Penrose P2 Tilings: A Study of Fully Leafed Induced Subtrees Mathieu Cloutier, Alain Goupil, Alexandre Blondin Massé Abstract We present new results about fully leafed induced subtrees in Penrose P2 tilings. We first determine the graph structure of these subtrees and show that they are caterpillars, up to an appendix of at most six tiles. We then study bi-infinite fully leafed induced caterpillars in P2 tilings and their geometric properties. In particular, we refute the conjecture proposed by C. Porrier, A. Goupil and A. Blondin Massé that there is a unique bi-infinite fully leafed caterpillar in Penrose P2 tilings. 1 Introduction Given a tiling, one may be interested in finding connected and acyclic sets S of n tiles that maximize the number of tiles of S adjacent to exactly one other tile of S . These tiles can be viewed as extremities of S . These structures are called fully leafed induced subtrees and we study them in Penrose P2 tilings. Fully leafed induced subtrees can be used in chemistry, for example to study adsorption on surfaces with a maximal number of boundary sites [ 7 ] . Since Penrose tilings model quasicrystals, the results presented here may be of interest in the study of adsorption on quasicrystal surfaces [ 6 ] [ 8 ] [ 12 ] . Section 2 presents the definitions needed for this work. In Section 3, we establish the graph structure of fully leafed induced subtrees in Penrose P2 graphs. Section 4 describes the construction of fully leafed induced subtrees, taking into account the geometry of the tilings in which they are embedded. Section 5 studies the construction of bi-infinite fully leafed induced caterpillars. We conclude in Section 6. Complete proofs of the new results presented here, as well as additional results on the same topic, can be found in the extended version of this paper [ 3 ] . 2 Definitions Basic definitions from graph theory and tilings can be found in [ 4 ] and [ 5 ] . The vertices of degree one in a tree T are called the leaves of T . For a graph G , the graph G^{\prime} obtained by removing the leaves from G is called the derived graph of G . A caterpillar in G is a subgraph \mathcal{C} of G such that \mathcal{C}^{\prime} is a chain. Definition 1 . ( [ 2 ] ) A fully leafed induced subtree of order n of a graph G is an induced subtree of G that maximizes the number of leaves among all induced subtrees of order n in G . We want to describe the graph structure of fully leafed induced subtrees in P2-graphs. A Penrose P2 tiling, shortly a P2 tiling, is a tiling of the plane constructed by two sorts of tiles, called prototiles , described in Figure 1 . P2 tilings are aperiodic. We call P2-graph the dual graph of a P2 tiling. So we often use the term tile to denote a vertex of a P2 -graph. All graphs considered in this paper are induced subgraphs of P2 -graphs, so we may not mention it every time. (a) Kite (b) Dart Figure 1 : Prototiles of P2 tilings. Tiles are edge-adjacent only when black (resp. white) vertices coincide. \varphi=\frac{1+\sqrt{5}}{2} is the golden ratio, giving the ratio of long to short edges in both types of tile. A patch in a P2 tiling is a connected set of tiles. Two important patches of P2 tilings will be used throughout this paper: the star and the sun. They are presented in Figure 2 . (a) Star (b) Sun Figure 2 : Two P2 patches An important concept of Penrose tilings is inflation. Consider a kite (resp. a dart), that we have outlined by a blue dashed contour in Figure 3(a) (resp. 3(b) ). The inflation of this kite (resp. this dart) is the patch of 4 tiles (resp. 3 tiles) that are \varphi times smaller than the original tiles, represented in brown and orange in Figure 3(a) (resp. 3(b) ). (a) Inflation of a kite (b) Inflation of a dart Figure 3 : Inflation of P2 tiles By doing an infinite number of inflations from a given patch and by rescaling the dimension at each inflation, we can construct P2 tilings. A P2-graph is the dual graph of a Penrose tiling. 