A master bijection for planar maps and its applications Olivier - - PowerPoint PPT Presentation

a master bijection for planar maps and its applications
SMART_READER_LITE
LIVE PREVIEW

A master bijection for planar maps and its applications Olivier - - PowerPoint PPT Presentation

A master bijection for planar maps and its applications Olivier Bernardi (MIT) Joint work with Eric Fusy (CNRS/LIX) UCLA, March 2011 Planar maps. Definition A planar map is a connected planar graph embedded in the sphere considered up to


slide-1
SLIDE 1

A master bijection for planar maps and its applications

UCLA, March 2011 Olivier Bernardi (MIT) Joint work with ´ Eric Fusy (CNRS/LIX)

slide-2
SLIDE 2

A planar map is a connected planar graph embedded in the sphere considered up to continuous deformation. Planar maps. Definition

= =

slide-3
SLIDE 3
  • Algorithmic applications: efficient encoding of meshed surfaces.
  • Probability and Physics: random lattices, random surfaces.

Appears courtesy to Wikipedia Appears courtesy to G. Chapuy

Planar maps. Motivations

  • Representation Theory: factorization problems.
slide-4
SLIDE 4
  • Generating functions [Tutte 63]

Recursive description of maps recurrences.

  • Matrix Integrals [’t Hooft 74]

Feynmann Diagram ≈ maps.

  • Representation theory [Jackson 90]

Factorizations of permutations ≈ maps.

  • Bijections [Cori-Vauquelin 91, Schaeffer 98]

Maps decorated trees. Planar maps. Methods

slide-5
SLIDE 5
  • Triangulations (2n faces)
  • Quadrangulations (n faces)
  • Bipartite maps (ni faces of degree 2i)

Planar maps. Counting results Common features: Algebraic generating function. Asymptotic ∼ κ n−5/2Rn.

Loopless: 2n (n + 1)(2n + 1) 3n n

  • Simple:

1 n(2n − 1) 4n − 2 n − 1

  • General:

2 · 3n (n + 1)(n + 2) 2n n

  • Simple:

2 n(n + 1) 3n n − 1

  • 2 · ( i ni)!

(2 + (i − 1)ni)!

  • i

1 ni! 2i − 1 i ni

Planar maps. Counting results

slide-6
SLIDE 6
  • Triangulations (2n faces)
  • Quadrangulations (n faces)
  • Bipartite maps (ni faces of degree 2i)

[Sc97] [BoDiGu04] Loopless: 2n (n + 1)(2n + 1) 3n n

  • Simple:

1 n(2n − 1) 4n − 2 n − 1

  • General:

2 · 3n (n + 1)(n + 2) 2n n

  • Simple:

2 n(n + 1) 3n n − 1

  • 2 · ( i ni)!

(2 + (i − 1)ni)!

  • i

1 ni! 2i − 1 i ni

Planar maps. Counting results

[ScPo02] [B.07] [Sc97] [Sc98] [PoSc06] [FuPoSc08] [Sc98] [Fu07] Bo: Bouttier Di: Di Francesco Fu: Fusy Gu: Guitter Po: Poulalhon Sc: Schaeffer Abbreviations:

slide-7
SLIDE 7

Outline Benefits:

  • Simplifies/unifies the proofs.
  • Helps to find bijections for new classes of maps.

Describe a master bijection for planar maps which generalizes all the known bijections (of the red type).

slide-8
SLIDE 8

Outline

  • 1. Master bijection between a class of oriented maps and a class
  • f bicolored decorated trees.
  • 2. Specializations to classes of maps (via canonical orientations).

Degree of the faces

Girth 1 2 3 4

1 2 3 4 5

6

[FuPoSc08] [Sc98] [Sc98,BoDiGu04] [PoSc02]

7 8

slide-9
SLIDE 9

Master bijection

(simplified version)

slide-10
SLIDE 10

A plane orientation is a oriented map drawn in the plane. Plane orientations.

slide-11
SLIDE 11

A plane orientation is a oriented map drawn in the plane. Plane orientations. We consider the set O of plane orientations which are:

  • minimal: there is no counterclockwise directed cycle,
  • accessible: any internal vertex can be reached from an external

vertex, and have external vertices of indegree 1.

slide-12
SLIDE 12

A mobile is a plane tree with vertices properly colored in black and white, together with buds (half-edges) incident to black vertices. Mobiles The excess is the number of edges minus the number of buds.

slide-13
SLIDE 13

Mapping Φ for a plane orientation O in O:

  • Return the external face.
  • Place a black vertex vf inside each internal face f.

