Академический Документы
Профессиональный Документы
Культура Документы
Time Complexity
first-order O(n log n)
Time Complexity
first-order O(n log n)
Time Complexity
first-order O(n log n)
shortest path:
shortest path:
walk
shortest path:
walk
quickest path?
shortest path:
walk
quickest path?
walk
shortest path:
walk
quickest path?
walk
shortest path:
walk
quickest path?
walk
shortest path:
walk
quickest path?
walk
walk
shortest path:
walk
L1 metric
quickest path?
walk
walk
city metric
Institute of Theoretical Informatics
Andreas Gemsa, D.T. Lee, Chih-Hung Liu, Dorothea Wagner – Higher Order City Voronoi Diagrams. Prof. Dr. Dorothea Wagner
City Metric
shortest path:
walk
L1 metric
quickest path?
walk
walk
city metric
Institute of Theoretical Informatics
Andreas Gemsa, D.T. Lee, Chih-Hung Liu, Dorothea Wagner – Higher Order City Voronoi Diagrams. Prof. Dr. Dorothea Wagner
City Metric
city metric:
L1 metric and transportation network
shortest path:
walk
L1 metric
quickest path?
walk
walk
city metric
Institute of Theoretical Informatics
Andreas Gemsa, D.T. Lee, Chih-Hung Liu, Dorothea Wagner – Higher Order City Voronoi Diagrams. Prof. Dr. Dorothea Wagner
City Metric
city metric:
L1 metric and transportation network
G
Def.: transportation network
graph G = (VC , EC )
planar, straight-line embedding
only horizontal and vertical edges
G
Def.: transportation network
graph G = (VC , EC )
planar, straight-line embedding
only horizontal and vertical edges
speed off the network: 1
speed on the network: v > 1
G
Def.: transportation network
graph G = (VC , EC )
planar, straight-line embedding
only horizontal and vertical edges
speed off the network: 1
speed on the network: v > 1
G
Def.: transportation network
graph G = (VC , EC )
planar, straight-line embedding
only horizontal and vertical edges
speed off the network: 1
speed on the network: v > 1
c := |VC |
Complexity of G: O(c)
q G
Def.: transportation network
graph G = (VC , EC )
planar, straight-line embedding
only horizontal and vertical edges
speed off the network: 1
speed on the network: v > 1
c := |VC |
p Complexity of G: O(c)
q G
Def.: transportation network
graph G = (VC , EC )
planar, straight-line embedding
only horizontal and vertical edges
L1 speed off the network: 1
speed on the network: v > 1
c := |VC |
p Complexity of G: O(c)
q G
Def.: transportation network
graph G = (VC , EC )
planar, straight-line embedding
only horizontal and vertical edges
L1 speed off the network: 1
speed on the network: v > 1
c := |VC |
p Complexity of G: O(c)
q G
Def.: transportation network
graph G = (VC , EC )
planar, straight-line embedding
only horizontal and vertical edges
L1 speed off the network: 1
speed on the network: v > 1
c := |VC |
p Complexity of G: O(c)
Euclidean-Metric City-Metric
Structural Complexity Structural Complexity
first-order Θ(n) first-order Θ(n + c)
