57
$\begingroup$

What are some (research level) open problems in Euclidean geometry ?

(Edit: I ask just out of curiosity, to understand how -and if- nowadays this is not a "dead" field yet)

I should clarify a bit what I mean by "Euclidean geometry". By this term I mean, loosely, the study of the geometry of certain subsets of Euclidean space $\mathbb{E}^n$ from a point of view which is either the "classical" one (i.e. axiomatic), or one that involves more modern tools, but the problem in question has not to be "clearly" a problem within some other branch of maths such as differential or algebraic geometry, algebraic or general topology, analysis, or measure theory.

Some examples to clarify:

  • the study of configurations of lines or affine subspaces is EG; but the algebro-topological study of hyperplane arrangements is not.
  • plane conics as defined via their metric property are objects of EG; but "algebraic curves" are not, unless they're defined by some "elementary enough property" (intentionally vague) involving the Euclidean metric.
  • root systems of Lie algebras are EG.
  • polyhedral cones are EG.
  • polytopes are EG.
  • tessellations of space with polytopes or analogous objects are in EG.
  • minimal surfaces in $\mathbb{E}^3$ are not EG.
  • fractal geometry (Julia sets, self-affine fractals...) is not EG.
  • not sure about convex bodies. If they're polyhedral I'd say their study fits in EG.
  • packings of spheres are EG.
$\endgroup$

14 Answers 14

33
$\begingroup$

The Unit Distance Problem asks:

For a set of $n$ points in the plane, what is the maximal number $g(n)$ of unit distances realized among the ${n \choose 2}$ pairs?

A properly scaled square grid gives a lower bound of something like $g(n) \ge n^{1 + \frac{c}{\log \log{n}}}$, and a beautiful application of the crossing number lemma gives that $g(n) = O(n^{4/3})$.

A closely related problem where great progress was made very recently is the Distinct Distance problem, asking for the minimum number $f(n)$ of distinct distances among $n$ points in the plane. (Clearly $f(n)g(n) \ge {n \choose 2}$.)

Guth and Katz recently obtained a sharp exponent for $f(n)$. Terence Tao and János Pach wrote nice summaries of this work.

$\endgroup$
22
$\begingroup$

The happy ending problem ( https://en.wikipedia.org/wiki/Happy_Ending_problem ) says that any set of five points in the plane in general position has a subset of four points that form the vertices of a convex quadrilateral. More generally, Erdös and Szekeres proved that for any positive integer $N$, there is a minimal integer $f(N)$ such that any set of $f(N)$ points in the plane in general position has a subset of $N$ points that form the vertices of a convex polygon, and it is known that $f(N)$ is at least $1+2^{N-2}$.

An open question is: does $f(N)=1+2^{N-2}$ hold?

Added in edit: As pointed out by Suvrit in comment, a recent progress has been posted on the arXiv, showing that $f(N)=2^{N+o(N)}$.

$\endgroup$
1
  • 5
    $\begingroup$ I just saw this: arxiv.org/abs/1604.08657 which claims to "nearly settle" the cited open question... $\endgroup$
    – Suvrit
    May 2, 2016 at 1:11
20
$\begingroup$

Chromatic Number of the Plane or Hadwiger–Nelson problem asks for the minimum number of colors required to color the plane such that no two points at distance $1$ from each other have the same color. It is only known that $$5 ≤ \chi ≤ 7.$$

$\endgroup$
11
  • 8
    $\begingroup$ Aubrey de Grey improved the lower bound from $4$ to $5$ in April 2018. $\endgroup$
    – Mark S
    Apr 22, 2018 at 23:28
  • 3
    $\begingroup$ Is this really a problem in Euclidean geometry? I would call it a problem in combinatorics. The OP excluded "other branches of mathematics". They didn't mention combinatorics explicitly, but I don't see any difference. (The same comment might apply to several of the other answers.) $\endgroup$
    – HJRW
    Apr 23, 2018 at 13:37
  • 2
    $\begingroup$ The issue is "unit distance". That and the fact that the graphs are in the Euclidean plane suggest to me that it is a candidate. The easy upper bound is a geometric argument combined with a simple combinatorial observation. Although a resolution may end up using combinatorics and graph theory primarily, geometric assistance has by no means been ruled out. Gerhard "Let's Give Geometers A Chance" Paseman, 2018.04.23. $\endgroup$ Apr 23, 2018 at 15:09
  • 1
    $\begingroup$ @GerhardPaseman, the list of excluded topics suggests that merely using a geometric notion to set up the problem is not enough to qualify. In terms of the practice of research mathematicians, this is certainly a question in combinatorics -- note that de Grey didn't cross-post his preprint to any other subject. $\endgroup$
    – HJRW
    Apr 23, 2018 at 15:17
  • 2
    $\begingroup$ A nice article about de Grey’s solution: theguardian.com/science/2018/may/04/… . Note that both experts quoted work in combinatorics. $\endgroup$
    – HJRW
    May 5, 2018 at 12:11
19
$\begingroup$