Turning clockwise around f, draw an edge/bud from vf to the corners following forward/backward edges

  • Erase the edges of O and its external vertices.

Master bijection

slide-14
SLIDE 14

Master bijection Theorem [B.,Fusy]: The mapping Φ is a bijection between the set O

  • f plane orientations and the set of mobiles with negative excess.

Moreover, indegree of internal vertices ← → degree of white vertices degree of internal faces ← → degree of black vertices degree of external face ← → - excess

slide-15
SLIDE 15

Strategy for counting a class C of maps:

  • Define a “canonical” orientation in O for each map in C.

This identifies C with a subset OC of O.

  • Characterize (and count) the set of mobiles associated to OC via

the master bijection Φ. Using the master bijection to count classes of maps?

slide-16
SLIDE 16

Using the master bijection to count classes of maps? How to define an orientation in the set O (orientations minimal, ac- cessible with external vertices of indegree 1)?

slide-17
SLIDE 17

Using the master bijection to count classes of maps? Let G = (V, E) be a map and let α : V → N. How to define an orientation in the set O (orientations minimal, ac- cessible with external vertices of indegree 1)? Fact 1: If there exists an orientation of G with indegrees α(v), then there exists a unique minimal one. ⇒ Orientations O can be defined by specifying the indegrees.

slide-18
SLIDE 18

Using the master bijection to count classes of maps? Let G = (V, E) be a map and let α : V → N. How to define an orientation in the set O (orientations minimal, ac- cessible with external vertices of indegree 1)? Fact 2: There exists an orientation with indegrees α(v) if and only if

  • |E| =

v∈V α(v)

  • for any U ⊂ V ,

|EU| ≤

v∈U α(v).

Fact 1: If there exists an orientation of G with indegrees α(v), then there exists a unique minimal one. ⇒ Orientations O can be defined by specifying the indegrees.

slide-19
SLIDE 19

Using the master bijection to count classes of maps? Let G = (V, E) be a map and let α : V → N. How to define an orientation in the set O (orientations minimal, ac- cessible with external vertices of indegree 1)? Fact 2: There exists an orientation with indegrees α(v) if and only if

  • |E| =

v∈V α(v)

  • for any U ⊂ V ,

|EU| ≤

v∈U α(v).

Moreover, the orientation is accessible if and only if |EU| <

v∈U α(v)

whenever U does not contain all the external vertices. Fact 1: If there exists an orientation of G with indegrees α(v), then there exists a unique minimal one. ⇒ Orientations O can be defined by specifying the indegrees.

slide-20
SLIDE 20

Using the master bijection to count classes of maps? Conclusion: For a map G = (V, E), one can define an orientation in O by specifying a function α : V → N such that:

  • α(v) = 1 for every external vertex v,
  • |E| =

v∈V α(v),

  • ∀ U ⊂ V , |EU| ≤

v∈U α(v)

with strict inequality if U does not contain all external vertices. (*)

slide-21
SLIDE 21

Using the master bijection to count classes of maps? Conclusion: For a map G = (V, E), one can define an orientation in O by specifying a function α : V → N such that:

  • α(v) = 1 for every external vertex v,
  • |E| =

v∈V α(v),

  • ∀ U ⊂ V , |EU| ≤

v∈U α(v)

with strict inequality if U does not contain all external vertices. (*) Remark: Specifying an orientation in O by the indegrees of internal vertices is also convenient in view of applying the master bijection Φ: indegrees of internal vertices ← → degrees of white vertices.

slide-22
SLIDE 22

Example of specialization: simple triangulations

Degree of faces

Girth 1 2 3 4

1 2 3 4 5

6 7

slide-23
SLIDE 23

Proof: The numbers v, e, f of vertices edges and faces satisfy:

  • Incidence relation: 3f = 2e.
  • Euler relation: v − e + f = 2.
  • Triangulations

Fact: A triangulation with n internal vertices has 3n internal edges.

slide-24
SLIDE 24

Triangulations Natural candidate for indegree function: α : v →

  • 3 if v internal

1 if v external . Fact: A triangulation with n internal vertices has 3n internal edges. 1 1 1 3 3 3 3

slide-25
SLIDE 25

Triangulations New (easier) proof: Use the Euler relation + the incidence relation to show that α satisfies condition (*).

