Near-Linear Near-Linear Approximation Algorithms Approximation Algorithms for Geometric Hitting for Geometric Hitting Sets Sets Pankaj Agarwal Esther Ezra Micha Pankaj Agarwal Esther Ezra Micha Sharir Sharir Duke Duke Duke Duke Tel-Aviv University Tel-Aviv University University University University University
46
Embed
Near-Linear Approximation Algorithms for Geometric Hitting Sets Pankaj Agarwal Esther Ezra Micha Sharir Duke Duke Tel-Aviv University University University.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Near-Linear Approximation Near-Linear Approximation Algorithms for Geometric Algorithms for Geometric
Hitting SetsHitting Sets
Pankaj Agarwal Esther Ezra Micha SharirPankaj Agarwal Esther Ezra Micha Sharir Duke Duke Tel-Aviv Duke Duke Tel-Aviv
University University University University University University
Range SpacesRange Spaces
Range space (X, R) :
X – Ground set.
R – Ranges: Subsets of X .
|R| 2|X|
Abstract form: Hypergraphs.
X – vertices.
R – hyperedges.
specification: X d, R = set of simply-shaped regions in d
X – Points on the real line.R – Intervals.
X – Points on the plane.R – halfplanes.
X – Points on the plane.R – Disks.
Geometric Range SpacesGeometric Range Spaces
X finite: Discrete model ,
X infinite: Continuous model
The hitting-set problemThe hitting-set problem
A hitting set for (X, R) is a subset H X, s.t., for any Q R , Q H .
Goal: find smallest hitting set.
Useful applications: art-gallery,sensor networking, and more.
Hardness of hitting setsHardness of hitting sets
Finding a hitting set of smallest size is NP-hard,
= size of the smallest hitting set.m = |X| , n = |R| .
Approximation factor: O(log ) .Sometimes, the approximation factor is even smaller!
Number of iterations: O( log (m/)) .
Running time per iteration: O*(n + m) (naïve).Overall: O*(m + n2 )
Goal: Improve the running time.A (near) linear-time algorithm?
In some cases, can be improved to
O*(n+m)
Improved!
Our resultOur resultPlanar regions:Obtain a O*(log n) approximation in near-linear time , when the union complexity of R is near-linear.
Applied in both discrete and continuous models.
Specifically:Union complexity of any subset R’ R, |R’| = r is O(r (r)) . Obtain a hitting set for (X, R) of size O( () log n)in (randomized expected) time O*(m + n) .
() is a slowly growing function.
The running time is O*(n) for the
continuous model.
The union boundary
Our result: Axis-parallel boxes in Our result: Axis-parallel boxes in dd
Approximation for geometric Approximation for geometric hitting setshitting sets
Geometric range spaces:
Achieve improved approximation factor!
Approximation factor: O(1 + log ) ,
Sometimes, the approximation factor is even smaller!
Points and disks or pseudo-disks in 2D: O(1) .
Points and halfspaces in 2D and 3D: O(1) .
Points and axis-parallel boxes in 2D and 3D: O(log log ) .
This is achieved via -nets
-nets for range spaces-nets for range spacesGiven:
• A range space (X, R) , assume X is finite, |X| = n .
• A parameter 0 < < 1 ,
An -net for (X, R) is a subset N X that hits every
range Q R, with |Q X| n .
N is a hitting set for all the ``heavy'' ranges.
Example:
Points and intervals on the real line: |N| = 1/ .
n
Bound does not depend on n.
Approximation for geometric hitting Approximation for geometric hitting setssets
The Bronimann-Goodrich technique / LP-relaxationIf (X, R) admits an -net of size f(1/ ) ,then there exists a polynomial-time approximation algorithm that reports a hitting set of size O(f()) .
Idea: Assign weights on X s.t each range Q R becomes heavy .Construct an -net for the weighed range space. Each range is hit by the -net.
Small-size -nets imply small approximation factors!
-net theorem-net theoremWhat is the size of -nets in geometric range spaces?
Theorem [Haussler-Welzl, 87]:If the ranges are simply-shaped regions, then, for any > 0, there exists an -net N X of size:
O(1/ log (1/ )) .
Moreover, a random sample of that size is an -net, with constant probability.
Theorem [Komlos, Pach, Woeginger 92]:The bound is tight!
No lower bound better than (1/ ) is known in geometry.
Bound does not depend on n.
Artificial construction(non-geometric!).
The Bronimann-Goodrich techniqueThe Bronimann-Goodrich technique
Number of iterations: O( log (m/)) .
