solvability of cubic graphs and
play

Solvability of Cubic Graphs and The Four Color Theorem Tony T. Lee - PowerPoint PPT Presentation

Solvability of Cubic Graphs and The Four Color Theorem Tony T. Lee Shanghai Jiao Tong University The Chinese University of Hong Kong July 1, 2013 Research Assistants: Yujie Wan, Hao Quan, Qingqi Shi Kempe Chain In 1879, Kempe provided the


  1. Solvability of Cubic Graphs and The Four Color Theorem Tony T. Lee Shanghai Jiao Tong University The Chinese University of Hong Kong July 1, 2013 Research Assistants: Yujie Wan, Hao Quan, Qingqi Shi

  2. Kempe Chain  In 1879, Kempe provided the first proof of four color theorem(4CT). Found to be flawed by Heawood in 1890.  Introduced a technique now called Kempe chains. A. B. Kempe, On the Geographical Problem of Four-Colors, Amer. J. Math . 2 (1879), 193-200.

  3. Tait Cycle  Tait’s proof published in 1880. Found to be flawed by Petersen in 1891.  Found an equivalent formulation of the 4CT in terms of three-edge coloring. P. G. Tait, Note on a theorem in geometry of position, Trans. Roy. Soc. Edinburgh 29 (1880), 657-660 .

  4. Computer-assisted Proof of 4CT  Kenneth Appel and Wolfgang Haken (1976).  Neil Robertson, Daniel Sanders, Paul Seymour, and Robin Thomas (1997). Another simpler proof.

  5. Computer-assisted Proof of 4CT  The computer-assisted proofs of the four color theorem caused great amounts of controversy because they can not be verified by human.  The search continues for a computer-free proof of the Four Color Theorem.

  6. Edge Coloring  Edge Coloring : an assignment of “colors” to the edges of the graph so that no two adjacent edges have the same color.

  7. Theorems on Edge Coloring  Petersen’s Theorem : Every bridgeless cubic graph contains a perfect matching.  Vizing’s Theorem [1]: Any simple graph is either Δ - or Δ + 1 -edge-colorable. Chromatic Index: 𝜓 𝑓 .  Holyer [2]: Deciding Δ - or Δ + 1 -edge-colorable is NP-Complete, even for Δ = 3 . 1. V. Vizing. On an estimate of the chromatic class of a p-graph. Diskret. Analiz , 3(7):25 – 30,1964 2. I. Holyer. The NP-completeness of edge-colouring. Siam J. Comput , 10(4):718 – 720, 1981.

  8. Outline  Operations of Complex Colors  Decomposition of Configurations  Solvability of Configurations  Generalized Petersen Configuration  Three-Edge-Coloring Theorem  Graph Theory versus Euclidean Geometry  Conclusions

  9. Constraints of Edge Coloring  Vertex constraint colors assigned to links incident to the same vertex are all distinct  Edge constraint  Variable-colored edge  Constant-colored edge

  10. Complex Coloring of Tetrahedron v v 1 1 e e 1,4 1,4 e e e e 1,2 1,3 1,2 1,3 e e v e e v 2,4 3,4 4 2,4 3,4 4 v v e v v e 2 2,3 3 2 2,3 3 Consistent Coloring Proper Coloring

  11. Color-Exchange Operation of Complex Colors           ( , ) ( , ) ( , ) ( , )  Color-exchange operation preserves the consistency of vertex constraint

  12. Kempe Walks Eliminate 2 Variables

  13. Variable Elimination by Kempe Walks Exhaustively eliminate variables by Kempe walks  Proper 3-edge-coloring if no variables remaining, or  All remaining variables are contained in odd cycles T. T. Lee, Y. Wan, H. Guan. Randomized ∆ -edge coloring via exchanges of complex colors, International Journal of Computer Mathematics 90 (2013), 228-245 .

  14. One-step move on Kempe path Case Next Step Operations Results KW1 step forward eliminate two KW2 variables eliminate one KW3 variable eliminate one KW4 variable

  15. Limitation of Kempe Walks  Kempe walks can only apply to two-colored sub-graphs H .  Kempe walks cannot change the topology of any two-colored sub-graphs H .  Variables are trapped within fixed two-colored odd cycles.

  16. Color Inversion of Complex Colors Color Inversion

  17. Outline  Operations of Complex Colors  Decomposition of Configurations  Solvability of Configurations  Generalized Petersen Configuration  Three-Edge-Coloring Theorem  Graph Theory versus Euclidean Geometry  Conclusions

  18. Petersen Matching and Configuration  Edges not in the perfect matching form a set of disjoint cycles, called Tait cycles .  Configuration 𝑈(𝐻) : assigning color 𝑑 to the edges in the perfect matching, and color 𝑏 or 𝑐 to the links in Tait cycles.  Every odd (𝑏, 𝑐) Tait cycle contains exactly one (𝑏, 𝑐) -variable.

  19. Color Configurations of A Cubic Plane Graph Two disjoint odd (𝑏, 𝑐) cycles. (𝑏, 𝑐) , (𝑐, 𝑑) , (𝑏, 𝑑) even cycles.

  20. Decomposition of Configuration Maximal Two-Colored Sub-graphs:  Locking Cycle : odd (𝑏, 𝑐) cycle contains one (𝑏, 𝑐) - variable.  Resolution Cycle : (𝑏, 𝑑) or (𝑐, 𝑑) even cycle.  Exclusive Chain : (𝑏, 𝑑) or (𝑐, 𝑑) open path connecting two (𝑏, 𝑐) -variables.