  • Thm [Schnyder 89]: A triangulation admits an orientation with in-

degree function α if and only if it is simple.

slide-26
SLIDE 26

Triangulations ⇒ The class T of simple triangulations is indentified with the class of plane orientation OT ⊂ O with faces of degree 3, and internal vertices

  • f indegree 3.

Thm [recovering FuPoSc08]: By specializing the master bijection Φ to OT one obtains a bijection between simple triangulations and mobiles such that • black vertices have degree 3

  • white vertices have degree 3
  • the excess is −3 (redundant).

Thm [Schnyder 89]: A triangulation admits an orientation with in- degree function α if and only if it is simple.

slide-27
SLIDE 27

Triangulations Counting: The generating function of mobiles with vertices of degree 3 rooted on a white corner is T(x) = U(x)3, where U(x) = 1 + xU(x)4. Consequently, the number of (rooted) simple triangulations with 2n faces is 1 n(2n − 1) 4n − 2 n − 1

  • .
slide-28
SLIDE 28

More specializations: d-angulations of girth d.

Degree of faces

Girth 1 2 3 4

1 2 3 4 5

6 7

slide-29
SLIDE 29

d-angulations of girth d Fact: A d-angulation with (d−2)n internal vertices has dn internal edges.

d = 5

slide-30
SLIDE 30

d-angulations of girth d Fact: A d-angulation with (d−2)n internal vertices has dn internal edges. Natural candidate for indegree function: α : v →

  • d/(d − 2) if v internal

1 if v external . . .

d = 5

slide-31
SLIDE 31

d-angulations of girth d Idea: We can look for an orientation of (d−2)G with indegree function α : v →

  • d if v internal

1 if v external . Fact: A d-angulation with (d−2)n internal vertices has dn internal edges. 5 5 5 5 5 5

d = 5

slide-32
SLIDE 32

d-angulations of girth d Proof: Use the Euler relation + incidence relation to show that α satisfies condition (*).

  • Thm [B., Fusy]: Let G be a d-angulation. Then (d−2)G admits an
  • rientation with indegree function α if and only if G has girth d.

2 1 2 1 2 1 2 1

d = 5

slide-33
SLIDE 33

There are now white-white edges in the mobile, with two positive weights summing to d − 2. Master bijection for fractional orientations

2 1 2 1 1 2 21 2 1 2 1 1 2 21 2 1 2 1 1 2 21

slide-34
SLIDE 34

Master bijection for fractional orientations Theorem [B.,Fusy]: Φ is a bijection between the set Ok of plane k- fractional orientations and the set of mobiles (with white-white edges having weights summing to k) with negative excess. Moreover, degree of internal faces ← → degree of black faces indegree of internal vertices ← → indegree of white vertices degree of external face ← → - excess

2 1 2 1 1 2 21 2 1 2 1 1 2 21 2 1 2 1 1 2 21

slide-35
SLIDE 35

d-angulations of girth d ⇒ The class Td of d-angulations of girth d can be indentified with the class of fractional orientations Ωd ⊂ Od

− 2 with faces of degree d, and

internal vertices of indegree d.

2 1 2 1 2 1 2 1 2 1 1 2 21 21

Thm [B., Fusy]: Let G be a d-angulation. Then (d−2)G admits an

  • rientation with indegree function α if and only if G has girth d.
slide-36
SLIDE 36

d-angulations of girth d

  • black vertices have degree d
  • white vertices have indegree d
  • the excess is −d (redundant).