Euclidean-Metric City-Metric
Structural Complexity Structural Complexity
first-order Θ(n) first-order Θ(n + c)
arrow
arrow
arrow
q q0
p activation point
arrow
q q0
p activation point
arrow
q q0
p
arrow
q q0
p
arrow
q q0
p
arrow
q q0
p
arrow
q q0
p
arrow
q q0
p
arrow
r
’quickest path’ from q to r
q q0
p
dC (p, q)
arrow
r
’quickest path’ from q to r
q q0
p
dC (p, q)
q q0
Institute of Theoretical Informatics
Andreas Gemsa, D.T. Lee, Chih-Hung Liu, Dorothea Wagner – Higher Order City Voronoi Diagrams. Prof. Dr. Dorothea Wagner
Structural Complexity – Overview
BC (p, q)
BC (p, q)
BC (p, q)
BC (p, q)
Mixed Vertices
# of mixed vertices
p
BC (p, q)
Mixed Vertices
# of mixed vertices
p
BC (p, q)
Mixed Vertices
# of mixed vertices
p
# of mixed vertices
q
BC (p, q)
Mixed Vertices
# of mixed vertices
p
# of mixed vertices
q
BC (p, q)
Mixed Vertices
# of mixed vertices
p
# of mixed vertices
q
r
q
r
q
r
q
BC (p, q)
Mixed Vertices
# of mixed vertices
p
# of mixed vertices
q
BC (p, q)
Mixed Vertices
# of mixed vertices
p
# of mixed vertices
q
BC (p, q)
BC (p, q)
BC (p, q)
m4
p
m3
m2
q
m1
BC (p, q)
m4
p
m3
m2
q
m1
m4
p
m3
m2
q
m1
m4
p
m3
m2
q
m1
m2
q
m1
m2
q
m1
m2
Lemma 1
Voronoi edge e has m mixed q
vertices. m1
m2
Lemma 1
Voronoi edge e has m mixed q
vertices. m1
Then e consists of O(m + 1)
segments.
Institute of Theoretical Informatics
Andreas Gemsa, D.T. Lee, Chih-Hung Liu, Dorothea Wagner – Higher Order City Voronoi Diagrams. Prof. Dr. Dorothea Wagner
Mixed Voronoi Vertices and An Upper Bound
Lemma 2
|Vk (S)| = O(k(n − k) + M ), M = #mixed vertices of Vk (S)
#edges is O(k(n − k)) [Lee, 1982] BC (p, q)
|ei | = O(mi + 1) (Lemma 1)
P
M= mi m4
p
m3
m2
Lemma 1
Voronoi edge e has m mixed q
vertices. m1
Then e consists of O(m + 1)
segments.
Institute of Theoretical Informatics
Andreas Gemsa, D.T. Lee, Chih-Hung Liu, Dorothea Wagner – Higher Order City Voronoi Diagrams. Prof. Dr. Dorothea Wagner
Structural Complexity – Overview
BC (p, q)
Mixed Vertices
# of mixed vertices
p
# of mixed vertices
q
BC (p, q)
Mixed Vertices
# of mixed vertices
p
# of mixed vertices
q
V1 (S)
q1 q3
q2 p
q4
q6 q5
Euclidean Metric
Institute of Theoretical Informatics
Andreas Gemsa, D.T. Lee, Chih-Hung Liu, Dorothea Wagner – Higher Order City Voronoi Diagrams. Prof. Dr. Dorothea Wagner
Iterative Construction of Vk
V1 (S)
q1 q3
q2 p
q4
q6 q5
Euclidean Metric
Institute of Theoretical Informatics
Andreas Gemsa, D.T. Lee, Chih-Hung Liu, Dorothea Wagner – Higher Order City Voronoi Diagrams. Prof. Dr. Dorothea Wagner
Iterative Construction of Vk
Q = S \ {p}
V1 (S) V1 (Q)
q1 q3 q1 q3
q2 p q2
q4 q4
q6 q5 q6 q5
Euclidean Metric
Institute of Theoretical Informatics
Andreas Gemsa, D.T. Lee, Chih-Hung Liu, Dorothea Wagner – Higher Order City Voronoi Diagrams. Prof. Dr. Dorothea Wagner
Iterative Construction of Vk
Q = S \ {p}
V1 (S) V1 (Q)