3 Graph structure It can be shown that every fully leafed induced subtree of a P2-graph that has 8 internal tiles (i.e. tiles of degree 2 or more) or fewer is a subcaterpillar of a caterpillar shown in Figure 4 [ 10 ] . We call a prime caterpillar such a fully leafed induced subtree of 8 internal tiles. We notice that internal tiles of prime caterpillars are all of degree 3. We can easily verify that a fully leafed induced subtree of 9 internal tiles or more must have at least one tile of degree 2. One could verify that an induced subtree of a P2-graph is fully leafed when its number of degree-3 tiles is maximal, or equivalently, when its number of degree-2 tiles is minimal, because a P2 tile of a tree has degree 3 or less [ 9 ] [ 10 ] [ 3 ] . Since prime caterpillars are the largest fully leafed induced subtrees with all internal tiles of degree 3, in order to construct a fully leafed induced subtree, we want to maximize the number of prime caterpillars in it. Thus, prime caterpillars are primitive structures that we want to join together to form fully leafed induced subtrees. (a) PC_{1} (b) PC_{2} (c) PC_{3} (d) PC_{4} (e) PC_{5} (f) PC_{6} Figure 4 : The six possible prime caterpillars, up to isometry and choice of leaves. Light green tiles are choices for leaves. Definition 2 ( [ 2 ] and [ 10 ] ) . Let I_{1} , I_{2} be two fully leafed induced subtrees and t a tile such that t is a leaf of I_{1} and I_{2} , I_{1}\cup I_{2} is a fully leafed induced subtree and I_{1}\cap I_{2}=\{t\} . Then we say that I=I_{1}\cup I_{2} is a grafting of I_{1} and I_{2} at t and we write I=I_{1}\diamond_{t}I_{2} , or just I=I_{1}\diamond I_{2} . There are two possibilities for extending a prime caterpillar PC_{k} into a bigger fully leafed induced subtree. We can extend it into a bigger caterpillar or into a tree that is not a caterpillar. In the second case, the fully leafed induced subtree grafted to PC_{k} is called an appendix of PC_{k} . We can prove the following results. Lemma 1 ( [ 3 ] ) . Let \mathcal{C}=A\diamond PC_{k} be a fully leafed induced subtree obtained by grafting the appendix A to the prime caterpillar PC_{k} . Then A has at most 2 internal tiles. Theorem 1 ( [ 3 ] ) . Every fully leafed induced subtree of a P2-graph belongs to one of the following structures: 1. A subcaterpillar of a prime caterpillar; 2. A caterpillar obtained by grafting several prime caterpillars with at most one proper subcaterpillar of a prime caterpillar structure; 3. A caterpillar obtained by grafting several prime caterpillars, to which an appendix of at most 2 internal tiles and 4 leaves is grafted. Thus, in order to study fully leafed induced subtrees in P2 tilings, it is sufficient to study caterpillars obtained by grafting prime caterpillars, since any fully leafed induced subtree can be obtained from such a caterpillar by adding a small number of tiles. Caterpillars obtained by grafting prime caterpillars are precisely the fully leafed induced subtrees that have the property of saturation [ 3 ] . Let us explain what it means. The leaf function L_{P2}(n) gives the number of leaves of a fully leafed induced subtree of order n . We give here an exact expression for L_{P2}(n) that corrects the one given in [ 10 ] . Proposition 1 ( [ 3 ] ) . L_{P2}(n)=\begin{cases}0&\text{if }0\leq n\leq 1,\\ \left\lfloor\dfrac{n}{2}\right\rfloor+1&\text{if }2\leq n\leq 18,\\ 8\left\lfloor\dfrac{n}{17}\right\rfloor+\left\lfloor\dfrac{n\bmod 17}{2}\right\rfloor+1+\mathbbm{1}(n\bmod 17=1)&\text{if }n\geq 19.\end{cases} Let \overline{L_{P2}} be the lowest linear function that upperbounds L_{P2} . We say that a fully leafed induced subtree of order n is saturated if L_{P2}(n)=\overline{L_{P2}}(n) [ 1 ] . 