4.1 Set of circumscribed circles

Definition 4.1.1.

Let S2S\subset\mathbb{Z}^{2}. The set of centres of circumscribed circles of SS with radius rr, cr(S)c_{r}(S) is the set of all points (x,y)(x,y) such that the integer circle Sr(x,y)S_{r}(x,y) is a circumscribed circle of the set SS.

Remark 4.1.2.

For S=∅︀S=\emptyset, we have cr(S)=2c_{r}(S)=\mathbb{Z}^{2}.

Definition 4.1.3.

Let S2S\subset\mathbb{Z}^{2}. The set of circumscribed centres of SS, C(S)C(S) is the union of sets:

C(S)=rcr(S).C(S)=\bigcup_{r\in\mathbb{Z}}c_{r}(S).

The set of circumscribed centres is the set of all points which are the centre of a circumscribed circle of SS. In other terms, this is the set of all points that are equidistant to all vertices of SS.

Example 4.1.4.

Let pp be an integer point. Then

C({p})=2{p}.C(\{p\})=\mathbb{Z}^{2}\setminus\{p\}.
Remark 4.1.5.

Recall that in Euclidean geometry, a line is the locus of points equidistant from two points. In integer geometry, the locus of points equidistant from p1p_{1} and p2p_{2} is given by

L=C({p1,p2})).L=C(\{p_{1},p_{2}\})).

In this chapter, we study properties of cr(S)c_{r}(S) and C(S)C(S). In particular, we compare the sets SS and cr(cr(S))c_{r}(c_{r}(S)).

Proposition 4.1.6.

Let SS be a set of integer points, and let rr be an integer. Then Scr(cr(S))S\subset c_{r}(c_{r}(S)).

Proof.

Let xSx\in S. Then for all points pcr(S)p\in c_{r}(S) we have that l(x,p)=r\operatorname{l\ell}(x,p)=r. Thus xcr(cr(S))x\in c_{r}(c_{r}(S)). Hence, Scr(cr(S))S\subset c_{r}(c_{r}(S)). ∎

Corollary 4.1.7.

Let S2S\subset\mathbb{Z}^{2}. Then C(C(S))SC(C(S))\subset S.

Proof.

This follows from the fact cr(S)C(S)c_{r}(S)\subset C(S)

Remark 4.1.8.

Since cr(∅︀)=2c_{r}(\emptyset)=\mathbb{Z}^{2}, if SS does not admit a circumscribed circle of radius rr, we have that cr(cr(S))=2c_{r}(c_{r}(S))=\mathbb{Z}^{2}.

Definition 4.1.9.

Let SS be an integer set with (0,0)S(0,0)\in S and let m=maxΛ(S)m=\max\Lambda_{\mathbb{Q}}(S) when it exists. Otherwise, let m=1m=1. Then the set

m(c1(c1(1mS)))m\left(c_{1}\left(c_{1}\left(\frac{1}{m}S\right)\right)\right)

is the principle circular-congruent set to SS we denote it by c¯(S)\overline{c}(S).

Let S^\hat{S} be an integer set. Then if c¯(S^)=c¯(S)\overline{c}(\hat{S})=\overline{c}(S), we say that S^\hat{S} and SS are circular-congruent.

Remark 4.1.10.

In the case when SS has only one element, Λ(S)\Lambda_{\mathbb{Q}}(S) is not bounded.

Remark 4.1.11.

We define c¯\overline{c} with maxΛ(S)\max\Lambda_{\mathbb{Q}}(S) so that it is defined for sets SS which only admit rational circumscribed circles.

Remark 4.1.12.

If SS and S^\hat{S} are integer-congruent subsets of 2\mathbb{Z}^{2} then SS and S^\hat{S} are circular-congruent.

Definition 4.1.13.

We say that SS is maximal if S=c¯(S)S=\overline{c}(S).

Example 4.1.14.

A one-point set is maximal.

For a two-point set we have a direct generalisation of the classical definition of a line as the set of points at the same distance from a given pair of points.

Problem 2.

Is it true that c(c(S))=c¯(S)c(c(S))=\overline{c}(S) in case of c(S)c(S) is non-empty? What about cr(cr(S))?c_{r}(c_{r}(S))?

Definition 4.1.15.

Let SS be a set. We say that SS^{\prime} is the toric complement to SS if:

  • SS^{\prime} is tori-transparent at all points in SS,

  • SS^{\prime} includes all points in 2\mathbb{Z}^{2} that are not tori-equivalent to a point in SS.

Remark 4.1.16.

If max(Λ(S))=1\max(\Lambda_{\mathbb{Q}}(S))=1 then the toric complement to SS is c1(S)c_{1}(S).

Theorem 4.1.17.

Let SS be a finite set, then c¯(S)=S\overline{c}(S)=S.

Proof.

Without loss of generality we suppose that max(Λ(S))=1\max(\Lambda(S))=1.

Since set of centres of SS is the toric complement to SS, we write

c1(S)=aS2[a]p.c_{1}(S)=\bigcap_{a\in S}\mathbb{Z}^{2}\setminus\cup[a]_{p}.

Hence

c¯(S)=aS[a]p.\overline{c}(S)=\bigcup_{a\in S}\cap[a]_{p}.

Since there is only one point in 2\mathbb{Z}^{2} which intersects each integer torus at the same point, we find that c¯(S)=S.\overline{c}(S)=S.

Example 4.1.18.

Let SS be a finite set. Then the complement S¯\overline{S} has no circumscribed circles and hence cc(S¯)=2cc(\overline{S})=\mathbb{Z}^{2}. Therefore S¯cc(S¯)\overline{S}\neq cc(\overline{S}). Hence the statement of Problem 4.1.17 is not necessarily true for an infinite set SS.

For the next proposition, we need the following definition:

Definition 4.1.19.

An exhaustive filtration \mathcal{F} is affine-invariant if for any S2S\subset\mathbb{Z}^{2} and AAff(2,)A\in\operatorname{Aff}(2,\mathbb{Z}) we have the following. If the density of SS exists, then the density of A(S)A(S) exists and

#S=#A(S).\#_{\mathcal{F}}S=\#_{\mathcal{F}}A(S).
Proposition 4.1.20.

Let FF be an affine-invariant exhaustive filtration. Let CC be a unit circle and let CCC^{\prime}\subset C such that the density of CC^{\prime} is not less than the density of intersection of two integer unit circles with respect to some exhaustive filtration. Then c1(c1(C))=Cc_{1}(c_{1}(C^{\prime}))=C.

Proof.

Let pp be the centre of CC. Then pc1(C)p\in c_{1}(C) and hence pc1(C)p\in c_{1}(C^{\prime}). Assume that any qpq\neq p is also in c1(C)c_{1}(C^{\prime}). Therefore c1(C)(S1(p)S1(q))c_{1}(C^{\prime})\subset(S_{1}(p)\cap S_{1}(q)). This is impossible due to our assumption. Hence c1(C)={p}c_{1}(C^{\prime})=\{p\}. Therefore c1(c1(C))=Cc_{1}(c_{1}(C^{\prime}))=C. ∎

Remark 4.1.21.

Let CC be a unit circle and let CCC^{\prime}\subset C such that the density #C#(S)\#_{\mathcal{I}}C^{\prime}\geq\#_{\mathcal{I}}(S) Then c1(c1(C))=Cc_{1}(c_{1}(C^{\prime}))=C.

Problem 3.

Is cc(C)=uu(C)cc(C)=uu(C), here uu(C)uu(C) is the intersection of circles centered at the the intersection of circles centered at SS.