q1 q3 q1 q3
q2 p q2
q4 q4
q6 q5 q6 q5
Euclidean Metric
Institute of Theoretical Informatics
Andreas Gemsa, D.T. Lee, Chih-Hung Liu, Dorothea Wagner – Higher Order City Voronoi Diagrams. Prof. Dr. Dorothea Wagner
Iterative Construction of Vk
Q = S \ {p}
V1 (S) V1 (Q)
q1 q3 q1 q3
q2 p q2
q4 q4
q6 q5 q6 q5
Euclidean Metric
Institute of Theoretical Informatics
Andreas Gemsa, D.T. Lee, Chih-Hung Liu, Dorothea Wagner – Higher Order City Voronoi Diagrams. Prof. Dr. Dorothea Wagner
Iterative Construction of Vk
Q = S \ {p}
V1 (S) V1 (Q)
q1 q3 q1 q3
q2 p q2
q4 q4
q6 q5 q6 q5
Euclidean Metric
Institute of Theoretical Informatics
Andreas Gemsa, D.T. Lee, Chih-Hung Liu, Dorothea Wagner – Higher Order City Voronoi Diagrams. Prof. Dr. Dorothea Wagner
Iterative Construction of Vk
Q = S \ {p}
V1 (S) V1 (Q)
q1 q3 q1 q3
q2 p q2
q4 q4
q6 q5 q6 q5
Euclidean Metric
Institute of Theoretical Informatics
Andreas Gemsa, D.T. Lee, Chih-Hung Liu, Dorothea Wagner – Higher Order City Voronoi Diagrams. Prof. Dr. Dorothea Wagner
Iterative Construction of Vk
Q = S \ {p}
V1 (S) V1 (Q)
q1 q1 q3 q3
q2 q2 p
q4 q4
q6 q6 q5 q5
Euclidean Metric
Institute of Theoretical Informatics
Andreas Gemsa, D.T. Lee, Chih-Hung Liu, Dorothea Wagner – Higher Order City Voronoi Diagrams. Prof. Dr. Dorothea Wagner
Iterative Construction of Vk
Q = S \ {p}
V1 (S) ∩ V1 (Q)
q1 q3
q2 p
q4
q6 q5
Euclidean Metric
Institute of Theoretical Informatics
Andreas Gemsa, D.T. Lee, Chih-Hung Liu, Dorothea Wagner – Higher Order City Voronoi Diagrams. Prof. Dr. Dorothea Wagner
Iterative Construction of Vk
Q = S \ {p}
V1 (S) ∩ V1 (Q)
q1 q3
q2 p
q4
q6 q5
V2 (S)
Euclidean Metric
Institute of Theoretical Informatics
Andreas Gemsa, D.T. Lee, Chih-Hung Liu, Dorothea Wagner – Higher Order City Voronoi Diagrams. Prof. Dr. Dorothea Wagner
Iterative Construction of Vk
V1 (S) V1 (Q)
q1 q3 q1 q3
q2 p q2
q4 q4
q6 q5 q6 q5
V2 (S)
Euclidean Metric
Institute of Theoretical Informatics
Andreas Gemsa, D.T. Lee, Chih-Hung Liu, Dorothea Wagner – Higher Order City Voronoi Diagrams. Prof. Dr. Dorothea Wagner
Iterative Construction of Vk – Wavefront
Q: How to transfer this approach to wavefront propagation?
Propagate a wavefront from every qi and not from p
q1
q3
q2
p
q4
q6 q5
Euclidean Metric
Institute of Theoretical Informatics
Andreas Gemsa, D.T. Lee, Chih-Hung Liu, Dorothea Wagner – Higher Order City Voronoi Diagrams. Prof. Dr. Dorothea Wagner
Iterative Construction of Vk – Wavefront
Q: How to transfer this approach to wavefront propagation?
Propagate a wavefront from every qi and not from p
q1
q3
q2
p
q4
q6 q5
Euclidean Metric
Institute of Theoretical Informatics
Andreas Gemsa, D.T. Lee, Chih-Hung Liu, Dorothea Wagner – Higher Order City Voronoi Diagrams. Prof. Dr. Dorothea Wagner
Iterative Construction of Vk – Wavefront
Q: How to transfer this approach to wavefront propagation?
