Problems of Enumeration and Realizability on Matroids, Simplicial - - PowerPoint PPT Presentation

problems of enumeration and realizability on matroids
SMART_READER_LITE
LIVE PREVIEW

Problems of Enumeration and Realizability on Matroids, Simplicial - - PowerPoint PPT Presentation

Problems of Enumeration and Realizability on Matroids, Simplicial Complexes, and Graphs Yvonne Kemper August 6, 2014 Yvonne Kemper PoEaRoMSCaG In Honor of a Diagram g s r a h p Yvonne Kemper PoEaRoMSCaG Wait, what were those things


slide-1
SLIDE 1

Problems of Enumeration and Realizability

  • n Matroids, Simplicial Complexes, and Graphs

Yvonne Kemper August 6, 2014

Yvonne Kemper PoEaRoMSCaG

slide-2
SLIDE 2

In Honor of a Diagram

g r a p h s

Yvonne Kemper PoEaRoMSCaG

slide-3
SLIDE 3

Wait, what were those things again?

Definition

A graph G = (V , E) is a set of vertices V = {v1, . . . , vn} and a set of edges E = {vivj : vi, vj ∈ V }.

Example

Here is a graph! 1 2 3 4 G = (V , E) = ({1, 2, 3, 4}, {12, 13, 14, 23, 24})

Yvonne Kemper PoEaRoMSCaG

slide-4
SLIDE 4

In Honor of a Diagram

g r a p h s

Yvonne Kemper PoEaRoMSCaG

slide-5
SLIDE 5

In Honor of a Diagram

g r a p h s m a t r

  • i

d s

Yvonne Kemper PoEaRoMSCaG

slide-6
SLIDE 6

Wait, what were those things again?

Definition

A matroid M = (E(M), I(M)) consists of a ground set E(M) and a family of subsets I(M) ⊆ 2E(M) called independent sets such that (1) ∅ ∈ I; (2) if I ∈ I and J ⊂ I, then J ∈ I; and (3) if I, J ∈ I, and |J| < |I|, then there exists some e ∈ I \ J such that J ∪ {e} ∈ I.

Yvonne Kemper PoEaRoMSCaG

slide-7
SLIDE 7

Wait, what were those things again?

Example

Here is a graph matroid! 3 1 4 5 2 M = (E(M), I(M)) = ({1, 2, 3, 4, 5}, {∅, 1, 2, 3, 4, 5, 12, 13, 14, 15, 23, 24, 25, 34, 35, 45, 124, 125, 134, 135, 145, 234, 235, 245})

Yvonne Kemper PoEaRoMSCaG

slide-8
SLIDE 8

Wait, what were those things again?

Example

Here is a graph matroid! 3 1 4 5 2 M = (E(M), I(M)) = ({1, 2, 3, 4, 5}, {∅, 1, 2, 3, 4, 5, 12, 13, 14, 15, 23, 24, 25, 34, 35, 45, 124, 125, 134, 135, 145, 234, 235, 245})

Yvonne Kemper PoEaRoMSCaG

slide-9
SLIDE 9

In Honor of a Diagram

g r a p h s m a t r

  • i

d s

Yvonne Kemper PoEaRoMSCaG

slide-10
SLIDE 10

In Honor of a Diagram

g r a p h s m a t r

  • i

d s s i m p l i c i a l c

  • m

p l e x e s

Yvonne Kemper PoEaRoMSCaG

slide-11
SLIDE 11

Wait, what were those things again?

Definition

An (abstract) simplicial complex ∆ on a vertex set V is a set

  • f subsets of V . These subsets are called the faces of ∆, and we

require that (1) for all v ∈ V , {v} ∈ ∆, and (2) for all F ∈ ∆, if G ⊆ F, then G ∈ ∆.

Yvonne Kemper PoEaRoMSCaG

slide-12
SLIDE 12

Wait, what were those things again?

Example

Here is a graph matroid simplicial complex! 1 2 3 4 ∆ = (V , F) = ({1, 2, 3, 4}, {∅, 1, 2, 3, 4, 12, 13, 14, 23, 24})

Yvonne Kemper PoEaRoMSCaG

slide-13
SLIDE 13

Wait, what were those things again?