4 Prime caterpillar graftings We study more precisely here how prime caterpillars can be grafted together. The only two possibilities for the grafting of two prime caterpillars are illustrated in Figure 5 [ 3 ] . (a) (b) Figure 5 : The only two derived paths of graftings of two prime caterpillars. Light brown tiles are possibilities for degree-3 tiles. Blue tiles have degree 2 . We add a segment that joins the centers of the stars adjacent to the prime caterpillars. Figure 6 : A region of an HBS tiling superimposed on a P2 tiling. Vertices at the center of a P2 star are color-coded red, green and blue depending on whether the star is adjacent to 0, 1, or 2 suns, respectively. For a given P2 tiling T , if we connect the centers of the stars in T with edges as in Figure 5 , we obtain a new graph that we call a star-graph . The tilings with vertices and edges of the star-graphs are precisely the HBS tilings [ 11 ] . We illustrate a region of an HBS tiling in Figure 6 . A vertex of a star-graph is colored red, green or blue if it is the vertex in the center of a P2 star adjacent to 0, 1 or 2 suns, respectively. Then, the problem of finding fully leafed caterpillars in P2-graphs becomes the problem of finding paths in star-graphs that determine fully leafed caterpillars. For \mathcal{C} a fully leafed caterpillar obtained by grafting prime caterpillars, the path p of \mathcal{C} in the star-graph can be extended into an acyclic path P in the star-graph that splits the Euclidian plane into two distinct sides. If a prime caterpillar C in \mathcal{C} is such that the internal tiles of C^{\prime\prime} all lie on the same side s of P , then we say that C lies on side s of p , since the choice of P from p does not matter here. Proposition 2 ( [ 3 ] ) . If \mathcal{C} is a fully leafed caterpillar, for C_{j} and C_{j+1} two grafted prime caterpillars in \mathcal{C} , C_{j} and C_{j+1} are on different sides of the star-graph path of \mathcal{C} . We are now interested in the angle formed by a prime caterpillar in a star-graph. Let C_{1},C_{2},C_{3} be three prime caterpillars such that C_{1}\diamond C_{2}\diamond C_{3} defines two incident edges E_{1},E_{2} in a star-graph. From Proposition 2 we know that C_{1} and C_{3} are on the same side of a star-graph path that contains the two edges E_{1},E_{2} and that divides the Euclidian plane into two distinct regions. We define the angle \alpha of C_{2} as the angle determined by E_{1},E_{2} and measured on the same side as C_{2} . Since we know that E_{1} and E_{2} do not depend on C_{1} and C_{3} (see Figure 5 ), we can measure the angle of any prime caterpillar, and not only prime caterpillars that are grafted to two prime caterpillars. In the next proposition we see that there are precisely three possible measures of the angle of a prime caterpillar. Proposition 3 ( [ 3 ] ) . The angle measured in radians of the prime caterpillars PC_{1} , PC_{3} and PC_{6} is \frac{4\pi}{5} , the angle of PC_{2} and PC_{5} is \frac{6\pi}{5} and the angle of PC_{4} is \frac{8\pi}{5} . We claim that if the star-graph path of a fully leafed saturated caterpillar \mathcal{C} contains at least one pair of edges forming an angle of \frac{8\pi}{5} , then this star-graph path uniquely determines the saturated caterpillar \mathcal{C} . Indeed, an angle of measure \frac{8\pi}{5} has the conjugate angle \frac{2\pi}{5} . Since there exists no prime caterpillar with an angle of \frac{2\pi}{5} , an angle measuring \frac{8\pi}{5} on one side and \frac{2\pi}{5} on the other side can correspond only to a prime caterpillar forming an angle of \frac{8\pi}{5} . The only possible prime caterpillar for such an angle is PC_{4} . By Proposition 2 , the rest of the star-graph path of \mathcal{C} completely determines the saturated caterpillar \mathcal{C} . Hence, there is no ambiguity in using a star-graph path containing at least one angle that measures \frac{8\pi}{5} on one side to represent a fully leafed saturated caterpillar. (a) Sail (b) Hull (c) Dock 1 (d) Dock 2 (e) Dock 3 (f) Dock 4 (g) Cape 1 (h) Cape 2 (i) Cape 3 (j) Cape 4 Figure 7 : Sea caterpillars (up to rotation). We keep the color code of the previous figures. Just as we did with prime caterpillars, we find paths in the star-graphs that are primitives blocks to construct bigger fully leafed caterpillars. We call the caterpillars corresponding to these paths sea caterpillars , and we present them in Figure 7 . 