Propagate a wavefront from every qi and not from p
q1
q3
q2
p
q4
q6 q5
Euclidean Metric
Institute of Theoretical Informatics
Andreas Gemsa, D.T. Lee, Chih-Hung Liu, Dorothea Wagner – Higher Order City Voronoi Diagrams. Prof. Dr. Dorothea Wagner
Iterative Construction of Vk – Wavefront
Q: How to transfer this approach to wavefront propagation?
Propagate a wavefront from every qi and not from p
q1
q3
q2
p
q4
q6 q5
Euclidean Metric
Institute of Theoretical Informatics
Andreas Gemsa, D.T. Lee, Chih-Hung Liu, Dorothea Wagner – Higher Order City Voronoi Diagrams. Prof. Dr. Dorothea Wagner
Iterative Construction of Vk – Wavefront
Q: How to transfer this approach to wavefront propagation?
Propagate a wavefront from every qi and not from p
q1
q3
q2
p
q4
q6 q5
Euclidean Metric
Institute of Theoretical Informatics
Andreas Gemsa, D.T. Lee, Chih-Hung Liu, Dorothea Wagner – Higher Order City Voronoi Diagrams. Prof. Dr. Dorothea Wagner
Iterative Construction of Vk – Wavefront
Q: How to transfer this approach to wavefront propagation?
Propagate a wavefront from every qi and not from p
q1
q3
q2
p
q4
q6 q5
Euclidean Metric
Institute of Theoretical Informatics
Andreas Gemsa, D.T. Lee, Chih-Hung Liu, Dorothea Wagner – Higher Order City Voronoi Diagrams. Prof. Dr. Dorothea Wagner
Iterative Construction of Vk – City Metric
before: wavefront propagation only from q
BC (p, q)
BC (p, q)
q2
q1 q
q10
q2
q1 q
q10
q2
q1 q
q10
q2
q1 q
q10
q2
q1 q
q10
q2
q1 q
q10
q2
q1 q
q10
q2
q1 q
q10
q2
q1 q
q10
q1 q
q10
Theorem 1
|Vk (S)| = O(k(n − k) + kc)
Theorem 1
|Vk (S)| = O(k(n − k) + kc)
BC (p, q)
Mixed Vertices
# of mixed vertices
p
# of mixed vertices
q
BC (p, q)
Mixed Vertices
# of mixed vertices
p
# of mixed vertices
q
q2
q1
q10 q
Computing V1 (·) with algorithm by Bae et al. [Bae, Kim, Chwa, 2009]
Computing V1 (·) with algorithm by Bae et al. [Bae, Kim, Chwa, 2009]
Theorem 2
The k -th order city Voronoi diagram can be computed
in O(k 2 (n + c) log(n + c)) time.
City-Metric
first-order: impact of c is an Structural Complexity
additive constant first-order Θ(n + c)
farthest site: impact of c is
not an additive constant k th -order O(k(n − k) + kc)
Ω(n + kc)
farthest-site Θ(nc)
Open questions:
Tight bound for the structural Time Complexity
complexity? first-order O((n + c) log n)
Transportation network under
k th
-order O(k2 (n + c) log(n + c))
the Euclidean metric?
2
farthest-site O(nc log n log (n + c))
City-Metric
first-order: impact of c is an Structural Complexity
additive constant first-order Θ(n + c)
farthest site: impact of c is
not an additive constant k th -order O(k(n − k) + kc)
Ω(n + kc)
farthest-site Θ(nc)
Open questions:
Tight bound for the structural Time Complexity
complexity? first-order O((n + c) log n)
Transportation network under
k th
-order O(k2 (n + c) log(n + c))
the Euclidean metric?
2
farthest-site O(nc log n log (n + c))
Thanks!
Institute of Theoretical Informatics
Andreas Gemsa, D.T. Lee, Chih-Hung Liu, Dorothea Wagner – Higher Order City Voronoi Diagrams. Prof. Dr. Dorothea Wagner