computing covers of plane forests luis barba alexis
play

Computing Covers of Plane Forests Luis Barba Alexis Beingessner - PowerPoint PPT Presentation

Computing Covers of Plane Forests Luis Barba Alexis Beingessner Prosenjit Bose Michiel Smid Previous Work Given a set T = { T 1 , T 2 , . . . , T m } of m pairwise non-crossing geometric trees with a total of n vertices in general position. The


  1. Computing Covers of Plane Forests Luis Barba Alexis Beingessner Prosenjit Bose Michiel Smid

  2. Previous Work Given a set T = { T 1 , T 2 , . . . , T m } of m pairwise non-crossing geometric trees with a total of n vertices in general position. The coverage of these trees is the set of all points p in R 2 such that every line through p intersects at least one of the trees.

  3. Previous Work

  4. Previous Work

  5. Previous Work Beingessner and Smid 2012: ◮ Coverage can be computed in O ( m 2 n 2 ) time ◮ Worst case example with coverage of size Ω( n 4 ). ◮ Problem is Θ( n 4 )

  6. Previous Work Is slowness a consequence of bad inputs being “contrived”? Optimization to be had in structure of “real” inputs?

  7. Observations Coverage of a single tree is it’s convex hull

  8. Observations If two coverages overlap, their combined convex hull is covered

  9. The hull-cover ◮ Compute the convex hull, CH ( T i ), of every tree T i ∈ T ◮ If any two convex hulls overlap, replace them with their convex hull ◮ Repeat until all convex hulls computed thusly are disjoint ◮ Resulting set of convex polygons is the hull-cover of T

  10. The hull-cover

  11. The hull-cover

  12. The hull-cover

  13. The hull-cover

  14. The hull-cover

  15. Approximation Does this approximate the coverage? ◮ A terrible approximation (for already hard inputs) ◮ A great approximation (for natural inputs)

  16. Computing the hull-cover Challenges: ◮ Finding pairwise intersection is fairly expensive ◮ Computing convex hulls is fairly expensive

  17. Weakly Disjoint Polygons Let a weakly disjoint pair of convex polygons P , Q be a pair of convex polygons such that P \ Q and Q \ P are both connected sets of points, and P does not share a vertex with Q .

  18. Weakly Disjoint Polygons A pair of polygons that are weakly disjoint, but not disjoint

  19. Weakly Disjoint Polygons A pair of polygons that are not weakly weakly disjoint

  20. Weakly Disjoint Polygons Lemma If two convex polygons P , Q are weakly disjoint, then their boundaries intersect at at most two points.

  21. Weakly Disjoint Polygons

  22. Weakly Disjoint Polygons

  23. Weakly Disjoint Polygons

  24. Weakly Disjoint Polygons

  25. Weakly Disjoint Polygons Lemma If two convex polygons P , Q are weakly disjoint, but not disjoint, then one contains a vertex of the other.

  26. Weakly Disjoint Polygons

  27. Lemma The convex hulls of two disjoint trees are weakly disjoint.

  28. Weakly Disjoint Polygons

  29. Blocked or Nested Lemma Assume R and S are two non-crossing trees whose convex hulls intersect. Then the convex hull of one is strictly inside the other, or there exists a pair of adjacent vertices on the convex hull of one whose visibility is blocked by the other tree.

  30. Blocked or Nested

  31. Shoot and Insert

  32. Shoot and Insert Ishaque et al. 2012: n pairwise disjoint polygonal obstacles can be preprocessed in O ( n log n ) time and space to support m permanent ray shootings in O (( n + m ) log 2 n + m log m ) time

  33. Algorithm

  34. Algorithm

  35. Algorithm

  36. Algorithm 1

  37. Algorithm 1 2

  38. Algorithm 3 1 2

  39. Algorithm 3 1 2 4

  40. Algorithm 5 3 1 2 4

  41. Algorithm 6 5 3 1 2 4

  42. Algorithm 6 5 7 3 1 2 4

  43. Algorithm 6 5 7 3 1 2 8 4

  44. Algorithm 6 5 7 9 3 1 2 8 4

  45. Algorithm

  46. Algorithm

  47. Analysis O ( n log 2 n ) time O ( n log n ) space

  48. The End Thank you! Questions?

Recommend


More recommend