Koiller observed that the sums of the distances from each focus to its antipedal vertices agree. This note proves that equality by turning the difference on each edge into a vertical displacement. Around a closed orbit, those displacements cancel.

The distances are ordinary positive Euclidean norms. The proof applies to every regular closed orbit with a strictly nested confocal elliptic caustic, including admitted odd primitive periods, star orbits and repeated traversals. It does not assert that either focal sum is separately constant as the orbit varies.

Read the paper

Open the original PDF · Download the PDF · Zenodo record and DOI · BibTeX

The unchanged manuscript and native record identify version 1.0, dated 6 October 2026.

The exact result

Take an ellipse E and a strictly nested confocal ellipse Eλ:

E: x²/a² + y²/b² = 1,
Eλ: x²/(a² − λ) + y²/(b² − λ) = 1,
a > b > 0,   0 < λ < b².

Write c = √(a² − b²), with foci f± = (±c, 0). A regular closed orbit is a cyclic list P1, …, PN on E with distinct consecutive endpoints and every edge tangent to the same Eλ. At each vertex it continues along the other caustic tangent, so immediate retracing is excluded. The vertices need not form a convex polygon or follow their order around the ellipse.

For each focus fσ and edge PiPi+1, intersect the full lines through the endpoints perpendicular to their respective focus-to-endpoint vectors. Call this antipedal vertex Qσ,i, and put qσ,i = ‖Qσ,i − fσ‖. Every such vertex exists uniquely, and every q is finite and strictly positive. The theorem is

∑i=1N q+,i = ∑i=1N q−,i,   hence   (∑ q+,i)/(∑ q−,i) = 1.

The assertion covers every admitted period and winding, including odd primitive periods, self-intersecting stars, reversal and repetitions of a shorter orbit. It is conditional on a regular closed orbit existing; it does not say every integer period occurs for every caustic.

Why the difference telescopes

For a chord AB and a focus f off its line, the antipedal intersection satisfies

‖Q − f‖ = ‖A − f‖ ‖B − f‖ / h,
h = dist(f, line(AB)) > 0.

This follows directly from the two perpendicular-line equations and the triangle-area formula, taking the positive square root. The other-tangent rule lets the whole orbit be oriented with the caustic on the left, even when the orbit is a star.

For an outward unit normal (u,v), put K = a²u² + b²v² and ρ² = K − λ, with ρ > 0. The two focal heights are h± = ρ ∓ cu. Their product is b² − λ > 0 and their sum is 2ρ > 0, so both are actual positive distances. In particular no admitted chord passes through a focus. The chord formulas give

qσ = (a²hσ + b²λ/hσ)/K,
q+ − q− = Γ(yB − yA),
Γ = c/(ab√λ) · (−a² + b²λ/(b² − λ)).

The same Γ applies to every edge in this consistent orientation. Summing the vertical increments gives yN+1 − y1 = 0 by closure. No even-period pairing or central symmetry is needed.

No division by u, vertical displacement or Γ occurs. Horizontal chords and the zero-coefficient caustic λ = a²b²/(a² + b²) are included; neither makes the ordinary norms vanish. Reversing the whole orbit preserves the antipedal vertices and their norms. Its displacement coefficient changes to −Γ in the reverse orientation.

What the invariant concerns

The equality concerns antipedal vertices of the original orbit polygon. They are not pedal feet, vertices of the outer tangent polygon, inverse side lengths or side perimeters. The two positive sums have ratio one; neither individual sum is claimed constant across a Poncelet family.

The strict elliptic-caustic assumptions matter. Hyperbolic caustics, degenerate caustics, two-bounce limits and immediate-retracing lists lie outside the theorem. No extension to λ = 0, λ = b² or the circular case a = b is asserted here.

The source archive includes exact rational-chord checks of the defining line equations, positive norms and edge identity. These finite checks supplement the written argument; they do not establish its all-real-parameter or all-admitted-period conclusion by sampling.