2 1 2 1 2 1 2 1 2 1 1 2 21 21

Thm [B., Fusy]: By specializing the master bijection one obtains a bijection between d-angulations of girth d and mobiles (with white-white edges having weights summing to d − 2) such that Thm [B., Fusy]: Let G be a d-angulation. Then (d−2)G admits an

  • rientation with indegree function α if and only if G has girth d.
slide-37
SLIDE 37

d-angulations of girth d: counting Thm[B., Fusy]: Let W0, W1, . . . , Wd−2 be the power series in x defined by: Wd−2 = x(1 + W0)d−1 and ∀j < d − 2, Wj =

  • r
  • i1,...,ir>0

i1+···+ir=j+2

Wi1 · · · Wir. The generating function Fd of d-angulations of girth d satisfies Fd = Wd−2 −

d−3

  • i=0

WiWd−2−i, and F ′

d(x) = (1 + W0)d.

slide-38
SLIDE 38

d-angulations of girth d: counting Thm[B., Fusy]: Let W0, W1, . . . , Wd−2 be the power series in x defined by: Wd−2 = x(1 + W0)d−1 and ∀j < d − 2, Wj =

  • r
  • i1,...,ir>0

i1+···+ir=j+2

Wi1 · · · Wir. The generating function Fd of d-angulations of girth d satisfies Fd = Wd−2 −

d−3

  • i=0

WiWd−2−i, and F ′

d(x) = (1 + W0)d.

Example d=5: W3 = x(1 + W0)4 W0 = W 2

1 + W2

W1 = W 3

1 + 2W1W2 + W3

W2 = W 4

1 + 3W 2 1 W2 + 2W1W3 + W 2 2

slide-39
SLIDE 39

More specializations: Maps of girth d.

Degree of faces

Girth 1 2 3 4

1 2 3 4 5

6 7

slide-40
SLIDE 40

Maps of girth d We consider maps of girth d with external face of degree d.

slide-41
SLIDE 41

Maps of girth d Orientation: It is possible to characterize girth d by looking at

  • rientations on (d−2)G + 2Q.

We consider maps of girth d with external face of degree d. Thm [B., Fusy]. Let G be a map with external face of degree d. The map G has girth d if and only if (d−2)G + 2Q admits an orientation such that inner vertices of G have indegree d and vertices of Q in faces

  • f degree δ have indegree δ + d.

1 1 1 1 1 1

slide-42
SLIDE 42

Maps of girth d Orientation: It is possible to characterize girth d by looking at

  • rientations on (d−2)G + 2Q.

We consider maps of girth d with external face of degree d.

  • Bijection. The orientation can be transfered to the map G and gives

a canonical orientation in O. One can characterize the mobiles obtained by the master bijection. (There are weights on the half-edges of the mobile. There are white- white and black-black edges. . . )

slide-43
SLIDE 43

Maps of girth d Orientation: It is possible to characterize girth d by looking at

  • rientations on (d−2)G + 2Q.

We consider maps of girth d with external face of degree d.

  • Bijection. The orientation can be transfered to the map G and gives

a canonical orientation in O. One can characterize the mobiles obtained by the master bijection. Counting: The generating function Fd(xd, xd+1, . . .) of maps of girth d and outer degree d is Fd = Wd−2 −

d−3

  • j=−2

WjWd−2−j. Where ∀j ∈ [−2..d−3], Wj =

  • r
  • i1,...,ir>0

i1+···+ir=j+2

Wi1 · · · Wir, and ∀j ∈ [d−2..d], Wj = [uj+1]

  • i

xi (u + uW0 + W−1 + u−1)i−1.

slide-44
SLIDE 44

Closed formulas Prop [B.,Fusy]: The number of simple bipartite maps with ni faces

  • f degree 2i is

2((i + 1) ni − 3)! ( ini − 1)!

  • i≥2

1 ni! 2i − 1 i + 1 ni 2 ( i ni)! ((i − 1)ni + 2)!

  • i≥1

1 ni! 2i − 1 i ni This can be compared with the formula obtained by Schaeffer (97) for unconstrained bipartite maps:

slide-45
SLIDE 45

Related results and extensions

slide-46
SLIDE 46

Orientations for d-angulations as Schnyder decompositions Thm [B., Fusy] If G is a d-angulation of girth d, then one can partition the internal edges of (d − 2)G into d forests such that:

  • each forest covers the internal vertices and d − 2 external vertices,
  • the forests cross each other in a specific way.
slide-47
SLIDE 47

Further extensions

  • 1. The master bijection generalizes two other known bijections:

irreducible triangulations [FuPoSc08] and quadrangulations [Fu09].

