Cet article est le résumé d’une conférance donnée à Harvard dans le cadre du colloque Process in media res.

There is a process involved behind every artistic and scientific productions. These processes can evolve, change directions and motivations, but at some point when the exact procedure is defined, automated processes can be constructed. The automated procedure is then available for others to be experimented and modified in order to find new applications and results. As this extra step is taken, an extended distance appears between the original creator of the process and the final result. Although, as pointed out by Einstein, when great specialisation is involved, the scientific and the artist merge into one identity (Calaprice 245) We show in this article that this double position between art and science is particularly present when creating automated processes. When creating abstract trends of patterns and procedures, the full extent of its applications rarely stands at reachable glance. On the other hand, the creation of subdivisions as copyright and patents leads the path for creators to think about the exact applications for their creations prior to their concretisation. This paper will explore the problematic involved in such a subdivision, especially in the paradigm of modern automated technologies. Various examples involving conceptual mathematic models, automated processes and visual art will be discussed in order to clarify the problematic.

As a first step, we compare different movies implying some mathematical concepts*: Zorns Lemma *(1970) by Hollis Frampton, *Last Year in Marienbad* (1961) by Alain Resnais and *Pi* (1998) by Darren Aronofski. These movies use different strategies to include mathematical concepts. The movie *Pi* is emblematic of the use of mathematics as a topic within its diegetic world. In this case, some concepts can be explained to the audience; the mathematical concepts are use in quotations since they don’t interfere with the structure of the movie itself. To a certain extent, these concepts could be changed for others and the structure would remain intact. As an example, the relation between the stock market and the value π could be exchange for the golden ratio to obtain a similar movie. It would remain an excellent movie with outstanding visuals aesthetic, only part of the semantic would be altered since the myth around pi differs largely from the myth around the golden ratio. These perceivable modifications would be linked to these specific numbers’ reputation outside the movie. For instance, the golden ration often being related to beauty, its use would charge scenes with a different emotional impact than the profoundly anxious and neurotic feeling that underline the whole movie. The value π does not work as a framing structure, it adds a mythological symbolism to its content and mark the film with a peculiar color coherent with the movie’s topic.

The film *Zorns Lemma* proposes a different appropriation of mathematical concept as a main constituent of art’s paradigm. The Zorn lemma is an important result in the foundations of modern logic and axiomatic set theory. It states that for a strictly partially ordered set, if every ordered subset has an upper bound in the original set, then the latest has a maximal element. The lemma has been proved independently by Kuratowski and by Bochner in 1922, but its popular appellation sticks to Zorn who proved it in 1935. (Munkres, p. 70)

Figure 1: Images from Zorns Lemma. Source: http://www.cia.edu/cinematheque/film-schedule/2013/02/zorns-lemma

The movie does not make apparent use of the lemma itself, although Frampton explicitly works its visual content from a set theoretical approach: groups of letters are combined as different sets to form words. As an example, in the second section groups of words appear ‘’organized alphabetically into sets of twenty-four and conforming to the Roman alphabet by combining *i *and *j* with *u *and *v.’’ *(Jenkins, p. 21) In this case, the abstract frame is calked from of a given field; set theory. Secondly, the object has a similar background question; how to organise elements of a set? In this case, the question is organise letters from the alphabet. The Zorn lemma appears as more than a mere abstract reference and its substitution for another theorem would note guarantee its correspondence with the movie structure. A title linked to the Pythagoras theorem, Fermat’s theorem or Gödel’s theorem would not be suitable references for Frampton’s work since we could not see a correspondence between the movie’s structure and the results of these theorems.

Figure 2: Last Year In Marienbad (Alain Resnais, 1961)

A slightly different approach is explored in Alain Resnais’s *Last Year in Marienbad*. In this film, the main character, interpreted by Giorgio Albertazzi, often plays the game of Nim -sometimes called the game of Marienbad after the movie (1)– and asserts that by starting first this would ensure him victory. On the mathematical side, the game was proved to be solvable, meaning that there is an algorithm leading inevitably to victory. (Bouton, 1902) The victorious pattern is presented multiple times during the movie and its logic is scaled to the overall frame of interplay with memory between to two main characters. The solvability of the game is implied in the movie as the dry output of destiny: the inevitable reconstitution of the forbidden, and maybe false, memory. The hunt for this blurred memory is ended before it started as the game of Nim is won before every game. As a result, the equivalence relation between the mathematics of the game and the movie’s structure is constructed by narrative means.

