Squares, three fleas, sporadic integer sets, and squares

Giedrius Alkauskas · 2025

Read Squares, three fleas, sporadic integer sets, and squares by Giedrius Alkauskas (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2508-05347.txt.

Introduction

Giedrius Alkauskas published this paper in 2025. It studies a jumping game with three fleas on a plane. At each move, two fleas leap to complete a square built on the segment between them. Starting from three set points, the question is which areas the flea triangle can ever span. Every positive half-integer area turns up. Integer areas hit every remainder class for every modulus.

Yet one family is missing entirely: no perfect square area ever occurs. The author explains the ban with quadratic reciprocity, a classic number theory law. Digging deeper, he ties the flea moves to integral Apollonian circle packings, a fractal circle geometry. A computer search up to a trillion found only five other missing non-square areas. The paper blends play, proof, and computation in a small, neat package.

Catalog record

CREATORGiedrius Alkauskas
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE23 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2508-05347

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Squares, three fleas, sporadic integer sets, and squares Giedrius Alkauskas Abstract . In the plane, three integer points (“fleas”) are given. At every tick of time, two of them (say, P , Q P,Q ) instantly jump to two vacant points R , S R,S , so that P ​ Q ​ R ​ S PQRS is a square with that order of vertices. Description of all integers and half-integers which occur as areas of spanned triangles turns out to be unexpectedly intricate problem. For example, if one starts from a triple ( 0 , 0 ) (0,0) , ( 2 , 0 ) (2,0) , ( 4 , 1 ) (4,1) , these positive integers are missed: 2 ⋅ { 0 , 1 , 4 , 15 , 16 , 20 , 79 , 84 , 95 , 119 , 156 } 2\cdot\{{0,1,4,15,16,20,79,84,95,119,156}\} , with no other up to 3 ⋅ 10 6 3\cdot 10^{6} (and seemingly, none at all); half-integers which are missed seem to form a 39 39 -element set (the largest of them being 11365 / 2 11365/2 ). However, there exist certain starting setups which have an “integrable” component as part of the answer. We demonstrate that for the initial triple ( 0 , 0 ) (0,0) , ( 2 , 1 ) (2,1) , ( 3 , 2 ) (3,2) , these integers are missed as areas: { N 2 : N ∈ ℕ 0 } \{N^{2}:N\in\mathbb{N}_{0}\} , along the sporadic set { 5 , 29 , 80 , 99 , 179 } \{{5,29,80,99,179}\} , with no other elements ≤ 3 ⋅ 10 6 \leq 3\cdot 10^{6} (most likely, no other at all; all half-integers serve as areas). Based on the evidence collected, we raise a corresponding conjecture. 1. Formulation Can a junior high school-like mathematical problem hit a dark side of mathematics: undecidability and/or incomputability? [ 3 ] . Sure, one can explain a notion of a tiling , or even the terms Heesch number , isohedral number for a connected monotile (see [ 5 ] ) to a pupil. Related problems are (or seem to be) on the border of “the bright side” of mathematics. Truly, Turing machine or Post correspondence system themselves can be presented in a disguise of a school-level game. Here we present a problem which might be a candidate. It occurred to an author in March 2024 [ 1 ] : Problem . In the plane, three points with integer coordinates are given, spanning a triangle \mathbf{K} . At every second, two of them (say, P and Q ) relocate to two points R and S , so that PQRS is a square with that order of vertices. Describe all integers – call this set \mathscr{I}(\mathbf{K}) – and half integers – call this set \mathscr{H}(\mathbf{K}) – which never occur as areas of spanned triangles. In this paper we contrive a method which allows to prove rigorously that if a certain value is missed in a finite number of steps, it is missed for good. A surprising fact is an “integrable” component of this set. Namely, there is a number-theoretic obstruction for an area value to be square of an integer. We will show when this occurs. Definition 1 . If three vertices of triangle \mathbf{K} do not belong any proper square sublattice of \mathbb{Z}^{2} , we call such a triangle primitive . Concerning the structure of sets of missed values, we pose the following Conjecture . For primitive triangles, the sets \mathscr{I}(\mathbf{K})\setminus\{n^{2}:n\in\mathbb{N}\} and \mathscr{H}(\mathbf{K}) are finite. Problem . If the conjecture is true, let f(N) be the maximal missed non-perfect square or half integer over all primitive triangles with side lengths \leq N . Is the function N\mapsto f(N) computable? We know that the function which tells the maximal number of steps needed for the Turing machine to stop over all Turing machine which eventually stop and which need N bits of information to be described, is incomputable (that is, eventually grow faster than any function one can actually write down). In the course of unravelling the problem, we will discover relations to Furstenberg topology [ 2 ] , covering systems [ 6 ] , matrix groups with torsion and freely generated semigroups [ 4 ] . We start from the basics. 