Randomized Algorithms for Higher-Order Voronoi Diagrams

39
Maksym Zavershynskyi 1 joint work with Cecilia Bohler 2 , Chih-Hung Liu 2 and Evanthia Papadopoulou 1 1 University of Lugano, Faculty of Informatics, CH-6904 Lugano, Switzerland 2 University of Bonn, Institute of Computer Science I, D-53113 Bonn, Germany This work was supported by the European Science Foundation (ESF) in the EUROCORES collaborative research project EuroGIGA/VORONOI, SNF 20GG21-134355. The work of the 1rd and 4rth author was supported in part by the Swiss National Science Foundation grant 200020-149658. Randomized Algorithms for Higher-Order Voronoi Diagrams

description

One of many important generalizations of ordinary Voronoi diagrams is the higher-order Voronoi diagram. The order-k Voronoi diagram is the partitioning of the plane into regions, such that each point within a fixed region has the same k nearest sites. Many algorithms have been developed that construct the higher-order Voronoi diagram of point-sites. In this talk we will discuss randomized algorithms that can be used for a larger class of sites—specifically, polygonal objects and the abstract setting. We describe the algorithms in combinatorial rather than geometric terms, which makes it possible to construct higher-order Voronoi diagrams that have bisectors satisfying certain combinatorial properties.

Transcript of Randomized Algorithms for Higher-Order Voronoi Diagrams

Page 1: Randomized Algorithms for Higher-Order Voronoi Diagrams

Maksym Zavershynskyi1joint work with

Cecilia Bohler2, Chih-Hung Liu2 and Evanthia Papadopoulou1

1University of Lugano, Faculty of Informatics, CH-6904 Lugano, Switzerland2University of Bonn, Institute of Computer Science I, D-53113 Bonn, Germany

This work was supported by the European Science Foundation (ESF) in the EUROCORES collaborative research project EuroGIGA/VORONOI, SNF 20GG21-134355. The work of the 1rd and 4rth author was supported in part by the Swiss National Science Foundation grant 200020-149658.

Randomized Algorithmsfor Higher-Order Voronoi Diagrams

Page 2: Randomized Algorithms for Higher-Order Voronoi Diagrams

Nearest Neighbor Voronoi Diagram

The nearest neighbor Voronoi diagram is the partitioning of the plane into regions, such that all points within a region have the same closest site.

Page 3: Randomized Algorithms for Higher-Order Voronoi Diagrams

Higher Order Voronoi Diagram

The order-k Voronoi diagram is the partitioning of the plane into regions, such that all points within an order-k region have the same k nearest sites.

VR2({p, q}, S)

p q

[Bohler et al.’13, Lee’82, Papadopoulou&Zavershynskyi’12]

Page 4: Randomized Algorithms for Higher-Order Voronoi Diagrams

Abstract Voronoi Diagrams

Defined by bisecting system rather than concrete geometric sites or distance measures.

J(p, q)D(p, q)

D(q, p)

[Bohler et al.’13, Klein et al.’09, Klein et al.’93]

Page 5: Randomized Algorithms for Higher-Order Voronoi Diagrams

Abstract Voronoi Diagrams

First-order Voronoi region:

VR1({p}, S) =�

q∈S,q �=p

D(p, q)

For each

A1) is pathwise connectedA2) Each point in belongs to some A3) is not emptyA4) is unbounded Jordan-curveA5) and have finitely many intersection points that are transversal.

S� ⊆ S

VR1(p, S�)

VR1(p, S�)R2

VR1(p, S�)

J(p, q)

J(p, q) J(s, t)

[Bohler et al.’13, Klein et al.’09, Klein et al.’93]

Page 6: Randomized Algorithms for Higher-Order Voronoi Diagrams

Abstract Voronoi Diagrams

Order-k Voronoi region:

VRk(H,S) =�

p∈H,q∈S\H

D(p, q)

• Points in convex distance metrics, Karlsruhe metric

• Disjoint line segments in Lp

• Disjoint convex polygons in Lp

Structural complexity is

[Bohler et al.’13, Klein et al.’09, Klein et al.’93]

≤ 2k(n− k)

Page 7: Randomized Algorithms for Higher-Order Voronoi Diagrams

Construction Algorithms

Construction Time Reference

Chazelle&Edelsbrunner’87

Clarkson’87

Aurenhammer’90

Mulmuley’91

Boissonnat et al.’93

Agarwal et al.’98

Chan’98

Ramos’99

O�n2 + b log2 n

O�n1+�

k�

O�nk

2 + n log n�

O�nk

3 + n log n�

O�n log3 n+ nk log n

O (n log n+ nk log n)

O

�n log n+ nk2c log

∗ k�

O�nk

2 log n�

Page 8: Randomized Algorithms for Higher-Order Voronoi Diagrams

Our Results

We Present 3 Algorithms for Abstract Diagrams

1.Randomized Divide-n-Conquer Algorithm

• Expected time complexityfor any constant

• Based on Clarkson-Shor technique and Clarkson’s algorithm for points.