Figure 3: Time Structure of Last Year in Marienbad by Resnais

In between these poles of mathematic as subject, as structure and as narrative construction stands the automated processes. In the last 60 years or so, computers have galvanised and specialised the precision of the relations between the abstract mathematical procedures and visual content. Indeed, it has been possible by means of automated processes, especially in the construction of geometrical operations. The rest of the article focus on structures defined on abstract art instead of figurative or narratives as in the case of *Last year in Marienbad*. As a result, we are interested in authorship in arts and sciences from a double perspective: as creator of an aesthetic geometrical result and as inventor of an abstract structure. A clear and simple example of such a problematic objects can be found in the Ulam spiral. Bored during a meeting Stanislaw Ulam started to organise numbers in a spiral and in this structures some patterns seem to appear for prime numbers. This simple object of number disposition leads to beautiful imagery when focusing on the prime numbers disposition and to some new mathematical results about these prime numbers.

The signature over the aesthetic constituents being often available, we need to address the question to find the source of the structure and its authorship.

In order to comprehend this relation tied between a creator and an automated process, we need to distinguish between the different tributary relations linking an artistic visual object and an abstract automated process. It is important to underline the implied relation might appears in both directions; an artistic object can be obtain by applying an automated process, and oppositely, an automated process can be discovered by trying to solve an artistic problem. Both sides of this equation share the common ground of creation and the results, no matter what is the original paradigm, lay on shared space of double probability: the result stands in the midway between pure technicality and art. The next step of application of the automated process is fundamentally unpredictable. For this reason, the automated process is in equal rights as much an invention as an artistic creation. Of course, once a seed bloomed, layers and layers of artistic objects, related automated process, solutions to various problems and, finally, new problems might add to the complexity of the object. We study some examples in the following paragraphs.

A practice of tiling the planarity of a wall or a floor is maybe as old as architecture itself. There exist infinitely many ways to tile the plane, but these can be grouped in finite sets when restrictions are added or when classifications are needed. If we restrict the tiling to congruent tiles, then a classification is made possible by considering reflections, rotations, translations and glide reflections of the original tile. The artist Maurelius C. Escher studied these different patterns of tiling and tried to find all possible patterns. Escher found an article by Polya and Haag on crystallography giving the complete classifications of such tilings and Escher based his next experimentations on these observations. Even if Escher have found by himself almost all the patterns, it still give a good example of an abstract mathematical problem including automatic process related to an art object. In this case, the automated process constitutes of applying infinitely many translations, rotations, reflexions or glide-reflexions, to cover a space harmonically.(Figure 4) (Schattschneider, p.23-30)

Figure 4: Polya’s representation of the wallpaper groups. Source: Visions of Symmetry, p. 23

The story does not end here. Of course, different types of tessellations not involving congruent tiles have been explored as a legacy to Escher’s work and covering problem, like the Penrose aperiodic tiling and fractal tilings. The problem even evolved to include other surfaces; mathematicians and artists have explored the tiling of the sphere and this led even to tessellations on other surfaces as the hyperbolic plane or the projective plane (2). (Figure 5) Therefore, the creation of the tiling problem is double, it includes the eventual creation of a mathematical knowledge as much as of series of artistic creations. Moreover, it creates the space of discussion in which both disciplines challenge each other.

Figure 5: Jos Leys Hyperbolic 1

A similar story is hidden behind conformal mappings. Conformal mappings are functions that project images between surfaces, possibly from itself to itself, by preserving angles of intersection between lines. Conformal mappings arises as a main interest in the study of projections and the complex plane where they naturally arise as differentiable functions. A commonly used conformal mapping from the sphere to the plane is called the stereographic projection. To obtain this projection, we can imagine we set a sphere on a plane, and from the North Pole, i.e. the more distant point from the sphere, we traces rays crossing the sphere at a point and reaching the plane at second point. The stereographic projection is obtained when mapping the whole sphere to the plane in that respect.

