Abstract 1 Introduction 2 Our Results 3 Conclusion References

Reeb Lobsters Are 1-Planar

Maarten Löffler ORCID Department of Information and Computing Sciences, Utrecht University, The Netherlands Miriam Münch ORCID Faculty of Computer Science and Mathematics, University of Passau, Germany Ignaz Rutter ORCID Faculty of Computer Science and Mathematics, University of Passau, Germany
Abstract

Very recently, Chambers, Fasy, Hosseini Sereshgi and Löffler [3] showed that every Reeb caterpillar admits a crossing-free drawing. It turns out that this does not hold for Reeb lobsters but we show that these graphs admit drawings with at most one crossing per edge.

Keywords and phrases:
Reeb graphs, layered drawings, local crossing number
Category:
Poster Abstract
Copyright and License:
[Uncaptioned image] © Maarten Löffler, Miriam Münch, and Ignaz Rutter; licensed under Creative Commons License CC-BY 4.0
2012 ACM Subject Classification:
Mathematics of computing Graphs and surfaces
; Mathematics of computing Topology
Funding:
Funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – 541433306.
Editors:
Vida Dujmović and Fabrizio Montecchiani

1 Introduction

A layered graph (G,h) is a graph together with pairwise-distinct preassigned heights of its vertices. In a drawing of a layered graph each vertex v is mapped to a unique point (x,h(v))2 and every edge is mapped to a continuous y-monotone curve [1, 7]. For a vertex v in a layered graph we denote the horizontal line through v by v. Let deg(v) denote the number of neighbors w of v with h(w)<h(v). Similarly we let deg(v) denote the number of neihgbors w of v with h(w)>h(v). A Reeb graph (G,h), introduced in [6], is a layered graph such that for every vertex v we have deg(v){1,3}, deg(v)2 and deg(v)2. Reeb graphs can be computed effieciently [4, 5] and capture the evolution of level sets of functions from a topological space to the real numbers. They have many applications in topological data analysis and visualization [8, 2].

A caterpillar is a tree C such that removing all leaves from C yields a path, called the spine of C. A tree L such that removing all leaves yields a caterpillar is a lobster. The spine of a lobster L is the spine of the caterpillar we obtain by removing all leaves from L. The local crossing number lcr(G) of a (layered) graph G is the smallest integer k such that G admits a drawing with at most k crossings per edge. A graph G is k-planar if lcr(G)k.

2 Our Results

First we show that crossings are needed to draw Reeb lobsters.

Theorem 1.

Not every Reeb lobster admits a crossing-free drawing.

Proof.

Consider the lobster G shown in Figure 1(a). Assume that G admits a crossing-free drawing Γ. Let G be the graph we obtain from G by adding two new independent vertices x,y with h(x)>h(v) and h(y)<h(v) for every vV such that x is adjacent exactly to the three vertices with greatest heights and y is adjacent exactly to the three vertices with lowest heights; see Figure 1(b). Then it is easy to extend Γ to a crossing-free drawing of G but G is a K3,3-subdivision and thus non-planar; a contradiction.

Figure 1: (a) A Reeb lobster and (b) its extension to a K3,3-subdivision.

Next we show that the local crossing-number of a layered lobster may be arbitrarily large.

Theorem 2.

For every k there is a layered lobster L such that lcr(L)>k.

Proof.

Figure 2: Illustration of the proof of Theorem 2.

For a fixed k consider the lobster L sketched in Figure 2(a). We call a set F of l edges an l-fan if all edges in F share a common endpoint c, the remaining endpoints have degree 1 and the heights of these degree-1 endpoints lie either all above h(c) or all below h(c). For each of the four fans in L the heights of the corresponding degree-1 vertices are chosen such that each other vertex either lies below all of them or above all of them. Assume for the sake of contradiction that there exists a drawing Γ of L with at most k crossings per edge. Let u be a degree-1 vertex adjacent to u, whose incident edge does not cross an edge incident to z; such a vertex exists as there are 3k+1 degree-1 vertices adjacent to u and the edges incident to z have at most 3k crossings. Analogously, there exists a degree-1 vertex w adjacent to w with h(w)>h(w), whose incident edge does not cross an edge incident to z. Similarly, we can find sets V′′ and W′′ each consisting of 2k+1 degree-1 vertices adjacent to v and w, respectively, that lie below w and whose incident edges do not cross an edge incident to z. Let Γ denote the drawing induced by {u,u,v,w,w,z}V′′W′′, where we possibly reroute zu and zw so that they do not cross each other. Note that all edges incident to z are crossing-free in Γ. Consider the edges e=uu and f=ww. Without loss of generality assume that zu crosses w left of w. Since the path uzw is crossing-free, f crosses z to the right of z and since zv is crossing-free, f crosses v to the right of v. We distinguish two cases based on whether e crosses v on the left or on the right side of v.