• Uses two other algorithms for subroutines.

2.Traversal-based Algorithm

3. Iterative Algorithm

O(kn1+ε)ε > 0

Page 9: Randomized Algorithms for Higher-Order Voronoi Diagrams

Construction Algorithms

O(kn1+ε)

O(k2n log n)O(n22α(n) log n)

Randomized divide-n-conquer algorithm

Iterative algorithmTraversal-based algorithm

uses

uses

good for large k good for small k

Page 10: Randomized Algorithms for Higher-Order Voronoi Diagrams

Additional Axiom

Additional Axiom:

A6) Each bisector has constant number of points of vertical tangencies.

Page 11: Randomized Algorithms for Higher-Order Voronoi Diagrams

Divide-n-Conquer Algorithm

Page 12: Randomized Algorithms for Higher-Order Voronoi Diagrams

Vertical Decomposition

Vertical decomposition of :

1.Consider

2. Shoot vertical rays from each vertex and vertical tangent point.

Vk(S)

Vk(S)

Vk(S) ∪ Vk+1(S)

Page 13: Randomized Algorithms for Higher-Order Voronoi Diagrams

Vertical Decomposition

Vertical decomposition of :

1.Consider

2. Shoot vertical rays from each vertex and vertical tangent point.

Vk(S)

Vk(S) ∪ Vk+1(S)

Vk(S) ∪ Vk+1(S)

Page 14: Randomized Algorithms for Higher-Order Voronoi Diagrams

Vertical Decomposition

Vertical decomposition of :

1.Consider

2. Shoot vertical rays from each vertex and vertical tangent point.

Vk(S)

Vk(S) ∪ Vk+1(S)

Vk(S) ∪ Vk+1(S)

Page 15: Randomized Algorithms for Higher-Order Voronoi Diagrams

Random Sampling

Draw random sample such that it is a good estimator of .

R

S

Trapezoid is defined by at most elements of Rd

e.g.: d = 5s1

s2

s3

s4

p - dominatorp�

[Clarkson’87]

Page 16: Randomized Algorithms for Higher-Order Voronoi Diagrams

Random Sampling

Site strongly conflicts with if

d = 5s1

s2

s3

s4

p - dominatorp�

s � � ⊂ D(s, p)

D(s, p)

Page 17: Randomized Algorithms for Higher-Order Voronoi Diagrams

Random Sampling

Site weakly conflicts with if

d = 5s1

s2

s3

s4

p - dominatorp�

s �

D(s, p)

�∩D(s, p) �= ∅

Page 18: Randomized Algorithms for Higher-Order Voronoi Diagrams

Lemma 1

Lemma 1

For a random sample , with probability at least

1) number of strong conflicts with is

2) number of weak conflicts with is

for each trapezoid in vert. decomposition of�

R 1/2

S

≤ α|S|S

≥ |S|/(r − 5)

r = |R|,α = O(log r),β = O(log r/ log log r)

Allows to efficiently “bracket” the space!

[Clarkson’87]

Vβ(R)

Page 19: Randomized Algorithms for Higher-Order Voronoi Diagrams

Lemma 2

r = |R|,α = O(log r),β = O(log r/ log log r)[Clarkson’87]

- set of sitesS

Vk(S) - order-k diagramv - vertex

v

Page 20: Randomized Algorithms for Higher-Order Voronoi Diagrams

Lemma 2

r = |R|,α = O(log r),β = O(log r/ log log r)[Clarkson’87]

- set of sitesS

- random sampleR ⊂ S

where each in vert. decomp. of has strong conflicts

Vβ(R)

Vk(S) - order-k diagram

≥ k

v - vertex

v

Page 21: Randomized Algorithms for Higher-Order Voronoi Diagrams

Lemma 2

r = |R|,α = O(log r),β = O(log r/ log log r)[Clarkson’87]

- set of sitesS

- random sampleR ⊂ S

where each in vert. decomp. of has strong conflicts

Vβ(R)

Vk(S) - order-k diagram

v - vertex

S� ⊂ S- sites in weak conflictVk(S

�)then is vertex ofv

v

Allows to perform divide-n-conquer!

≥ k

Page 22: Randomized Algorithms for Higher-Order Voronoi Diagrams

Divide-n-Conquer Algorithm

Construction of can be reduced to finding all the vertices of .

Vk(S)

Vk(S)

[Clarkson’87]

Page 23: Randomized Algorithms for Higher-Order Voronoi Diagrams

Divide-n-Conquer Algorithm

If then use traversal-based algorithm.

Else choose “good” random sample .

• Construct and the decomposition, using iterative algorithm.

• For each trapezoid in the decomposition recursively compute vertices in .