5 Bi-infinite fully leafed caterpillars Since we now know how to graft caterpillars to construct large fully leafed caterpillars, the natural question that comes to mind is to ask if there are infinite such caterpillars. Since a caterpillar can be extended in two directions, we are interested in finding bi-infinite fully leafed caterpillars. A first one was found in [ 10 ] , and we formalise its construction in [ 3 ] . In [ 10 ] , it is conjectured that this bi-infinite caterpillar is the unique bi-infinite fully leafed caterpillar of P2 -graphs. To approach this conjecture, we focus on results that allow us to identify caterpillars that cannot be extended to bi-infinite fully leafed caterpillars. Lemma 2 ( [ 3 ] ) . There exists no bi-infinite fully leafed caterpillar that contains two consecutive \frac{4\pi}{5} angles. Proposition 4 ( [ 3 ] ) . The prime caterpillar PC_{1} cannot belong to any bi-infinite fully leafed caterpillar. Proposition 5 ( [ 3 ] ) . Neither a cape 2 nor a cape 3 can belong to a bi-infinite fully leafed caterpillar. Other similar results, in particular about graftings of sea caterpillars, can be found in [ 3 ] . These results allow us to construct a new bi-infinite fully leafed caterpillar and refute the conjecture claiming the unicity of the bi-infinite fully leafed caterpillar presented in [ 10 ] . Theorem 2 ( [ 3 ] ) . There exist a bi-infinite fully leafed caterpillar that contains a cape 4. Proof. The proof sketch is as follows. We consider an extension of a cape 4, that we call a Super cape 4 , which is illustrated in Figure 8(a) . By inflating the region in which it is embedded, we obtain a new region of a P2 tiling where we can construct a larger fully leafed caterpillar that contains a Super Cape 4, which we denote by \psi(\text{Super Cape 4}) and show in Figure 8(b) . We then proceed to find similar constructions for the new caterpillars in \psi(\text{Super Cape 4}) , and associate them with larger caterpillars in the third inflation of the tiling region where they are embedded. There are 8 such constructions to find, and they must be defined so that for all n\in\mathbb{N} , \psi^{n}(\text{Super cape 4}) remains connected and acyclic. (a) Super cape 4 (b) \psi(\text{Super cape 4}) Figure 8 : Super cape 4 and \psi(\text{Super cape 4}) . In Figure (b), the black path is the star-graph chain of a Super cape 4 before applying 3 inflations and is used as a reference in the tiling and the purple path is the star-graph chain of \psi(\text{Super cape 4}) . ∎ 6 Conclusion We now have a better understanding of the structure of fully leafed induced subtrees in P2 tilings. By Theorem 1 , they are caterpillars, up to an appendix of at most six tiles, and the saturated ones are caterpillars. We disproved the conjecture from [ 10 ] that there is a unique bi-infinite fully leafed caterpillar in P2 tilings. Since we now know that there exists more than one bi-infinite fully leafed caterpillar in Penrose P2 tilings, we aim to find them all. We have already eliminated several patterns. It is also possible to consider the star-graph chains of these bi-infinite structures as words over a 3-letter alphabet whose letters are either the colors of the vertices in these chains (red, green, and blue) or the angles of the prime caterpillars, and to study the properties of these words. We have begun this work in [ 3 ] . Of course, it would also be interesting to describe the fully leafed induced subtrees in other aperiodic tilings. Acknowledgements: We thank Kevin Bertman for developing the website https://www.mrbertman.com/penroseTilings.html , which we used to construct tile patches of P2 tilings. *