2. The setup and basics 2.1. Notations Let \mathbf{P}_{i}=(x_{i},y_{i}) , i\in[1,2,3] , be three given points. Let \Delta be the area of the spanned triangle, and s=2\Delta\in\mathbb{Z}\cup\{0\} . The area will be positive or negative depending on whether points \mathbf{P}_{1},\mathbf{P}_{2},\mathbf{P}_{3} go round a triangle in a positive or negative direction. If all initial points belong to a proper square sublattice of \mathbb{Z}^{2} of index I>1 , then all subsequent points also do. Consequently, the set of possible areas is obtained from a set \Sigma , only each term multiplied by I . Here \Sigma is a set of all spanned areas, where our sublattice is interpreted as a new unit lattice. Therefore, without loss of generality, the initial triangle is required to be primitive. Define A,B,C to be the squares of the corresponding side-lengths: \displaystyle A=(x_{2}-x_{3})^{2}+(y_{2}-y_{3})^{2},\quad B=(x_{3}-x_{1})^{2}+(y_{3}-y_{1})^{2},\quad C=(x_{1}-x_{2})^{2}+(y_{1}-y_{2})^{2}. These expressions imply that A+B+C is even. One number out of these is even, the other two are odd (otherwise all points belonged to a index 2 square sublattice), and the odd ones \equiv 1\text{ (mod }4) . The following numbers are therefore integers: \displaystyle\frac{B+C-A}{2}=a,\quad\frac{C+A-B}{2}=b,\quad\frac{A+B-C}{2}=c. Figure 1. Possible position of fleas after 3 moves (and a possible fourth jump): \Delta=\frac{31}{2},s=31,A=17,B=65,C=58,a=53,b=5,c=12 . In what is to follow, we will need the following result. Lemma 2 . Let p|A , where p\equiv 3\,\mathrm{(mod}\,4) is a prime. Then, if p^{s}||A , s is even. This is classics, since A is a sum of two perfect squares. 2.2. Relations We will prove the formula \displaystyle s^{2}=ab+ac+bc. (1) In essence, this is just algebraically transformed Heron’s formula. Indeed, if \alpha is the angle of the triangle at the vertex \mathbf{P}_{1} , the cosine theorem gives \displaystyle A=B+C-2\sqrt{BC}\cos\alpha\Rightarrow a^{2}=\frac{(B+C-A)^{2}}{4}=BC\cos^{2}\alpha=BC-BC\sin^{2}\alpha=BC-s^{2}. Since B=a+c , C=a+b , this implies the needed identity. For future reference, \displaystyle a+b+c=\frac{A+B+C}{2},\quad s^{2}=(a+c)(b+c)-c^{2}=AB-c^{2}. (2) 2.3. Transformation We will now find a transformation rule for the quadruple (a,b,c;s) after the jump (\mathbf{P}_{2},\mathbf{P}_{3})\mapsto(\mathbf{P}_{2}^{\prime},\mathbf{P}_{3}^{\prime}) . There exist two squares with a given side \mathbf{P}_{2}\mathbf{P}_{3} . Depending on the choice, the new quadruple is given by \displaystyle(a^{\prime},b^{\prime},c^{\prime};s^{\prime})=\Big(a+(b+c)\pm 2s,b,c;\pm(b+c)+s\Big). (3) To prove this, note that b^{\prime}=b and c^{\prime}=c . Indeed, drawing a hypotenuse from \mathbf{P}_{1} to the line \mathbf{P}_{2}^{\prime}\mathbf{P}_{3}^{\prime} implies |\mathbf{P}_{1}\mathbf{P}_{3}|^{2}-|\mathbf{P}_{1}\mathbf{P}_{2}|^{2}=|\mathbf{P}_{1}\mathbf{P}_{3}^{\prime}|^{2}-|\mathbf{P}_{1}\mathbf{P}_{2}^{\prime}|^{2} (see Figure 1 ). Second, the area of the new triangle increases (or decreases) by half of the area of the square \mathbf{P}_{2}\mathbf{P}_{3}\mathbf{P}_{3}^{\prime}\mathbf{P}_{2}^{\prime} . Consequently, the equality s^{\prime}=s\pm A=s\pm(b+c) also holds. Lastly, ( 1 ) implies \displaystyle a^{\prime}=\frac{s^{\prime 2}-bc}{b+c}=\frac{s^{2}-bc}{b+c}\pm 2s+(b+c)=a\pm 2s+(b+c). Note the following consequence: the parity of a+b+c+s is an invariant. Applying ( 3 ) (with the “+” sign) t\in\mathbb{N} times, gives a general transformation (a,b,c;s)\mapsto(a^{\prime},b^{\prime},c^{\prime};s^{\prime}) . Explicitly, \displaystyle{\,\Big(a,b,c;s\Big)\mapsto\Big(a+(b+c)t^{2}+2st,b,c;(b+c)t+s\Big)\,} (4) A choice of the sign ``-" gives the same result, only with t\in\-\mathbb{N} . Therefore, in ( 4 ) we may assume t\in\mathbb{Z} . This will be our main formula. No need to write another two: one can just permute values a,b,c and apply ( 4 ) with an arbitrary integer t . 2.4. Two

Read the complete text