Degree of the faces

Girth 1 2 3 4

1 2 3 4 5

6

[FuPoSc08] [Sc98] [Sc98,BoDiGu04] [PoSc02]

7 8

slide-48
SLIDE 48

Further extensions

  • 2. The master bijection can be extended to maps of higher genera.

[B., Chapuy 10].

  • 1. The master bijection generalizes two other known bijections:

irreducible triangulations [FuPoSc08] and quadrangulations [Fu09].

slide-49
SLIDE 49

Further extensions

  • 2. The master bijection can be extended to maps of higher genera.

[B., Chapuy 10].

  • 1. The master bijection generalizes two other known bijections:

irreducible triangulations [FuPoSc08] and quadrangulations [Fu09]. 3. The master bijection also generalizes some bijections for maps with matter.

slide-50
SLIDE 50

Thanks.

On the ArXiv:

  • A bijection for triangulations, quadrangulations, pentagulations, etc.
  • Counting maps by girth and degree I. (number II in preparation)

Related:

  • Schnyder decompositions for regular plane graphs and application to

drawing.

  • A bijection for covered maps, or a shortcut between Harer-Zagier’s and

Jackson’s formulas, with Guillaume Chapuy (to appear in JCTA).

slide-51
SLIDE 51

And now for something completely different. . . Schnyder woods and drawing algorithms

slide-52
SLIDE 52

A Schnyder wood of a triangulation is a partition of the internal edges into 3 trees,

  • each tree covers all the internal vertices and one external vertex,
  • the trees cross in a specific manner around each internal vertex.

Schnyder woods for triangulations

slide-53
SLIDE 53

Schnyder woods for triangulations Theorem [Schnyder 89,90].

  • Any simple triangulation T admits a Schnyder wood.
  • The Schnyder woods of T are in bijection with 3-orientations

(i.e. there exists a unique way of coloring each 3-orientation).

  • It is possible to find a Schnyder wood of T in linear time.
  • If T has n vertices, Schnyder woods can be used to compute a

planar straight-line drawing of T on the grid [2n] × [2n].

slide-54
SLIDE 54

Prop: Any d-angulation of girth d admits a d/(d-2)-orientation (orientation of (d − 2)G such that the indegree of each vertex is d). Schnyder decompositions for d-angulations 5 5 5 5 5 5

slide-55
SLIDE 55

Schnyder decompositions for d-angulations A Schnyder decomposition is a partition of (d − 2)E into d forests,

  • each forest covers all the internal vertices and d − 2 external

vertices,

  • the trees cross in a specific manner around each internal vertex.
slide-56
SLIDE 56

Schnyder decompositions for d-angulations Theorem [B. Fusy].

  • Any simple d-angulation T admits a Schnyder decomposition.
  • The Schnyder woods of T are in bijection with its d/(d − 2)-
  • rientations

(i.e. there exists a unique way of coloring each orientation).

  • A Schnyder decomposition can be computed in polynomial time.
slide-57
SLIDE 57

Dual of Schnyder decompositions (for d-regular graphs) A dual Schnyder decomposition is a collection of d spanning trees,

  • each internal edge has two trees in opposite directions,
  • the trees 1, 2, . . . , d arrive in this clockwise order at each vertex.

v∗

3 2 d d−1 1

slide-58
SLIDE 58

Dual of Schnyder decompositions (for d-regular graphs) A dual Schnyder decomposition is a collection of d spanning trees,

  • each internal edge has two trees in opposite directions,
  • the trees 1, 2, . . . , d arrive in this clockwise order at each vertex.

Moreover, when d is even one can impose that for each edge e, the two trees using a of different parity. Trees 1 and 3 Trees 2 and 4

slide-59
SLIDE 59

Application: drawing algorithm for 4-regular graphs Algorithm:

  • For each vertex v, compute x(v)= number of faces on the left of v,

and y(v)=number of faces below v.

  • Draw each vertex v at the point (x(v), y(v)) of the grid [n] × [n],

and draw each edge as a line segment. v x(v) faces

slide-60
SLIDE 60

Application: drawing algorithm for 4-regular graphs Thm [B. Fusy]. The algorithm gives a straight-line planar drawing. The same vertex placement also gives an orthogonal planar drawing (with 1 bend per edge).