Covering a square by congruent squares

György Dósa, Zsolt Lángi, Zsolt Tuza · 2026

Read Covering a square by congruent squares by György Dósa, Zsolt Lángi, Zsolt Tuza (2026) 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-2601-16535.txt.

Introduction

György Dósa, Zsolt Lángi and Zsolt Tuza ask how big a square can be clad by unit squares. Their 2026 preprint sets a clean maths task. Take a count n of unit squares, each of edge one. They must clad a big square with no gaps left. The aim is the most large edge the clad square can have.

That most edge is named S of n in the work. Small n can be solved by hand with care. Big n needs firm bounds and build tricks. Clean tasks of this kind test pack and clad maths. The team seeks sharp values and bounds for S of n.

Catalog record

CREATORGyörgy Dósa, Zsolt Lángi, Zsolt Tuza
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE30 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2601-16535

Browse the full library · About the Tower of Babel

Full text

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

Covering a square by congruent squares Thanks: Partially supported by the National Research, Development and Innovation Office, NKFI, K-147544 grant, the ERC Advanced Grant “ERMiD”, and the Project no. TKP2021-NVA-09 with the support provided by the Ministry of Innovation and Technology of Hungary from the National Research, Development and Innovation Fund and financed under the TKP2021-NVA funding scheme. György Dósa, Zsolt Lángi, Zsolt Tuza Abstract . The main goal of this paper is to address the following problem: given a positive integer n n , find the largest value S ⁡ ( n ) S(n) such that a square of edge length S ⁡ ( n ) S(n) in the Euclidean plane can be covered by n n unit squares. We investigate also the variant in which the goal is to cover only the boundary of a square. We show that these two problems are equivalent for n ≤ 4 n\leq 4 , but not for n = 5 n=5 . For both problems, we also present the solutions for n = 5 n=5 . 2020 Mathematics Subject Classification 52C15, 52A40, 52A38 1. Introduction Our main goal is to investigate the following problem from [ 8 ] : given a positive integer n , find the largest value S(n) such that a square of edge length S(n) in the Euclidean plane can be covered by n unit squares. The case n=3 of this problem was proposed by Dudeney [ 5 ] in 1931, and a sketch of its solution, namely that S(3)=\sqrt{\frac{1+\sqrt{5}}{2}} , can be found in a paper of Friedman and Paterson [ 8 ] . In their paper, they also give a short argument to show that S(2)=1 , and remark that a similar method can be applied to prove that S(5)=2 and S(10)=3 . Complete proofs for the last two equalities were published by Januszewski [ 13 ] three years later. Some efficient coverings with larger values of n , in particular for n=7 and n=8 , are given in [ 8 ] . A weaker version of this problem was proposed by Soifer (see [ 23 ] or [ 21 , 22 ] ): for a positive integer n , find the smallest number \Pi(n) such that \Pi(n) unit squares can cover a square of edge length n+\varepsilon , he gave asymptotic estimates on this number as n\to\infty . To our knowledge, the best known bounds in this direction are \Pi(1)=3 ; 6\leq\Pi(2)\leq 7 ; and 11\leq\Pi(3)\leq 14 , where the inequalities 6\leq\Pi(2) and 11\leq\Pi(3) follow from the result in [ 13 ] . A different square covering problem was investigated in [ 3 ] . There, the authors consider squares of sizes 1,2,\dots,n , one of each size, and are interested in covering the largest possible square with translates of these squares. For small values of n , this problem is solved by a solver, for larger values of n by a heuristic, and for very large values of n by an asymptotically optimal algorithm. A higher-dimensional related question was posed by W. Kuperberg [ 17 ] . He asks for the smallest number C(d) of unit cubes in the d -dimensional Euclidean space that cover a cube of edge length 1+\varepsilon . For d=1 , this question is trivial and C(1)=2 , and for d=2 we have C(2)=\Pi(1)=3 . Kupenberg conjectures that C(d)=d+1 holds in any dimension. As a dual version of the above, there is a rich literature considering not covering, but packing squares into a larger square. One of the first publications regarding this problem is a paper of Erdős and Graham in 1975 [ 6 ] , who proposed the problem of finding a square with minimum edge length s(n) that can contain a packing of n unit squares for a positive integer n . In fact, Erdős considered the problem much earlier (as we are informed by [ 19 ] ). He showed that the sum of circumferences of two non-overlapping squares (of any size) inside a unit square is at most 4 . This observation was given by Erdős around 1932 as a problem for high-school students in Hungary. This is actually the simplest case of a more general conjecture in [ 6 ] , which states that the total circumference of any k^{2}+1 non-overlapping squares inside a unit square is at most 4k . The survey paper [ 7 ] of Friedman, published in 1998, presents many configurations that might be optimal for the problem in [ 6 ] , and also gives some results. Despite the work of many mathematicians (e.g. Stromquist, Cottingham, Wainwright, Stenlund, Trump, etc.), the problem is still open for almost all values of n . We recommend the paper [ 7 ] to the interested reader. Consider n squares of edge lengths 1,2,\ldots,n , respectively. The problem of finding the smallest square that can contain a packing of translates of these squares is discussed on the page of A005842 on the website of integer sequences [ 1 ] . This question is originated in the columns written by Martin Gardner in 1966 and 1975 [ 9 , 10 ] . Based on earlier works, Gardner [ 10 ] listed the best possible (tight) results up to n=17 . Some further research on this problem can be found e.g. in the publications [ 15 , 16 , 12 ] . The current champion results of upper bounds UB(n) up to n=56 can be found on the [ 1 ] website; most of those values are tight. However, tightness is still not proved for the cases n=38,40,42,48,52,53 and 55 . Hougardy [ 12 ] showed that UB(28)=89 , UB(32)=108 , UB(33)=113 , UB(34)=118 , and UB(47)=190 . The paper of Balogh et al. [ 2 ] considers this problem where n is large. Their result is the first one with asymptotic-type guarantees to the problem of packing consecutive squares into a square. In the paper both unrestricted axis-parallel packings and guillotine-type packings are considered. For both cases the paper introduces asymptotically optimal greedy algorithms. A very unique special case of the problem is n=24 . Watson mentioned already in 1918 in [ 24 ] that n=24 is the only case (different from n=1 ) where the sum of the areas of the consecutive squares from 1 to n , i.e. \frac{n(n+1)(2n+1)}{6} is a perfect square, namely 70\ast 70 . However, the optimum is 71 as proved by Korf [ 16 ] applying a computer-aided proof. The first purely theoretical proof was recently published by Sgall et al. [ 20 ] . A related question in the literature is the problem of packing an arbitrary set of squares with a total area of at most 1 into a rectangle of minimum area, see [ 18 ] and [ 14 ] . Possibly the best result so far is given in [ 11 ] , proving that a rectangle with an area smaller than 1.4 suffices to cover all those squares. A similar question, where the container is a square, was answered in [ 4 ] . Here the authors proved that any set of squares can be packed into the unit square if (i) the total

Read the complete text