Title: University of California Santa Barbara Department of Computer Science
1University of CaliforniaSanta BarbaraDepartment
of Computer Science
- Minimum Length Corridor and Related Problems
- Ph. D. Thesis Proposal
Committee Teofilo F. Gonzalez (Chair) Peter
Cappello Ömer Egecioglu
Arturo Gonzalez-Gutierrez
2Outline
- Minimum Length Corridor (MLC) Problem
- Related Problems
- NP-completeness
- Approximation Algorithm TRA-MLC-R
- Future Work
3Minimum Length Corridor (MLC) Problem
- Motivation
- Applications
- Formal Definition
- Background
- Hierarchy
4MLC Problem Motivation
MLC solution
TRA-MLC solution
5MLC Problem Applications
- Circuit Board Layout Design
- Wires for power supply
- Building Wiring Design
- Optical Fiber for Data Communication Networks
- Wires for Electrical Networks
- Trip Planning Queries in Spatial Databases
6MLC Problem Formal Definition
- INPUT Given a pair (F,R), where F is a
rectangular boundary whose interior is
partitioned into rectilinear polygons
RR1,R2,,Rm. - OUTPUT A corridor for (F,R) consisting of a set
of connected line segments each of which
completely overlap with the line segments that
form F or the boundary of the rectilinear
polygons, and include at least one point of F and
at least one point from each of the m rooms. - OBJECTIVE FUNCTION Minimize the total edge
length of the corridor.
7MLC Problem Background
- Demaine, E.D., and ORourke, J. Open problems
from CCCG 2000. In Proceedings of the 13th.
Canadian Conference on Computational Geometry
(2001), (by Naoki Katoh). http//theory.csail.mit.
edu/edemaine/papers/CCCG2000Open/ - Eppstein, D. Some open problems in graph theory
and computational geometry. http//www.ics.uci.edu
/eppstein/200-f01.pdf, November 2001. - Jin, L. Y., and Chong, O. W. The minimum touching
tree problem. http//www.yewjin.com/research/Minim
umTouchingTrees.pdf, National University of
Singapore, School of Computing, 2003. - No polynomial time algorithm known
- Not even a constant ratio approximation algorithm
- Seems likely to be NP-complete but no proof known
8MLC Problem Hierarchy
9Related Problems
- Tree Feedback Node Set (TFNS)
- Tree Vertex Cover (TVC)
- Group Steiner Tree (GST)
- Tree Errand Cover (TEC)
10Related Problems Tree Feedback Node Set
Generalization of the N-MLC
- INPUT A connected undirected edge-weighted graph
G(V,E,w), where wE?? is an edge-weight
function. - OUTPUT A tree T(V,E), where E?E, and V?V is
a feedback node set(every cycle in G includes at
least one vertex in V) and the total edge-weight
is minimized.
11Related Problems Tree Feedback Node Set Tree
Vertex Cover Problem
- INPUT A connected undirected edge-weighted graph
G(V,E,w), where wE?? is an edge-weight
function. - OUTPUT A tree TG of G ? its vertices cover all
the edges E, and the total length
is minimized.
12Related Problems Tree Vertex Cover
- A polynomial approximation algorithm with
performance 3.78 Arkin, E.M. et. al. (1993) - Uses the approximation algorithm for the weighted
vertex cover and then uses an approximation
algorithm for the Steiner tree problem. - Replace the approximation algorithm for the
weighted vertex cover by a constant ratio
approximation algorithm for the weighted FNS
problem Chudak, F. A. et. al. (1998).
13Related Problems Group Steiner Tree
ProblemGeneralization of the MLC Problem
- INPUT A connected undirected edge-weighted graph
G(V,E,w), where wE?? is an edge-weight
function, a non-empty subset N, N?V, of
terminals and a partition N1,N2,...,Nk of N. - OUTPUT A tree TG(N) of G ? at least one terminal
from each set Ni is in the tree TG(N) and the
total length is minimized.
14Related Problems Group Steiner Tree
ProblemReducing the TRA-MLC to the GST
15Related Problems Group Steiner Tree
ProblemReducing the TRA-MLC to the GST
16Related Problems Group Steiner Tree
ProblemReducing the TRA-MLC to the GST
17Related Problems Group Steiner Tree Problem
18Related Problems Tree Errand Cover
- INPUT A connected undirected edge-weighted graph
G(V,E,w), where wE?? is an edge-weight
function, a non-empty set Ut1,t2,...,tm of
errands. Associated with each vertex i?V is a set
Si?U, where . - OUTPUT A tree TG of G ? and
the total length is minimized
19Related Problems Tree Errand CoverTRA-MLC?TEC
UR1,R2,R3 ,R4,R5,R6,R7,R8,R9
S1R1 S2R1,R2 S3R2,R3 S4R3 S5R1,R2,R5
S6R2,R3,R5 S7R3,R5,R6 S8R3,R6 S9R1,R4
S10R1,R4,R5
S11R5, R6, R8 S12R6,R8 S13R4,R7 S14R4,
R5, R7 S15R5,R8,R9 S16R8,R9 S17R7 S18R
7,R5 S19R5,R9 S20R9
S4R3 S5R1,R2,R5 S6R2,R3,R5 S7R3,R5,R6
S8R3,R6 S10R1,R4,R5 S12R6,R8 S14R4,
R5, R7 S16R8,R9
20Related Problems Tree Errand CoverPerformance
ratio
- Slavik, P. (1998)
- The TEC problem can be approximated to within a
factor of 2? in polynomial time, when each errand
is assigned to at most ? vertices. - Still, applying this case even to the TRA-MLC-R
problem does not give us a constant approximation
algorithm. - There are errands that may be assigned to any
number of vertices.