  21. Locking Cycle Configuration 𝑈(𝐻) . Two locking (𝑏, 𝑐) cycles.

  22. Essential Resolution Cycle The (𝑏, 𝑑) exclusive chain. An essential (𝑏, 𝑑) cycle. Two even (𝑏, 𝑐) cycles after negating (𝑏, 𝑑) cycle.

  23. Nonessential Resolution Cycle The (𝑐, 𝑑) exclusive chain. A nonessential (𝑐, 𝑑) cycle. Two odd (𝑏, 𝑐) cycles after negating (𝑐, 𝑑) cycle.

  24. State Transitions within A Configuration 𝑈 𝐻  State transitions of 𝑈 𝐻 :  Negate any 𝑏, 𝑐 cycle, either even or odd.  Move any 𝑏, 𝑐 -variable within its locking cycle.  State transitions retain the sub-graphs of all 𝑏, 𝑐 cycles intact, but change 𝑏, 𝑑 and 𝑐, 𝑑 exclusive chains and resolution cycles.

  25. Outline  Operations of Complex Colors  Decomposition of Configurations  Solvability of Configurations  Generalized Petersen Configuration  Three-Edge-Coloring Theorem  Graph Theory versus Euclidean Geometry  Conclusions

  26. Solvability of Configurations  A state 𝜊 ∈ 𝑇 𝑈(𝐻) is solvable if one of the (𝑏, 𝑑) or (𝑐, 𝑑) cycle in the state 𝜊 is essential. Otherwise, the state 𝜊 ∈ 𝑇 𝑈(𝐻) is unsolvable .  The configuration 𝑈(𝐻) is solvable if one of the state 𝜊 ∈ 𝑇 𝑈(𝐻) is solvable. Otherwise, the configuration 𝑈(𝐻) is unsolvable if all states are unsolvable.

  27. Transitions of A Configuration  Local operation : (𝑏, 𝑐) color exchanges, move a state to another state within the same configuration 𝑈(𝐻) .  Global operation : (𝑏, 𝑑) and (𝑐, 𝑑) color exchanges, transform configuration 𝑈(𝐻) into another configuration 𝑈′(𝐻) .

  28. Transition Diagram of Configurations

  29. Petersen Graph

  30. A Configuration of Petersen Graph  A configuration contains 2 variables A state 𝜊 1 of Petersen graph. (𝑏, 𝑑) sub-graph of 𝜊 1 . (𝑐, 𝑑) sub-graph of 𝜊 1 .

  31. A Configuration of Petersen Graph  Another state of the same configuration A state 𝜊 2 of Petersen graph. (𝑏, 𝑑) sub-graph of 𝜊 2 . (𝑐, 𝑑) sub-graph of 𝜊 2 .

  32. Tutte’s Conjecture  Tutte (1966): Every snark has the Petersen graph as a graph minor.  Neil Robertson and Robin Thomas announced in 1996 that they proved this conjecture, but did not publish the result.  This conjecture implies the four color theorem. W.T. Tutte. On the algebraic theory of graph colorings. J. of Combinatorial Theory 1 (1966), 15 – 50.

  33. Contract Flower Snark to Petersen Graph

  34. Contract Loupekine’s First Snark

  35. Contract Double Star Snark

  36. Petersen Graph as Graph Minor Petersen graph as graph minor is NOT a characterization of a snark

  37. Proposition of Unsolvability  A bridgeless cubic graph 𝐻(𝑊, 𝐹) is a class 2 graph if and only if 𝐻 has a closed set of unsolvable configurations.

  38. Outline  Operations of Complex Colors  Decomposition of Configurations  Solvability of Configurations  Generalized Petersen Configuration  Three-Edge-Coloring Theorem  Graph Theory versus Euclidean Geometry  Conclusions

  39. Generalized Petersen Configuration A generalized Petersen configuration 𝑄(𝐻) satisfies:  The configuration 𝑄(𝐻) contains two (𝑏, 𝑐) - variables.  The two 𝑏, 𝑐 -variables are on the boundary of a pentagon in some state 𝜊 of 𝑄(𝐻) .

  40. Generalized Petersen Configuration

  41. Generalized Petersen Configuration

  42. Resolution Cycle of Generalized Petersen Configuration

  43. Proposition of Solvability  Every generalized Petersen configuration 𝑄(𝐻) of a bridgeless cubic plane graph 𝐻(𝑊, 𝐹) is solvable.  Verified by more than 100,000 instances generated by computer.  Don’t have a logical proof of this assertion.

  44. 𝑄(𝐻) with 284 solvable states

  45. An Unsolvable Configuration  The two (𝑏, 𝑐) -variables are NOT on the boundary of the same pentagon in any states 𝜊 .

  46. Negating (a,c) Cycle

  47. Negating (b,c) Cycle

  48. Outline  Operations of Complex Colors  Decomposition of Configurations  Solvability of Configurations  Generalized Petersen Configuration  Three-Edge-Coloring Theorem  Graph Theory versus Euclidean Geometry  Conclusions

  49. Three-edge Coloring Theorem  Lemma 1 : The girth of a bridgeless cubic plane graph 𝐻(𝑊, 𝐹) is less than or equal to 5.  Lemma 2 : Any face of a bridgeless cubic plane graph 𝐻(𝑊, 𝐹) has at least one admissible edge.  Theorem : Every bridgeless cubic plane graph 𝐻(𝑊, 𝐹) has a 3-edge-coloring.

  50. The Girth of G Equals 3

  51. The Girth of G Equals 4

  52. The Girth of G Equals 4

Recommend


More recommend