TY - JOUR
T1 - Approximating the Generalized Minimum Manhattan Network Problem
AU - Das, Aparna
AU - Fleszar, Krzysztof
AU - Kobourov, Stephen
AU - Spoerhase, Joachim
AU - Veeramoni, Sankar
AU - Wolff, Alexander
N1 - Funding Information:
A preliminary version of this paper appeared in Proc. 24th International Symposium on Algorithms and Complexity (ISAAC’13), volume 8283 of Lect. Notes Comput. Sci., pp. 722–732. This work was supported by the ESF EuroGIGA project GraDR (DFG Grant Wo 758/5-1).
Publisher Copyright:
© 2017, Springer Science+Business Media New York.
PY - 2018/4/1
Y1 - 2018/4/1
N2 - We consider the generalized minimum Manhattan network problem (GMMN). The input to this problem is a set R of n pairs of terminals, which are points in R2. The goal is to find a minimum-length rectilinear network that connects every pair in R by a Manhattan path, that is, a path of axis-parallel line segments whose total length equals the pair’s Manhattan distance. This problem is a natural generalization of the extensively studied minimum Manhattan network problem (MMN) in which R consists of all possible pairs of terminals. Another important special case is the well-known rectilinear Steiner arborescence problem (RSA). As a generalization of these problems, GMMN is NP-hard. No approximation algorithms are known for general GMMN. We obtain an O(log n) -approximation algorithm for GMMN. Our solution is based on a stabbing technique, a novel way of attacking Manhattan network problems. Some parts of our algorithm generalize to higher dimensions, yielding a simple O(log d + 1n) -approximation algorithm for the problem in arbitrary fixed dimension d. As a corollary, we obtain an exponential improvement upon the previously best O(nε) -ratio for MMN in d dimensions (ESA 2011). En route, we show that an existing O(log n) -approximation algorithm for 2D-RSA generalizes to higher dimensions.
AB - We consider the generalized minimum Manhattan network problem (GMMN). The input to this problem is a set R of n pairs of terminals, which are points in R2. The goal is to find a minimum-length rectilinear network that connects every pair in R by a Manhattan path, that is, a path of axis-parallel line segments whose total length equals the pair’s Manhattan distance. This problem is a natural generalization of the extensively studied minimum Manhattan network problem (MMN) in which R consists of all possible pairs of terminals. Another important special case is the well-known rectilinear Steiner arborescence problem (RSA). As a generalization of these problems, GMMN is NP-hard. No approximation algorithms are known for general GMMN. We obtain an O(log n) -approximation algorithm for GMMN. Our solution is based on a stabbing technique, a novel way of attacking Manhattan network problems. Some parts of our algorithm generalize to higher dimensions, yielding a simple O(log d + 1n) -approximation algorithm for the problem in arbitrary fixed dimension d. As a corollary, we obtain an exponential improvement upon the previously best O(nε) -ratio for MMN in d dimensions (ESA 2011). En route, we show that an existing O(log n) -approximation algorithm for 2D-RSA generalizes to higher dimensions.
KW - Approximation algorithms
KW - Computational geometry
KW - Minimum Manhattan Network
UR - http://www.scopus.com/inward/record.url?scp=85014075563&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85014075563&partnerID=8YFLogxK
U2 - 10.1007/s00453-017-0298-0
DO - 10.1007/s00453-017-0298-0
M3 - Article
AN - SCOPUS:85014075563
SN - 0178-4617
VL - 80
SP - 1170
EP - 1190
JO - Algorithmica
JF - Algorithmica
IS - 4
ER -