anchored drawings of planar graphs
play

Anchored Drawings of Planar Graphs Angelini, Da Lozzo, Di - PowerPoint PPT Presentation

22nd International Symposium on Graph Drawing 24-26 September 2014, Wrzburg, Germany Anchored Drawings of Planar Graphs Angelini, Da Lozzo, Di Bartolomeo, Di Battista, Hong, Patrignani, Roselli Applicative Context Drawing a graph on a


  1. 22nd International Symposium on Graph Drawing 24-26 September 2014, Würzburg, Germany Anchored Drawings of Planar Graphs Angelini, Da Lozzo, Di Bartolomeo, Di Battista, Hong, Patrignani, Roselli

  2. Applicative Context ● Drawing a graph on a geographical map ● Vertices have fixed positions

  3. Drawing Nicely ● Our idea: ○ Let vertices move “a bit” around their positions ○ Check if this allows a planar drawing of the graph

  4. Anchored Graph Drawing Problem ● Instance ○ Planar graph G ○ Initial vertex positions α ( v ) ○ Maximum distance δ ● Question ○ Does G admits a planar drawing ○ ...such that vertices move by distance at most δ ○ ...from their initial positions α ?

  5. Considered Settings Distance “Euclidean” “Manhattan” “Uniform” d = (d x 2 + d y 2 ) 1/2 d = d x + d y Function d = max(d x ,d y) Vertex d y d y d y d x d x d x Region Rectilinear Straight-line Drawing Style

  6. Previous work ● NP-hard: straight-line and disks of different size ○ Godau. On the difficulty of embedding planar graphs with inaccuracies . 1995 ● NP-hard: rectilinear and δ = inf ○ Garg, Tamassia. On the comp. compl. of upward and rectilinear planarity test . 2001 ● Application of force-directed algorithms ○ Abellanas et. al. Network drawing with geographical constraints on vertices . 2005 ● Iterative adjustments that preserve mental map ○ Lyons et. al. Algorithms for cluster busting in anchored graph drawing . 1998

  7. Assumption ● No overlap between vertex regions ○ Or two vertices may invert their positions ■ Very confusing for a user ○ Relationship with Clustered Planarity with drawn clusters

  8. Our Results Metric Straight-line Rectilinear Manhattan NP-hard NP-hard d x Euclidean NP-hard NP-hard Uniform NP-hard Polynomial

  9. Our Results Metric Straight-line Rectilinear Manhattan NP-hard NP-hard d x Euclidean NP-hard NP-hard Uniform NP-hard Polynomial

  10. Polynomial Case ● Connected graph ● Uniform distance ( regions) ● Rectilinear drawing

  11. Edge Pipes ● We call pipe the convex hull of two regions ○ Minus the regions ● An edge can be drawn only inside a pipe ● In this setting pipes “get rectilinear” too

  12. Rectilinear Edges ● An edge is either horizontal or vertical ● Can be deduced by the region positions ● Visibility is required between two endpoints

  13. Trimming ● Regions and pipes can trim each other ● A trimmed area cannot be used

  14. General Strategy 1. Start from the initial region/pipe configuration 2. While (a trim is possible): a. Trim unusable parts of pipes and regions b. Check if a negative configuration is obtained 3. Flag the instance as positive 4. Draw edges according to the current pipes

  15. Trimming Pipes ● VP-overlaps can trim a pipe

  16. Trimming Regions ● VP-overlaps can trim a region

  17. Negative Instances No visibility PP-overlap (Unavoidable crossing)

  18. An Example of Execution

  19. An Example of Execution

  20. An Example of Execution

  21. An Example of Execution

  22. An Example of Execution

  23. NP-hard Case ● Euclidean distance ( regions ) ● Straight-line drawing ● Reduction from Planar 3-SAT

  24. Planar 3-SAT (x 1 ˅ ¬x 2 ˅ x 5 ) ˄ (x 2 ˅ x 3 ˅ ¬x 4 ) ˄ (x 1 ˅ ¬x 3 ˅ x 5 ) ˄ (x 3 ˅ x 4 ˅ x 5 ) C 1 C 2 C 3 C 4 x 1 c 1 x 2 c 2 x 3 c 3 x 4 x 5 c 4

  25. Planar 3-SAT - Gadgets

  26. Planar 3-SAT - Variable Gadget

  27. Planar 3-SAT - Clause Gadget

  28. Planar 3-SAT - Truth Propagation

  29. Planar 3-SAT - Not Gadget

  30. Planar 3-SAT - Turn Gadget

  31. Planar 3-SAT - Split Gadget

  32. Variable Gadget True configuration False configuration F T F T

  33. Truth Propagation F F T T

  34. Not Gadget F F T T

  35. Turn Gadget F T F T

  36. Split Gadget T F F T F T

  37. Clause Gadget T T a F F x y F-T-T case The gadget is planar F T b

  38. Clause Gadget T T a F F x y F-F-F case The gadget is NOT planar F T b

  39. Clause Gadget T T F F F T

  40. Clause Gadget T T F F F T

  41. Clause Gadget T T F F F T

  42. Clause Gadget T T F F F T

  43. Clause Gadget T T F F F T

  44. Clause Gadget T T F F F T

  45. Open Problems ● Do the hard problems belong to NP? ● Still hard with biconnected gadgets. What if triconnected? ● What if we allow regions to partially overlap? ● What if we allow some crossings?

  46. Applicative Context ● Drawing a graph on a geographical map ● Vertices have fixed positions

  47. Clause Gadget (master slide) T T F F F T

  48. Challenges ● Vertex cluttering, edge crossings ● Techniques exist to mitigate cluttering ● However, crossings are still an issue

Recommend


More recommend