Bicentric quadrilaterals, the arbelos, coaxal circles and inversion in Lean 4
We give machine-checked proofs, in Lean 4 with Mathlib, of a body of classical theorems onquadrilaterals and circle systems in the Euclidean plane. For quadrilaterals we prove Bretschneider’sformula and Brahmagupta’s formula for a convex quadrilateral, Pitot’s theorem and itsconverse, Euler’s quadrilateral theorem, Ptolemy’s second theorem (the ratio of the diagonals),van Aubel’s theorem, Newton’s theorem that the midpoints of the diagonals of a tangentialquadrilateral and its incentre are collinear, and Fuss’s relation 1/(R − d)2 + 1/(R + d)2 = 1/r2for a convex bicentric quadrilateral whose incircle is given by its distances to the four side lines.For the arbelos we prove Archimedes’ twin circles, the area theorem for Lebesgue measure, thePappus chain for every n with its uniqueness and its inversion picture, the existence and radiusof Bankoff’s third twin circle, and the Steiner porism for any two nested circles and anychain length, both for cyclic tangent sequences that may repeat circles and for chains of n ≥ 3distinct circles bounding pairwise disjoint discs. For circle systems we prove the radical centretheorem, the basic theory of coaxal pencils and orthogonal circles, Johnson’s theorem on threeequal circles through a point, the Simson–Wallace theorem as an equivalence together with anidentity relating the area of the pedal triangle to the power of the point, and Steiner’s linethrough the orthocentre. For inversion we prove the basic distance laws, Ptolemy’s inequalityand its boundary case through inversion, the Peaucellier–Lipkin linkage, the Descartes circletheorem, Soddy’s five-sphere relation in space and the integrality of Apollonian curvatures underApollonian curvature moves. All proofs use only Lean’s three standard axioms. Many ofthese results have been formalised before, and we reprove them independently: Brahmagupta’sformula (three earlier Lean proofs, one in our own Paper B), Euler’s quadrilateral theorem,van Aubel, the Simson–Wallace theorem with its converse, Steiner’s line, Peaucellier–Lipkin,the Descartes and Soddy relations, and the basic inversion facts, which are in Mathlib. Fuss’srelation was proved earlier, for possibly non-convex quadrilaterals and by another route, in ourown Casey–Fuss paper. The pedal-area identity also has an earlier unit-circle complex formulation;here it is stated for an arbitrary circumcircle using Mathlib projections. For the remainingstatements we found no earlier formal proof in the sources we searched, subject to stated gaps.
Authors
- Joshua Bald (ORCID: https://orcid.org/0009-0002-1317-6489)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-17
- DOI
- https://doi.org/10.5281/zenodo.24296363
- Primary Topic
- Mathematics and Applications
- Type
- preprint