R

|S| ≤ k(r − 5)

Vβ(R)

�Vk(S

�)

where - the sites in weak conflict with .S� ⊂ S �[Clarkson’87]

Page 24: Randomized Algorithms for Higher-Order Voronoi Diagrams

Traversal-Based Algorithm

Page 25: Randomized Algorithms for Higher-Order Voronoi Diagrams

Influence Region

Influence region of site - union of all order-k Voronoi regions induced by the site .

s

s

Vk(S)

influence region of site s [Papadopoulou&Zavershynskyi, Bohler et al.’13]

Page 26: Randomized Algorithms for Higher-Order Voronoi Diagrams

Traversal-Based Algorithm

We can construct influence region of sitein time , using Har-Peled’s random walk method.

s

O(n2α(n) log n)

Vk(S)

influence region of site s [Har-Peled’00]

Page 27: Randomized Algorithms for Higher-Order Voronoi Diagrams

Traversal-Based Algorithm

Union of all influence regions is the order-k Voronoi diagram.

Total running time: O(n22α(n) log n)

[Har-Peled’00, Chazelle&Edelsbrunner’87]

Page 28: Randomized Algorithms for Higher-Order Voronoi Diagrams

Iterative Algorithm

Page 29: Randomized Algorithms for Higher-Order Voronoi Diagrams

Iterative Algorithm

Order-k Voronoi diagram can be computed from the order-(k-1) Voronoi diagram in expected time.O(k(n− k) log n)

Total running time: O(k2n log n)

[Bohler et al.’13, Lee’82, Papadopoulou&Zavershynskyi, Klein et al.’93]

Page 30: Randomized Algorithms for Higher-Order Voronoi Diagrams

SUMMARY

We have developed 3 randomized algorithms for abstract higher-order Voronoi diagrams.

1.Divide-n-conquer algorithm with time complexity.

2.Traversal-based algorithm with time complexity. Good for large orders.

3.Iterative algorithm with time complexity. Good for small orders.

O(kn1+ε)

O(n22α(n) log n)

O(k2n log n)

Page 31: Randomized Algorithms for Higher-Order Voronoi Diagrams

Thank you!

Page 32: Randomized Algorithms for Higher-Order Voronoi Diagrams

References

1) C. Bohler, P. Cheilaris, R. Klein, C.-H. Liu, E. Papadopoulou, and M. Zavershynskyi. On the complexity of higher order abstract Voronoi diagrams. 40th International Colloquium on Automata, Languages and Programming (ICALP’13), pp. 208–219, 2013.

2) K. L. Clarkson. New applications of random sampling in computational geometry. Discrete and Computational Geometry 2(1), pp. 195–222, 1987.

3) S. Har-Peled. Taking a walk in a planar arrangment. SIAM Journal on Computing 30(4), pp. 1341-1367, 2000.

4) D. T. Lee. On k Nearest Neighbor Voronoi Diagrams in the Plane. IEEE Trans. Computers 31(6), pp. 478-487, 1982.

5) E. Papadopoulou and M. Zavershynskyi. On Higher Order Voronoi Diagrams of Line Segments. 23rd International Symposium on Algorithms and Computation (ISAAC ’12), LNCS 7676, pp. 177–186, 2012.

6) R. Klein, E. Langetepe, and Z. Nilforoushan. Abstract Voronoi Diagrams Revisited. Computational Geometry: Theory and Applications 42(9), pp. 885-902, 2009.

7) R. Klein, K. Mehlhorn, and St. Meiser. Randomized Incremental Construction of Abstract Voronoi Diagrams. Computational Geometry: Theory and Applications 3(1), pp. 157–184 1993.

8) B. Chazelle and H. Edelsbrunner. An improved algorithm for constructing kth-order Voronoi Diagram. IEEE Transactions on Computers 36(11), pp. 1349–1454, 1987.

Page 33: Randomized Algorithms for Higher-Order Voronoi Diagrams

Misc

projection

Page 34: Randomized Algorithms for Higher-Order Voronoi Diagrams

Misc

planes that correspond to extremal positions of β-sets

Page 35: Randomized Algorithms for Higher-Order Voronoi Diagrams

Misc

planes that correspond to extremal positions of β-sets

Page 36: Randomized Algorithms for Higher-Order Voronoi Diagrams

Misc

planes that correspond to extremal positions of β-sets

normals

Page 37: Randomized Algorithms for Higher-Order Voronoi Diagrams

Misc

planes that correspond to extremal positions of β-sets

normals

Page 38: Randomized Algorithms for Higher-Order Voronoi Diagrams

Misc

planes that correspond to extremal positions of β-sets

normals

triangulation of the cone

Page 39: Randomized Algorithms for Higher-Order Voronoi Diagrams

Misc

normals

points that belong to the union of the halfplanes - analog of weak conflictpoints that belong to the intersection of the halfplanes - analog of strong conflict