21NP-Completeness
- TRA-MLC-R
- MLC-R
- MLCf-R
- MA- MLCf-R
- TRA-MLC
- MLC
- MLCf
- MA-MLCf
22NP-Completeness
- Complexity of the Minimum-Length Corridor
Problem, Submitted to Publication, March 2005.
23NP-Completeness
Theorem 1. TRA-MLC-R a MLC-R
24NP-Completeness
Theorem 2. TRA-MLC-R a MLCf-R. TRA-MLC-R a
MA-MLCf-R with k access points, k2.
25NP-completeness
- Theorem 3. The MLC, MLC-R, MLCf, MLCf-R, MA- MLCf
and MA- MLCf-R problems are NP-complete.
26NP-Completeness
- Theorem 4. The TRA-MLC problem is NP-complete.
- Theorem 5. The TRA-MLC-R problem is NP-complete
27NP-Completeness TRA-MLC
I ? 3SAT is satisfiable ? f(I)?TRA-MLC has a
corridor with length at most B
28NP-Completeness TRA-MLC
29NP-Completeness TRA-MLC-R
30NP-Completeness TRA-MLC-R
31NP-Completeness TRA-MLC-R
32Approximation Algorithm TRA-MLC-R
- Approximation Ratio
- Approach by Restriction
- Restriction to four points
- Restriction to five points
- Restriction to one and two points
- Restriction to three points
33Approximation Algorithm Approximation Ratio
- Use Slaviks Algorithm for the TEC problem for a
given ?
S?(I) Solution to the TRA-MLC-R? generated by
using Slaviks Algorithm O?(I) Optimal solution
to the TRA-MLC-R? S(I) Any solution to the
TRA-MLC-R O(I) Optimal solution to the TRA-MLC-R
For every corridor S(I) we can construct S?(I) ?
S?(I) c S(I) Applying this to O(I), S?(I) c
O(I) We know that O?(I) S?(I) Therefore,
O?(I) S?(I) c O(I) Slaviks solution shows
that S?(I)2? O?(I) And, S?(I) 2? O?(I) 2? c
O(I)
?maxV(Ri)
34Approximation Algorithm Restriction to one
pointOne out of four corners
TRA-MLC-R ? TRA-MLC-R4
35Approximation Algorithm Restriction to five
pointsFour corners and special point
TRA-MLC-R ? TRA-MLC-R5
36Approximation Algorithm Restriction to five
points Basic definitions
- SP(p,q) Length of the shortest path from vertex
p to vertex q using only boundary F and rectangle
line segments. - MinLen(p,Ri) Minimum edge length such that every
vertex q of rectangle Ri satisfies
SP(p,q)MinLen(p,Ri) - diminMinLen(vi,Rj)ri?j Min-connectivity
distance for rectangle Ri through vertex vi to
another rectangle Rx
37Approximation Algorithm Restriction to five
points Min connectivity distance for R5
38Approximation Algorithm Restriction to five
pointscpe and ncpe rectangles
V(Ri)5?Ri is a cpes
V(Ri)gt5?Ri is a ncpes
39Approximation Algorithm Restriction to five
points ncpes and pre-order traversal
S5(I)3S(I)
FEtraversing?Econexions of sp
40Approximation Algorithm Restriction to five
points Regions type 2,1,0.
41Approximation Algorithm Restriction to five
points
0
42Approximation Algorithm Restriction to five
points From here on we assume the first ncpe
rectangle is already connected
0 001
43Approximation Algorithm Restriction to five
points
0 0010010
44Approximation Algorithm Restriction to five
points
0 001001000100
45Approximation Algorithm Restriction to five
points
0 001001000100001000
0 0010
46 Approximation Algorithm Restriction to five
points
0 0010 0011
47Approximation Algorithm Restriction to five
points
0 0010 001100111
000100011
0001(01)
48Approximation Algorithm Restriction to five
points
0001(01) 01
49Approximation Algorithm Restriction to five
points
0001(01) 01010
50 Approximation Algorithm Restriction to five
points
0001(01) 01(01)
0001(01) 01010 011
51Approximation Algorithm Restriction to five
points Rules
- 0?OK
- 1 ?OK
- 1x?x
- 02x?x
- 0001(01) ?OK
- 01(01) ?OK
- 0001(01)2x ?x
- 01(01)2x ?x
S5(I)3S(I) S5253O(I)30O(I)
52Approximation Algorithm Restriction to five
pointsBy rule 5 0001(01) ?OK
53Approximation Algorithm Restriction to one and
two points
Special point and bottom-left corner
Special point
Special point and top-left corner
Special point and top-right corner
Special point and bottom-right corner
54Future work
- Is there a provably constant 30-approximation
algorithm for the TRA-MLC-R problem by
restricting maxV(Ri) to five vertices? - Is the decision version of TRA-MLC-R5 problem
NP-complete? - Is there a provably constant approximation
algorithm for the TRA-MLC-R problem by
restricting maxV(Ri) to three vertices
(TR,BL,SP)? - ?3,c5?23530
- Is the decision version of TRA-MLC-R3 problem
NP-complete? - Is it possible to have a O(constant)-approximation
algorithm for a given number of critical points
bounded by a constant l.
55Thank you!
- Questions
- Comments
- Suggestions