Example

Here is a graph matroid simplicial complex! 1 2 3 4 ∆ = (V , F) = ({1, 2, 3, 4}, {∅, 1, 2, 3, 4, 12, 13, 14, 23, 24})

Yvonne Kemper PoEaRoMSCaG

slide-14
SLIDE 14

In Honor of a Diagram

g r a p h s m a t r

  • i

d s s i m p l i c i a l c

  • m

p l e x e s

Yvonne Kemper PoEaRoMSCaG

slide-15
SLIDE 15

In Honor of a Diagram

g r a p h s m a t r

  • i

d s s i m p l i c i a l c

  • m

p l e x e s p r

  • b

l e m s

  • n

m a t r

  • i

d s , s i m p l i c i a l c

  • m

p l e x e s , a n d g r a p h s

Yvonne Kemper PoEaRoMSCaG

slide-16
SLIDE 16

In Honor of a Diagram

g r a p h s m a t r

  • i

d s s i m p l i c i a l c

  • m

p l e x e s p r

  • b

l e m s

  • f

e n u m e r a t i

  • n
  • n

m a t r

  • i

d s , s i m p l i c i a l c

  • m

p l e x e s , a n d g r a p h s e n u m e r a t i

  • n

Yvonne Kemper PoEaRoMSCaG

slide-17
SLIDE 17

In Honor of a Diagram

g r a p h s m a t r

  • i

d s s i m p l i c i a l c

  • m

p l e x e s e n u m e r a t i

  • n

r e a l i z a b i l i t y p r

  • b

l e m s

  • f

e n u m e r a t i

  • n

a n d r e a l i z a b i l i t y

  • n

m a t r

  • i

d s , s i m p l i c i a l c

  • m

p l e x e s , a n d g r a p h s

Yvonne Kemper PoEaRoMSCaG

slide-18
SLIDE 18

In Honor of a Diagram

g r a p h s m a t r

  • i

d s s i m p l i c i a l c

  • m

p l e x e s e n u m e r a t i

  • n

r e a l i z a b i l i t y p r

  • b

l e m s

  • f

e n u m e r a t i

  • n

a n d r e a l i z a b i l i t y

  • n

m a t r

  • i

d s , s i m p l i c i a l c

  • m

p l e x e s , a n d g r a p h s

Lookin’ good!

Yvonne Kemper PoEaRoMSCaG

slide-19
SLIDE 19

The Problems Three

◮ h-Vectors of Small Matroids ◮ Flows on Simplicial Complexes ◮ Polytopal Embeddings of Cayley Graphs

Yvonne Kemper PoEaRoMSCaG

slide-20
SLIDE 20

h-Vectors of Small Matroids

Yvonne Kemper PoEaRoMSCaG

slide-21
SLIDE 21

Simplicial Complexes: a Few More Definitions

Definition

The dimension of a face F is |F| − 1, and the dimension of ∆ is d = max{|F| : F ∈ ∆} − 1.

Definition

A simplicial complex is pure if all maximal elements of ∆ have the same cardinality. In this case, a facet is a maximal face, a ridge is a face of one dimension lower.

Yvonne Kemper PoEaRoMSCaG

slide-22
SLIDE 22

Faces: A Natural Quantity to Measure

◮ The f -vector of a simplicial complex ∆, dim ∆ = d − 1, is

f (∆) := (f−1(∆), f0(∆), . . . , fd−1(∆)), where fi(∆) := |{F ∈ ∆ : dim F = i}|.

Yvonne Kemper PoEaRoMSCaG

slide-23
SLIDE 23

Faces: A Natural Quantity to Measure

◮ The f -vector of a simplicial complex ∆, dim ∆ = d − 1, is

f (∆) := (f−1(∆), f0(∆), . . . , fd−1(∆)), where fi(∆) := |{F ∈ ∆ : dim F = i}|.

◮ The h-vector, h(∆) := (h0(∆), . . . , hd(∆)), is given by: d

  • j=0

hj(∆)λj =

d

  • i=0

fi−1(∆)λi(1 − λ)d−i.

Yvonne Kemper PoEaRoMSCaG

slide-24
SLIDE 24

Characterizations of f - and h-Vectors

Definition

Given two integers k, i > 0, write k = ni i

  • +

ni−1 i − 1

  • + · · · +

nj j

  • ,

where ni > ni−1 > · · · > nj ≥ j ≥ 1. Define k(i) = ni i + 1

  • +

ni−1 i

  • + · · · +

nj j + 1

  • .