Performance of the algorithm:
(Weighted) net-finder: O(m)
Verifier: O(n |N| + m) O*(n + m) (naïve)
Overall: O*(m + n2 )
Improvement in some cases:
Verifier: O*(m + n)
Overall: O*((n + m) )
|N| = O( log )
Axis-parallel rectangles andplanar regions with near-linear
union complexity
Arrangement and levelsArrangement and levels
R = {Q1, … Qn} set of planar regions.A(R) – the arrangement of R.
The depth (p, R) of a point p 2 in A(R) =# regions in R that contain p in their interior.
The-level of A(R) = set of all points with depth .In particular:The 0-level is the closure of the complement of the union of R.
A(R) = the points at level in A(R) .
p
(p,R)=3
Complexity of AComplexity of A(R)(R)
Using Clarkson & Shor:
The complexity of A() is:O(2 f(n/)) = O(n (n/) )
Vertical decomposition of A() (or A() ) :Partition each cell of A() (or A() ) intopseudo-trapezoidal cells.The decomposition has the same complexity as
A() (or A() ) .
Vertical decomposition
of level 1
Union complexity
= n/ (n/) .
Need to assume constant description
complexity
1/r-cuttings1/r-cuttingsr - the parameter of the cutting, 1 r n .
Construct (1/r)-cuttings:Choose a random sample K R of O(r log r) regions.Form the planar arrangement A(K) of KConstruct the vertical decomposition of A(K).
All cells cover A(R) .With high probability, each cell meets n/r boundary regions.
Improvement [Chazelle-Friedman]:: The number of cells can be decreased to O(r2).
# cells in the cutting is O(r2 log2 r)
(1/r)-cuttings for A(R)(1/r)-cuttings for A(R)r - the parameter of the cutting, 1 r n .
1. Choose a random sample K R of O(r log r) regions.
2. Form the planar arrangement A(k) of k: Overall complexity: O(r2 log2r).
3. Construct the vertical decomposition of A(K).Number of cells: O(r2 log2r)
Theorem [Clarkson & Shor], [Haussler & Welzl] :Each cell is crossed by n/r boundary regions of R, with high probability.
Improvement: Use two-level sampling [Chazelle-Friedman]: The number of cells can be decreased to O(r2).
Improved 1/r-cuttingsImproved 1/r-cuttingsTheorem [Chazelle-Friedman]:The size of the cutting can be improved to O(r2).
Proof sketch (apply a two-level sampling):First step: Choose a random sample K R of O(r) regions, and construct the vertical decomposition.
Second (repair) step: For each “heavy” cell that meets tn/r boundary regions, for t > 1, construct a (1/t)-cutting within :• Choose a random sample K of O(t log t) regions.
• Construct the vertical decomposition of A(K) .• Clip these cells within .
The number of the heavy cells is
small!
Shallow cuttingsShallow cuttings
A shallow cutting is a (1/r)-cutting that covers Al(R) .
Construct shallow cuttings:Discard all cells of the full (1/r)- cutting that do not meet Al(R) .
Theorem [Matousek, Agarwal etal.]The size of the shallow cutting is
O(q r (r/q)) ,where q = (r/n) + 1 .
When = n, # cells is O(r2) .
= max p X {(p, R)}, choose r = c n / , c > 0 sufficiently large constant .
• Construct a (1/r)-cutting for level of A(R) .
• For each cell with X , choose an arbitrary point p X .
• Eliminate all regions in R stabbed by p.
• R’ = set of remaining regions.Claim: max p X {(p, R’)} /2 .
• Recurse with R’.Bottom of recursion: c1 (some constant):Construct the entire A (R) , and choose a single point in each non-empty cell.
The algorithm for the discrete modelThe algorithm for the discrete model
The maximum depth in A(R’)The maximum depth in A(R’)
A (remaining) region in R’ cannot fully contain a non-empty cell of the cutting.
The boundary of these regions cross cells .
max p X {(p, R’)} = max # crossing boundary regions in a cell
max p X {(p, R’)} n/r = /c /2.
The algorithm terminates after log log n steps!
R’
The size of the hitting set in each The size of the hitting set in each stepstep
By the shallow cutting theorem, in each step,
the size of the hitting set is bounded by the number of cells:
O(q r (r/q)) , q = (r/n) + 1 .
Put = , r = c n / , the size is
O(n/ (n/)) .
Informal description:
The averaged depth ’ in the arrangement of the sample is O(1).
According to Clarkson & Shor, the complexity of that level is:
O(r ’ (r/’) ) = O(n/ (n/)) .
The approximation factorThe approximation factor
The overall size is: O(n/ (n/) log n) .
Observation: n/ .
Each point in the optimum can stab regions.
This property holds in each step of the algorithm.