In the last decades, photographers like Alexandre Duret-Lutz have used projection in order obtain pleasant photographs offering different spatial perspectives. The application of the stereographic projection lead to very peculiar pictures dubbed wee planets. In these photographs, objects are grotesquely deformed while keeping an overall readability due to the conformity of the projection. Ususally, the horizon surrounding the camera morphs into the circumference a small planet on the picture, resulting in pleasant cartoonesques scenes. (Figure 6) Modern photography contains more peculiar pictures calling for stronger mathematical notions. (Lambert, 2012)

Figure 6: wee planet Alexandre Duret-Lutz

The study of functions in the complex plane led August Ferninand Möbius to the definition of Möbius transforms, a group of conformal mappings constructed from translations, rotations, dilations and inversions (which inverts the inside and outside of a circle before rotating it). These functions are conformal and they can all be link to the stereographic projection through some motions of the sphere. (Arnold and Rogness) For instance, to obtain the inversion, it is equivalent to rotate the sphere upside down before applying the stereographic projection. The use of Möbius transformations is also recognisable in the photographs of Duret-Lutz, especially when the sky stands as a disk in the middle of the picture as a result of the inversion. Interestingly, artists are now applying similar techniques to video, Ryubin Tokuzawa (3). (http://www.ryubin.com/panolab/panoflash/#)

Other conformal mappings have been explores by photographs like Seb Pzbr or Josh Sommers. The utilisation by Sommers and Pzbr of a special composition of conformal mappings comes, though, from outside the mathematical discipline. In 1956, Escher worked on the highly complex *Printing Gallery*. The conformal mapping he tried to develop was so elaborate he could never finish his work, leaving a blank space in the middle. Half a century later, Lenstra and his team finally modeled the transformation Escher had in mind and, with the help of computers, they filled the blank spot. (Smit and Lenstra) The transformation, usually named the Droste effect -after on old advertisement using a self-referential figure- is now used by photographers to propose a wide range of new imageries, from self-portrait to the representation of abstract architecture. (Figure7)(4)

Figure 7 : Droste effect on architectural desing

JThe story of such photographs lies on multiple layers on each of which part of the authorship is diluted. It comes from a rich balance of complex numbers, functions, projections, Escher’s vision and programmers that integrated this process in code to obtain the results on photographs. This automated process and results from a 300 years old long dialogue where the authorship was constructed.

It is of prime importance to underline the presence in these pieces of art of the automated process: without the programs applying the conformal deformations, some photographs and videos, could never have existed. The creations, unreachable solely by humans, exist at the very limit of the creator’s capacity. It is the result of a tremendous collaboration where the sum worth more than the parts.

The epitome of art as resulting from automated process can be find in fractals. We will discuss two examples to underline two major components of the automated process; the structure, or skeleton, implied by the automata, and the theoretically possible infiniteness of its application in time and space. Fractals are geometrical figures defined by Mandelbrot in *Les objets fractals* (1975) in an attempt to describe the geometry of nature. These objects are often use by iterated processes and are self-similar for certain scale factors. Similar to the comments by Mandelbrot in his article *Fractals and an Art for the Sake of Science,* we can distinguished between two types of fractals, the organic and inorganic ones. The organic fractals identifies by and obvious similarity with nature whereas inorganic share the structural quality of independence of scale but keep evident traces of man’s hands.

Inorganic fractals usually results from a defined automated iterated process. For instance, the Koch curve is obtained by infinitely adding triangles on the middle thirds of each segments of its constitution. Surprises arise when one realises the same figure can be obtained from different automated processes. The Thue-Morse sequence is a sequence of zeros and ones built iteratively in such a way to avoid any triplet repetitions. It is constructed by the infinite concatenation of the complement of a binery sequence. The sequence is constructed as follow: 01, 0110, 01101001, 0110100110010110 etc.