W. Wernick has tabulated 139 triangle construction problems using a list of sixteen points associated with the triangle. In each case three points are given and the goal is to construct a triangle for which the three 'special' points listed are the points given. There are still some open problems in Wernick's list. Notation:

$A$, $B$, $C$ Three vertices,

$M_a$, $M_b$, $M_c$ Three midpoints of the sides,

$H_a$, $H_b$, $H_c$ Three feet of the altitudes,

$T_a$, $T_b$, $T_c$ Three feet of the internal angle bisectors,

$G$, $H$, $I$, $O$ The centroid, orthocenter, incenter and circumcenter.

All problems are divided into the following four distinct types:

$R$ – Redundant. Given the location of two of the points of the triple, the location of the third point is determined. An example would be: $A$, $B$, $M_c$.

$L$ – Locus Restricted. Given the location of two points, the third must lie on a certain locus. Example: $A$, $B$, $O$.

$S$ – Solvable. Known ruler and compass solutions exist for these triples.

$U$ – Unsolvable. By using algebraic means, it is possible to prove that no ruler and compass solution exists for these triples. Example: $O$, $H$, $I$.

enter image description here

Update:

Wernick's list has been completed by Pascal Mathis and Pascal Schreck, see Determining automatically compass and straightedge unconstructibility in triangles.

$\endgroup$
4
  • 1
    $\begingroup$ Frankly, many of these are not particularly interesting. Especially the feet of the angle bisectors are rather obscure objects in terms of standard triangle geometry (they have some interesting properties, such as the ones the Emelyanovy have proven, but they are no more natural than, say, the three excenters, the Brocard points, the points where the incircle touches the sides, the projections of the symmedian point on the sides, the vertices of the tangent triangle, etc.). $\endgroup$ Apr 9, 2015 at 19:08
  • 1
    $\begingroup$ I think that the feet of the angle bisectors is one the most intetresting problems from this list. $\endgroup$ Apr 10, 2015 at 6:24
  • $\begingroup$ "The requested page was not found on this server." easychair.org/publications/paper/cw seems to work. $\endgroup$ Apr 23, 2018 at 3:23
  • 1
    $\begingroup$ @GerryMyerson fixed $\endgroup$ Apr 23, 2018 at 10:17
18
$\begingroup$

I might as well air this question (first posed by Keith Ball) that has sweeping ramifications in convex geometry in high dimensions if the answer is yes:

Let $K$ be a centrally symmetric convex body in $\mathbb{R}^n$, and let $K^\circ$ be the polar or dual convex body. Define a statistic $e(K)$ as the expected value of $(\vec{x} \cdot \vec{y})^2$, where $\vec{x}$ is chosen randomly from $K$ and $\vec{y}$ is chosen randomly from $K^\circ$. Then for each fixed $n$, is $e(K)$ maximized when $K$ is an ellipsoid? The question is even open in two dimensions.

A much weaker conjecture is that the integral of $(\vec{x} \cdot \vec{y})^2$ over $K \times K^\circ$, as opposed to the average value, is maximized when $K$ is an ellipsoid. It is known that $K \times K^\circ$ has the most volume when $K$ is an ellipsoid; this fact is called Santaló's inequality.

It is known that the answer to the first conjecture is no if $K$ is not centrally symmetric, even if the origin is the only points fixed by the symmetries of $K$. (Central symmetry means specifically that $K = -K$.)

$\endgroup$
7
  • 3
    $\begingroup$ This will certainly be a deep problem, but I'm not sure it fits into the "euclidean geometry" label, as -i guess- a convex body which is not a priori a polytope or an ellipsoid may have a very complicated shape. It sounds more like an 'analysis thing', but someone more knowladgeble than me may judge about this. $\endgroup$
    – Qfwfq
    Jan 30, 2011 at 23:52
  • 2
    $\begingroup$ If you can prove it for convex polygons in the plane, then I will already be very interested. $\endgroup$ Jan 31, 2011 at 10:37
  • 2
    $\begingroup$ Moreover, in general convex polytopes are as good as the general case of the problem. $\endgroup$ Jan 31, 2011 at 10:37
  • $\begingroup$ Of course - as you well know, Greg - by the reasoning of your last comment most of the biggest open problems about convex bodies fit here (including the ones I'm surprised you refrained from mentioning in your answer). $\endgroup$ Jan 31, 2011 at 12:58
  • 1
    $\begingroup$ @Mark I decided to go for a question that I particularly care about, instead of offering a general review. $\endgroup$ Jan 31, 2011 at 22:29
16
$\begingroup$

An important open problem in combinatorial Euclidean geometry is the question of how many different halving lines a set of $2n$ points in the Euclidean plane may have, in the worst case. A halving line is a line through two of the points such that $n-1$ of the points are on each of its sides. The number of halving lines is known to be $O(n^{4/3})$, and there are examples of point sets for which this number is $n2^{\sqrt{\Omega(\log n)}}$, but there remains a large gap between these upper and lower bounds.