Assume that e crosses v on the left side of v; see Figure 2(b). As v lies between e and f on v and the edges incident to z are crossing-free in Γ, each edge connecting v to V′′ crosses e or f. Since |V′′|=2k+1, e or f has more than k crossings; a contradiction.

It remains to consider the case where e crosses v on the right side of v; see Figure 2(c). Note that in this case e crosses the horizontal line w to the right of w since the path wzv is uncrossed. Since the path zuu starts and ends above w, crosses w once to the left of w and once to the right of w, and zu is crossing-free, all edges connecting w to W′′ cross e. Since |W′′|=2k+1 it follows that e has more than k crossings; a contradiction.

Finally we show that Reeb lobsters admit a drawing with at most one crossing per edge.

Theorem 3.

Every Reeb lobster is 1-planar.

Proof.

Let L be a Reeb graph. We consider an extended spine of L; i.e, a path between degree-1 vertices of L that contains the spine. Let {v0,,vk} be an extended spine of L such that for every 0<i<k the two neighbors of vi are vi1 and vi+1. We start by drawing the extended spine x-monotone; see Figure 3 for an illustration. Next we insert all vertices of L that are adjacent to the extended spine crossing-free by assigning each such vertex u an x-coordinate close to the x-coordinate of its neighbor v on the extended spine. It remains to add the missing leaves of L that have a neighbor that is not adjacent to a vertex on the extended spine. Let w be such a leaf of L with neighbor u. Let vi be the neighbor of u on the extended spine. If h(vi) does not lie between h(w) and h(u), then we can easily insert w crossing-free by assigning it an x-coordinate close to or equal to the x-coordinate of u. This does not result in a crossing. Hence assume that h(w)<h(vi)<h(u) or h(w)>h(vi)>h(u). Now we assign w an x-coordinate between the x-coordinates of vi and vi1. Then wu crosses the edge vivi1. Hence by construction for every such leaf w, the edge wu receives at most one crossing. Recall that deg(u)2, deg(u)2. Thus every u that is neither a leaf of L nor part of the extended spine has at most one neighbor w with h(w)<h(vi)<h(u) or h(w)>h(vi)>h(u). Hence for every 0<i<k also vivi1 receives at most one crossing. Thus the construction described above yields a 1-planar drawing of L.

Figure 3: Illustration of the construction of a 1-planar drawing of a lobster that is a Reeb graph.

3 Conclusion

Our results show that the property to be a Reeb graph yields an upper bound of 1 for the otherwise unbounded local crossing number of layered lobsters. An interesting question is how this generalizes to more general Reeb trees.

References

  • [1] Oliver Bastert and Christian Matuszewski. Layered Drawings of Digraphs, pages 87–120. Springer Berlin Heidelberg, Berlin, Heidelberg, 2001. doi:10.1007/3-540-44969-8_5.
  • [2] Silvia Biasotti, Daniela Giorgi, Michela Spagnuolo, and Bianca Falcidieno. Reeb graphs for shape analysis and applications. Theoretical computer science, 392(1-3):5–22, 2008. doi:10.1016/j.tcs.2007.10.018.
  • [3] Erin W. Chambers, Brittany Terese Fasy, Erfan Hosseini Sereshgi, and Maarten Löffler. Drawing reeb graphs. In Combinatorial Algorithms - 36th International Workshop, IWOCA 2025, volume 15885 of Lecture Notes in Computer Science, pages 58–71. Springer, 2025. doi:10.1007/978-3-031-98740-3_5.
  • [4] William Harvey, Yusu Wang, and Rephael Wenger. A randomized O(mlogm) time algorithm for computing Reeb graphs of arbitrary simplicial complexes. In Proceedings of the 2010 annual symposium on Computational geometry, SoCG ’10, pages 267–276. ACM, 2010. doi:10.1145/1810959.1811005.
  • [5] Salman Parsa. A deterministic O (m log m) time algorithm for the reeb graph. Discret. Comput. Geom., 49(4):864–878, 2013. doi:10.1007/s00454-013-9511-3.
  • [6] Georges Reeb. Sur les points singuliers d’une forme de pfaff completement integrable ou d’une fonction numerique [on the singular points of a completely integrable pfaff form or of a numerical function]. Comptes Rendus Acad. Sciences Paris, 222:847–849, 1946. URL: https://cir.nii.ac.jp/crid/1571417125676878592.
  • [7] Kozo Sugiyama, Shôjirô Tagawa, and Mitsuhiko Toda. Methods for visual understanding of hierarchical system structures. IEEE Transactions on Systems, Man, and Cybernetics, SMC, 11(2):109–125, 1981. doi:10.1109/TSMC.1981.4308636.
  • [8] Lin Yan, Talha Bin Masood, Raghavendra Sridharamurthy, Farhan Rasheed, Vijay Natarajan, Ingrid Hotz, and Bei Wang. Scalar field comparison with topological descriptors: Properties and applications for scientific visualization. In Computer Graphics Forum, volume 40, pages 599–633. Wiley Online Library, 2021. doi:10.1111/cgf.14331.