Fractals as Sculptures
Şahin Koçak · 2025
Read Fractals as Sculptures by Şahin Koçak (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-2510-25949.txt.
Introduction
Sahin Kocak, working alone, gives a new way to view the attractor of an iterated function system as a sculpture like intersection. An iterated function system is a set of shrinking maps whose repeated use builds a fractal, a shape with fine detail at every scale. Usually the attractor is defined as the limit, in the Hausdorff sense, of any nonempty compact set under repeated iteration.
Kocak shows that on a boundedly compact metric space, such a system has compact, invariant subspaces with a special property. The attractor can then be written as the intersection of a decreasing chain of compact spaces. The note offers a clean alternative picture of a standard fractal construction.
Catalog record
| CREATOR | Şahin Koçak |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 8 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2510-25949
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Fractals as Sculptures Şahin Koçak Abstract The attractors of iterated function systems are usually obtained as the Hausdorff limit of any non-empty compact subset under iteration. In this note we show that an iterated function system on a boundedly compact metric space has compact, invariant subspaces so that the attractor of the IFS can also be expressed as the intersection of a sequence of decreasing compact spaces. Keywords: Iterated function systems MSC: 28A80 Painting and poetry have more to do with “adding”, but sculpture has more to do with “subtracting”. As the legend goes, Michelangelo must have said to the astounded spectators of David, “He was already in the marble block, I just had to chisel away the superfluous material”. Figure 1 : Michelangelo chiseling away the superfluous material ( [ 1 ] ) I remembered Michelangelo’s words as I for the first time encountered the Cantor set in my youth. Cantor was taking his thin and fragile block and chiseling away some superfluous material. He was first discarding the (open) middle third, without hesitating to make his block disconnected (in contrast to sculptors whose works are mostly connected), then discarding the middle thirds of the remaining two pieces, and was going on forever applying this recipe. What remained in his hands, was utterly disconnected. We called it later a fractal! These points surviving all the deletions are the intersection of all the intermediary stages. (see Fig. 2 ) Figure 2 : Cantor Set obtained by successive deletions Then Sierpinski took a triangle, discarded the (open) middle fourth, then the middle fourths of the remaining three, and so on… The surviving points gave another fractal. (see Fig. 3 ) Figure 3 : Sierpinski Triangle obtained by successive deletions Menger came maybe closer to the great masters of art as he took a 3-dimensional cube and chiseled away tunnels through it. (see Fig. 4 ) This marvellous fractal having zero volume contains nevertheless a homeomorphic copy of every compact topological space of topological dimension 1. It has been very recently ( [ 2 ] ) shown that it even contains a copy of every knot! Figure 4 : Chiseling tunnels through a cube In the seventies there were a large collection of “fractals” around and the term was coined and the notion was made popular to the public by Mandelbrot ( [ 3 ] , [ 4 ] ). Obviously a systematic treatment of these creatures was in need and Hutchinson devised the notion of iterated function systems ( [ 5 ] ), which comprised many of these fractals; though a subspecies emerging from a totally different source, from some innocent-seeming dynamical systems, resisted to pass into this scheme. Nevertheless, the iterated function systems were able to mimic many “fractals”, including even botanical shapes ( [ 6 ] ) and we became addicted to the newly upcoming computers to see the slowly emerging, startling fractals. You consider some contractions f_{1},f_{2},\dots,f_{N} on a complete metric space X and show that there exists a unique (non-empty) compact subspace A satisfying f_{1}(A)\cup f_{2}(A)\cup\dots\cup f_{N}(A)=A , which you declare to be a “fractal” (we use this term in a loose sense, calling to mind spaces with some kind of self-similarity and fractional dimension). The iterated function system (IFS) approach is constructive in the sense that you can start with any (non-empty) compact subspace B\subset X , build the image