$\endgroup$
3
  • 1
    $\begingroup$ No three points on a line? $\endgroup$ Apr 10, 2015 at 2:23
  • 1
    $\begingroup$ It doesn't make a difference: if you have a point set allowing collinearities you can always perturb it to have no collinearities and at least as many halving lines. $\endgroup$ Apr 10, 2015 at 19:08
  • 3
    $\begingroup$ Of course. I just wanted the problem to be stated cleanly, without readers stopping over trivial non-essential elements, even if only for a moment. $\endgroup$ Apr 10, 2015 at 23:14
14
$\begingroup$

I am reposting my own answer from Not especially famous, long-open problems which anyone can understand for it answers this question as well.

Are there eight points on the plane, no three on a line, no four on a circle, with integer pairwise distances?

The analogous question for seven points was posed by Paul Erdős and answered positively by Kreisel, Kurz 2008, who then asked this question.

Tobias Kreisel, Sascha Kurz, There Are Integral Heptagons, no Three Points on a Line, no Four on a Circle, Discrete & Computational Geometry 39/4 (2008), 786-790. (Internet Archive)

$\endgroup$
13
$\begingroup$

In recent years there have been a good amount of surveys and publications on "computational" or "combinatorial" geometry, and looking at them may give you a good idea of current questions. Specifically, there is the excellent recent book "Research Problems in Discrete Geometry" by Brass, Moser, and Pach. You may want to start by looking there and at the references it provides. Besides a good deal of information on classical questions, among many other topics, you find:

  • Density problems for packings and coverings.
  • Distance problems.
  • Lattice point problems.
  • Graph drawings and geometric graphs.
  • Geometric inequalities.
$\endgroup$
10
$\begingroup$

Among the many choices one might get from an Internet search, I suggest Unsolved Problems in Geometry by Hallard Croft, Kenneth Falconer, and Richard Guy (Springer-Verlag, 1991). It may include references to non-Euclidean geometries.

As an aside, I would like to see a geometric proof that the configuration of Pappus is implied by that of Desargues for finite geometric spaces.

Gerhard "Ask Me About System Design" Paseman, 2011.01.30

$\endgroup$
10
$\begingroup$

What is the largest integer $k$ such that any $k$ points in the plane, no matter how they are arranged, can always be covered with disks with pairwise-disjoint interior having radius $1$?

See Covering Points with Disjoint Unit Disks for the history.

$\endgroup$
2
  • 1
    $\begingroup$ This has also been asked as a question on our site: mathoverflow.net/q/20558/5340 What is the minimum N for which there exist N points in the plane that cannot be covered by any number of non-overlapping closed unit discs? $\endgroup$ Apr 14, 2015 at 9:38
  • $\begingroup$ @Zsbán Ambrus Thank you, I didn't see this question. $\endgroup$ Apr 14, 2015 at 12:28
9
$\begingroup$

The classification of all pentagons tiling the plane is not completely finished, see for example MR3037862. (Bagina, O. G. Convex pentagons that tile the plane (types: 11112, 11122). (Russian) Sib. Èlektron. Mat. Izv. 9 (2012), 478–530.)

Update: This problem is now probably solved: see e.g. https://en.wikipedia.org/wiki/Pentagonal_tiling and references therein.

$\endgroup$
1
  • 1
    $\begingroup$ Is it not? I thought this was solved a few years ago: quantamagazine.org/… — or do you mean not just the convex case but also concave pentagons? $\endgroup$ Dec 9, 2019 at 21:22
7
$\begingroup$

Polynomial Hirsch Conjecture?

$\endgroup$
5
$\begingroup$

In addition to the Croft-Falconer-Guy and Brass-Moser-Pach books others have mentioned, there's Victor Klee and Stan Wagon, Old and New Unsolved Problems in Plane Geometry and Number Theory, No. 11 in the MAA series, Dolciani Mathematical Expositions, from 1991.

$\endgroup$
-5
$\begingroup$

Let $L_0,L_1,L_2,L_3,L_4$ be five lines in general position on the Euclidean plane---think of the subscripts mod $5$ and draw $L_i$ as the consecutive lines of a (not necessarily regular!) pentagram. Let $C_i$ be the circle inscribed about the triangle formed by $L_i$, $L_{i-2}$, and $L_{i+2}$. Then $C_{i-1}$ and $C_{i+1}$ meet at the intersection of $L_{i-1}$ and $L_{i+1}$, and again at some other point $P_i$ (which we take to be the same point if the two circles are tangent there). Show that these five points $P_i$ are concyclic.
.
Actually, it isn't open, but some years ago presented to me as open, with computer-graphical evidence for it...

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.