The observation and the proof

Reznik, Garcia and Koiller introduced this observation as invariant k603. Their Table 7 records its experimental discovery in May 2020 and marks its proof unknown there. That dated marker is not a statement about the present literature. The observation and the antipedal construction are theirs.

Classical negative-pedal and polarity geometry, the known focal sideline-height product, centrally paired cases proved by central inversion, and ordinary telescoping are background. The contribution proposed in this note is the identity between the ordinary norm difference and a coordinate increment, giving the full-period proof, especially for odd primitive periods.

Focal inverse-perimeter invariants use different summands. Spatial-average formulas for other elliptic-billiard observables likewise differ from this finite antipedal norm identity. The historical comparison is bounded by the selected sources and versions: journal finals and precursors, and final books and author sources, must be distinguished; some historical and experimental material was not read. Neither the dated table nor the manuscript’s bounded search supplies an absolute-first, exhaustive-literature or continued-open-status guarantee.

  • D. Reznik, R. Garcia and J. Koiller, Eighty New Invariants of N-Periodics in the Elliptic Billiard, arXiv:2004.12497v11 (29 October 2020), §§3.5, 3.7 and Table 7. Author preprint.
  • D. Reznik, R. Garcia and J. Koiller, Fifty New Invariants of N-Periodics in the Elliptic Billiard, Arnold Mathematical Journal 7 (2021), 341–355, §3.7 and Table 7, p. 349. Journal article.
  • G. Salmon, A Treatise on the Higher Plane Curves, third edition (1879), Arts. 121–122, pp. 105–107: classical negative-pedal and polarity geometry.
  • M. Bialy and S. Tabachnikov, Dan Reznik’s identities and more, European Journal of Mathematics 8 (2022), 1341–1354, Lemma 4.2: the focal sideline-height product. Journal article.
  • P. Roitman, R. Garcia and D. Reznik, New Invariants of Poncelet–Jacobi Bicentric Polygons, Arnold Mathematical Journal 7 (2021), 619–637: different inverse-perimeter observables. Journal article.
  • R. Garcia, J. Koiller and D. Reznik, Estimating Elliptic Billiard Invariants with Spatial Integrals, Journal of Dynamical and Control Systems 29 (2023), 757–767, online 10 August 2022: spatial averages of other observables. Journal article.

Original abstract

Reznik, Garcia and Koiller observed that the sums of the distances from the two foci of an ellipse to their respective antipedal vertices agree along every closed elliptic billiard orbit. We prove this equality for every regular closed orbit tangent to a strictly nested confocal ellipse, including odd primitive periods, star orbits and repeated traversals. The difference of the two ordinary positive antipedal distances on each edge is a fixed coefficient times the vertical displacement of that edge. Closure therefore gives the equality without a parity restriction.

Original files and citation

The unchanged original files are available here and from Zenodo, in the native record’s order:

  1. focal_antipedal_sum.pdf (65,142 bytes) · Zenodo original PDF
  2. focal_antipedal_sum_support.zip (113,446 bytes) · Zenodo original ZIP

Kriebel, Alec. A telescoping proof of the focal antipedal sum invariant. Version 1.0, 6 October 2026. doi:10.5281/zenodo.23191247. Download BibTeX.

The manuscript and all original authored support files, including authored code, use CC BY 4.0. Preserve attribution and indicate adaptations. Cited third-party works retain their own copyright and license terms; their full bodies are not redistributed in the archive.

Status and collaboration

AI tools were used extensively in solving, deriving, drafting, literature checking, code generation, verification and adversarial checking. Alec Kriebel is the named human author and is responsible for the work. Independent checks include AI agents and are not conventional human peer review. This manuscript is unrefereed and has not received conventional independent human or journal peer review, journal acceptance or formal proof-assistant certification.

Independent verification, substantive mathematical extensions and work toward a journal submission are welcome. Collaboration and any authorship would reflect actual scholarly contributions.