F(B)=f_{1}(B)\cup f_{2}(B)\cup\dots\cup f_{N}(B) , then iterate this process obtaining a sequence B,F(B),F(F(B))=:F^{2}(B),\dots,F^{n}(B),\dots of which the longed-for fractal A is the limit point with respect to the Hausdorff metric. This metric is meanwhile an indispensible tool for pattern recognizers! (For a lovely exposition of the details we refer to [ 6 ] .) The only unhappy side of this construction is that it does not need to hold B\supset F(B)\supset F^{2}(B)\supset\cdots F^{n}(B)\supset\cdots\ . If this would be the case, the infinite intersection would be the Hausdorff limit and we could imagine like a sculptor that this limit is obtained by successive deletions from the initial compact set B . But generally this is not true. However we will show below that in fact such a compact subspace B , which is invariant under all the functions constituting the iterated function system can always be found, thus somehow reconciling the IFS approach with the classical deletion approach. Our candidate of such a compact subspace B will be a so-called multi-foci ellipse, a notion which might have interested the mathematicians of the antiquity, but apparently was investigated only in the 19. century by a teen no lesser than Maxwell, the creator of the famous electro-magnetic equations ( [ 7 ] ). His “oval curves” are defined as the locus of points on the plane whose distance-sum to a given (finite) set of points is a given number. The same definition applies surely to any metric space, where we allow the distance-sum to be less than or equal to a given number. We will use in the following multi-foci ellipses whose focal points will be the fixed points of the contractive functions of the given iterated function system. Figure 5 : Maxwell’s drawing of a 3-foci ellipse (with focal points A,B and C ) with a thread ( [ 7 ] ) (and without a computer, which was non-existent anyway!) In the following proposition we will make the slightly stronger assumption that the space X is boundedly compact which suffices for all applications and which implies completeness. Proposition 1 . Let (X,d) be a boundedly compact metric space and f_{i}:X\to X , i=1,2,\dots,N be contractions. Then, there exists a (non-empty) compact subspace B\subset X with f_{i}(B)\subset B for all i=1,2,\dots,N. Proof. Let the contractions f_{i} have the contraction ratios \lambda_{i} with 0<\lambda_{i}<1 , and a_{i}\in X\ (i=1,2,\dots,N) be the fixed points of f_{i} (i.e. f_{i}(a_{i})=a_{i} ). Let \displaystyle\lambda:=\max_{1\leq i\leq N}\{\lambda_{i}\},\ \displaystyle D_{i}:=\sum_{j=1}^{N}d(a_{i},a_{j}),\ D:=\max_{1\leq i\leq N}\{D_{i}\} and M:=\dfrac{1+\lambda}{1-\lambda}D . We consider the N- foci ellipse \displaystyle B=\{x\in X\mid\sum_{j=1}^{N}d(x,a_{j})\leq M\} . For x\in B , we show that f_{i}(x)\in B, i.e. \displaystyle\sum_{j=1}^{N}d(f_{i}(x),a_{j})\leq M for all i=1,2,\dots,N . \begin{array}[]{ll}\text{For }j=i:&d(f_{i}(x),a_{i})=d(f_{i}(x),f_{i}(a_{i}))\leq\lambda d(x,a_{i})\\ \text{For }j\neq i:&d(f_{i}(x),a_{j})\leq d(f_{i}(x),a_{i})+d(a_{i},a_{j})\\ &\leq\lambda d(x,a_{i})+d(a_{i},a_{j})\leq\lambda d(x,a_{j})+(\lambda+1)d(a_{i},a_{j})\end{array} Thus, \begin{array}[]{ll}\displaystyle\sum_{j=1}^{N}d(f_{i}(x),a_{j})&\leq\lambda d(x,a_{i})+\displaystyle\sum_{j\neq i}[\lambda d(x,a_{j})+(\lambda+1)d(a_{i},a_{j})]\\ &\leq\lambda\left(\displaystyle\sum_{j=1}^{N}d(x,a_{j})\right)+(\lambda+1)D_{i}\leq\lambda M+(\lambda+1)D=M.\end{array} ∎ Example 1 . We consider the IFS on \mathbb{R}^{2} consisting of two contractions f_{1} and f_{2} , whereby f_{1} is a rotation of angle \pi/6 around the origin in the counterclockwise direction with contraction ratio 1/2 and f_{2} is a rotation of angle \pi/6 around (1,0) in the clockwise direction with contraction ratio 3/5 . If we choose the disk with center (0,0) and radius 1 as a random starting set B of iteration, for example, then F(B)=f_{1}(B)\cup f_{2}(B) is not contained in B (See Fig.6). If however we choose the starting set as the ellipse \frac{(x-1/2)^{2}}{4}+\frac{y^{2}}{15/4}\leq 1 as suggested by the proof above, it holds F(B)=f_{1}(B)\cup f_{2}(B)\subset B . (See Fig.7a. We also show the second iteration of this IFS in Fig.7b) It seems that our sculptor will have a difficult job of laborious deletions to get a faint impression of the emerging fractal. To make him/her the life easier and to give a glimpse into the fractal we can offer to use an iteration á la Hutchinson and Barnsley, starting with a singleton, say \{(1,1)\}