Let \(\mathcal D\) be a set of disks and G be the intersection graph of \(\mathcal D\). A drawing of G is obedient to \(\mathcal D\) if every vertex is placed in its corresponding disk. We show that deciding whether a set of unit disks \(\mathcal D\) has an obedient plane straight-line drawing is \(\mathcal {NP}\)-hard regardless of whether a combinatorial embedding is prescribed or an arbitrary embedding is allowed. We thereby strengthen a result by Evans et al., who show \(\mathcal {NP}\)-hardness for disks with arbitrary radii in the arbitrary embedding case. Our result for the arbitrary embedding case holds true even if G is thinnish, that is, removing all triangles from G leaves only disjoint paths. This contrasts another result by Evans et al. stating that the decision problem can be solved in linear time if \(\mathcal D\) is a set of unit disks and G is thin, that is, (1) the (graph) distance between any two triangles is larger than 48 and (2) removal of all disks within (graph) distance 8 of a triangle leaves only isolated paths. A path in a disk intersection graph is isolated if for every pair A, B of disks that are adjacent along the path, the convex hull of \(A\cup B\) is intersected only by disks adjacent to A or B. Our reduction can also guarantee the triangle separation property (1). This leaves only a small gap between tractability and \(\mathcal {NP}\)-hardness, tied to the path isolation property (2) in the neighborhood of triangles. It is therefore natural to study the impact of different restrictions on the structure of triangles. As a positive result, we show that an obedient plane straight-line drawing is always possible if all triangles in G are light and the disks are in general position (no three centers collinear). A triangle in a disk intersection graph is light if all its vertices have degree at most three or the common intersection of the three corresponding disks is empty. We also provide an efficient drawing algorithm for that scenario.
[1]
Sándor P. Fekete,et al.
Deterministic boundary recognition and topology extraction for large sensor networks
,
2005,
SODA '06.
[2]
Balázs Keszegh,et al.
Drawing Planar Graphs of Bounded Degree with Few Slopes
,
2013,
SIAM J. Discret. Math..
[3]
Maarten Löffler,et al.
Data Imprecision in Computational Geometry
,
2009
.
[4]
Yan Zhang,et al.
Geometric ad-hoc routing: of theory and practice
,
2003,
PODC '03.
[5]
Jean Cardinal,et al.
Computational Geometry Column 62
,
2015,
SIGACT News.
[6]
Chandrajit L. Bajaj,et al.
The algebraic degree of geometric optimization problems
,
1988,
Discret. Comput. Geom..
[7]
Maarten Löffler,et al.
Recognizing a DOG is Hard, But Not When It is Thin and Unit
,
2016,
FUN.
[8]
David G. Kirkpatrick,et al.
Minimizing Co-location Potential of Moving Entities
,
2016,
SIAM J. Comput..
[9]
Giuseppe Di Battista,et al.
Anchored Drawings of Planar Graphs
,
2014,
Graph Drawing.
[10]
Jie Gao,et al.
Boundary recognition in sensor networks by topological methods
,
2006,
MobiCom '06.
[11]
Ivan Stojmenovic,et al.
Routing with Guaranteed Delivery in Ad Hoc Wireless Networks
,
2001,
Wirel. Networks.
[12]
Mark de Berg,et al.
Optimal Binary Space Partitions for Segments in the Plane
,
2012,
Int. J. Comput. Geom. Appl..
[13]
Roger Wattenhofer,et al.
Unit disk graph approximation
,
2004,
DIALM-POMC '04.