It has been shown it was intrinsically related to the Koch curve: by assigning directions values to the digits of the sequence, it is possible to obtain the iterations of the Koch curve (Ma and Holdener, 2005). In a similar fashion, by assigning another set of instruction to the digits it has been demonstrated that the same Thue-Morse sequence can serve to obtain a tamil *kolam* (types of ritual figure drawn with sand or rice powder) (Allouche, Allouche and Shallit, 2006). The three entities, the Thue-Morse sequence, the Koch curve and this particular *kolam* are simply different interpretation of a common genetic code hidden in their automated iterative processes. (Figure 8) An automated process could, therefore, generate three or more different objects that we, from a visual point of view, consider distinct.

Figure 8: A Kolam and its equivalence as Koch curve and Thue-Morse sequence

Finally, we present a set of fractals named Julia sets and the Mandelbrot set. Again, it all roots back to the idea of representing complex numbers in a plane. Complex numbers are have two components, and real part on which we add an imaginary part, or equivalently a multiple of the square root of -1, denoted *i*. We usually write them *a+bi*. To reprensent them on the plane, we give the real component value to the x-axis and the imaginary part to the y-axis. Therefore the point (3,4) represent the complex number 3+4*i*. This representation helped understanding the way complex numbers multiply themselves and led to studies of conformal mappings as previously seen.

With this coordinate equivalence for a complex number, we can represent complex numbers on the plane as vectors with a length and a direction. The new vectors obtained from the multiplication has an angle equals to the sum of the previous vectors and a length equals to the product of their length. As a result, a number bigger than one will spiral out to infinity if multiplied by itself an infinite number of times. At the end of the First World War, the Academy of Science of Paris promised a prize for the better paper on complex numbers’ dynamic (5). From his hospital room where he cured his injuries, Gaston Julia wrote many important papers on the topic. He defined his set by the set of complex numbers not diverging to infinity when iterated in rational functions. For *C* a nonzero complex constant, the Julia set of quadratic forms *f(z) = z² + C *forms a fractal. Indeed, at the time, Julia did not have the tools to visualize the complexity of these sets. When the computer entered universities in the 60’s and 70’s, researchers started to code programs that would automatically generates Julia sets. The results started to evoke, even if only slightly, how rich wew the images Julia was trying to draw decades ago.

Figure 8: Julia Set

Nowadays, colors are added to these figures to produce marvelous pictures. On top of having to compute a great number of points in the complex plane in order to obtain a single picture, by a step by step focusing figures on the border of these sets we can obtain fractal zooms. In theory, these zooms could produce infinitely many different forms and could last forever, such is the complexity of Julia sets. (Figure 9) (6) A French mathematician, decided two classify the Julia sets into connected and disconnected ones. His classification led him to another infinitely complex set now dubbed the Mandelbrot set on which infinite zooms are also possible. The exploration of fractals led mathematicians into trying to define three dimensional versions of Julia sets and the Mandelbrot set. Difficulty arises and multiplication for complex numbers are to be represented. Since the complex numbers are defined on two dimensions, the real and imaginary one, the complex multiplication can be visualised in the plane, the representation of two complex numbers would need four dimensions. Fortunately, Paul Nylander have found a way to represent such mapping in three dimensions and three dimensional fractals based on this operation have arised, as for example the Mandelbox and the Mandelbulb. (Figure 10) As for the planar versions of these fractals, three dimensional fractals are never fully seen since they are infinitely intricate. Although, with the arrival of 3D printers, there’s been many attempt the represent some fractals such as the Sierpinski triangle. As well, it worths mentioning Tom Beddard’s work with fractal sculptures with lasers (7).

Figure 9: Inside of the Mandelbox, image by Krzysztof Marczak

Fractals apply naturally to arts, but they can also find specific technical application. For instance, there are many attempt to construct virtual landscapes based on fractal oriented programs. This tradition find its roots in the work of Voss who was developing programs to generate infinite maps, which has been done as well by Mandelbrot.

Again, the question of author seems problematic. The multi-level architecture behind these zooms starts with the definition of complex numbers, the idea of the complex plane, the studies of Julia, the long story behind computers and their programs, and then gigantic calculations made by the computers to generate a fractal zoom. The choice of the zoom’s point and the applications of specific colors is what is left to the last person involved in line, the artist. In that case, what is the fractal, where does it stands between discovery, invention and piece of art. Mandelbrot put it in these words: ‘’ Thus fractal art seems to fall outside the usual categories of ‘invention’, ‘discovery’ and ‘creativity’.’’ (Mandelbrot 1993, 14)

If again, no simple solution can be drawn from these various examples, they can still be linked to the idea of author. The idea of authorship is not only to refer to the existence of a creative process, but as well to contain a certain mark, a certain signature proper to the author. Of course, in the previous examples, elements of signature could be grasp at different level, in the choice of topic for a conformal mapping photograph, in the program’s style of coding lines, in the idea behind the proofs a the different theorems leading to these constructions. Since traces of authorship could be found at all these level, it shows that the notion transcend the simple binary separation between what is art and what is patentable. I urges as well that research centers such like universities to allow more permeability between areas of sciences and arts and include more classes on cross-disciplinary classes where students and researchers from both groups can meet and work together not only to solve problems, but to propose new ones as well.

Félix Lambert (First ideas presented at Harvard in 2013, first draft for this paper finished may 2015)

Notes:

(1) It was also called Fan-Tan at the beginning of the 20^{th} century (Bouton, 1902)

(2) For a clear introduction to the topic the reader is invited to consult John Stillwell’s work: *Geometry of Surfaces*, Springer, 1992.

(3) http://www.ryubin.com/panolab/panoflash/#)

(5) For the detailed history, the reader is referred to Michèle Audin’s work: *Fatou, Julia, Montel: The Great Prize of Mathematical Science of 1918.*Springer, 2011.

(6) http://social-biz.org/2010/03/28/generating-chaos/

(7) http://www.visualnews.com/2013/06/23/faberge-fractals-by-tom-beddard/

**Mediagraphy :**

Allouche, Gabrielle, Jean-Paul Allouche et Jeffrey Shallit. 2006. « Kolam indiens, dessins sur le sable aux îles Vanatu, courbe de Sierpinski et morphismes de monoïde ». En Ligne *: Annales de L’Institut Fourier*, Tome 56, n°7, p. 2115-2130. Consulté le 07/02/12. http://aif.cedram.org/item?id=AIF_2006_56_7_2115_0

Arnold, Douglas N. and Johnathan Rogness. 2008. « Möbius Transforms Revealed ». *Notices of the American Mathematical Society*, Vol. 55, Nu. 10, p. 1226-1231.

Audin, Michèle. 2011. *Fatou**, Julia, Montel : The Great Prize of Mathematical Sciences of 1918, and Beyond*. New York: Springer, Lecture Notes in Mathematics 2014, History of Mathematics Subseries.

Bouton, Charles. 1902. « Nim, a Game with a Complete Mathematical Theory ». Annals of Mathematics, Second Series, Vol. 3, no. 1. P. 35-39.

Calaprice, Alice, Ed. 2000. *The Expandable Quotable Einstein*. Princeton: Princeton University Press.

Frampton, Hollis. 1970. *Zorns Lemma*. In *A Hollis Frampton Odyssey*, DVD 1, 59 min. Criterion Collection 2012.

Ma, Jun and Judy Holdener. 2005. « When Thue-Morse Meets Koch ». *Fractals: Complex Geometry, Patterns, and Scaling in Nature and Society*, vol. 13. n°3, p. 191-206.

Mandelbrot, Benoît*. Les Objets Fractals : Formes Hasard et Dimension*, 4th Ed. Paris: Flammarion, 1995.

Mandelbrot, Benoït. 1993. ‘’Fractals and an Art for the Sake of Sciences’’. In Michel Emmer Ed. The Visual Mind: Art and Mathematics. Cambridge: MIT Press, p.11-14.

Munkres, James R. *Topology*. 2^{nd} Ed. New Jersey: Prentice Hall, Inc., 2000.

Schattschneider, Doris. 1992 (1990). *Visions de la Symétrie: Les Cahiers, les Dessins Périodiques et les Oeuvres Corrélatives de M.C. Escher*. Traduit de l’américain par Marie Bouazzi. Paris : Éditions du Seuil.

Smit, B. de and H.W. Lenstra Jr. 2003. « Artful Mathematics: The Heritage of M.C. Escher». *Notices of the American Mathematical Society*, Volume 50, nu 4, p. 446-451.

Stillwell, John. *Geometry of Surfaces. *New York: Springer-Verlag, 1992.