Theorem (Sch¨ utzenberger, Kruskal, Katona)

A vector (1, f0, f1, . . . , fd−1) ∈ Zd+1 is the f -vector of some (d − 1)-dimensional simplicial complex ∆ if and only if 0 < fi+1 ≤ f (i+1)

i

, 0 ≤ i ≤ d − 2.

Yvonne Kemper PoEaRoMSCaG

slide-25
SLIDE 25

Other Characterizations?

Question

Can we characterize subclasses of simplicial complexes?

Example

◮ Cohen-Macaulay complexes ◮ Flag complexes ◮ Shifted complexes ◮ Independence complexes of matroids

Yvonne Kemper PoEaRoMSCaG

slide-26
SLIDE 26

Matroid Complexes: An Example

Let M be given by: I(M) = {∅, 1, 2, 3, 4, 5, 12, 13, 14, 15, 23, 24, 25, 34, 35, 45, 124, 125, 134, 135, 145, 234, 235, 245} The corresponding complex: 1 2 3 4 5 Then: f (M) = (1, 5, 10, 8) and h(M) = (1, 2, 3, 2).

Yvonne Kemper PoEaRoMSCaG

slide-27
SLIDE 27

O-Sequences

◮ A non-empty set of monomials M is a multicomplex if

m ∈ M and n|m ⇒ n ∈ M.

◮ A sequence h = (h0, h1, . . . , hd) of integers is an O-sequence

if there exists a multicomplex with precisely hi monomials of degree i.

◮ An O-sequence is pure if all maximal elements have the same

degree.

Yvonne Kemper PoEaRoMSCaG

slide-28
SLIDE 28

O-Sequences

◮ A non-empty set of monomials M is a multicomplex if

m ∈ M and n|m ⇒ n ∈ M.

◮ A sequence h = (h0, h1, . . . , hd) of integers is an O-sequence

if there exists a multicomplex with precisely hi monomials of degree i.

◮ An O-sequence is pure if all maximal elements have the same

degree.

Example

Let M = {1, x1, x2, x1x2, x2

1, x2 2, x1x2 2, x2 1x2}. Then, the

corresponding (pure) O-sequence is: O(M) = (1, 2, 3, 2).

Yvonne Kemper PoEaRoMSCaG

slide-29
SLIDE 29

Stanley’s Conjecture

Conjecture (Stanley, 1977)

The h-vector of a matroid complex is a pure O-sequence. Little progress was made for twenty years, but since 1997, the conjecture has been proved for matroids which are:

◮ of rank 4 (Klee, Samper), ◮ of rank less than or equal to 3 (Stokes, H´

a et al.),

◮ cographic (Biggs, Merino), ◮ lattice-path (Schweig), ◮ cotransversal (Oh), ◮ paving (Merino, et al.).

Yvonne Kemper PoEaRoMSCaG

slide-30
SLIDE 30

Results

Theorem (De Loera, K., Klee)

◮ Let M be a matroid of rank 2. Then h(M) is a pure

O-sequence.

◮ Let M be a matroid of corank 2. Then h(M) is a pure

O-sequence.

◮ Let M be a matroid of rank d ≥ 4. Then, the subsequence

(1, h1(M), h2(M), h3(M)) of h(M) is a pure O-sequence.

◮ Let M be a matroid of rank 3. Then h(M) is a pure

O-sequence.

◮ Let M be a matroid on at most 9 elements. Then h(M) is a

pure O-sequence.

Yvonne Kemper PoEaRoMSCaG

slide-31
SLIDE 31

An Experimental Result: Matroids on at Most Nine Elements

◮ Royle and Mayhew generated list of all matroids on at most

nine elements - why not check them all?

◮ Used database to generate all h-vectors for these matroids. ◮ Generated list of all possible O-sequences of multicomplexes

(up to maximal degree 9 on at most 9 variables), then checked that every h-vector appeared on this list.

Yvonne Kemper PoEaRoMSCaG

slide-32
SLIDE 32

Example: A Multicomplex in Two Variables

  • 1. Pick a point (a, b) ∈ Z2 on the

hyperplane x + y = r, where r is the rank of the matroid.

Yvonne Kemper PoEaRoMSCaG

slide-33
SLIDE 33

Example: A Multicomplex in Two Variables

Let’s say r = 3.

  • 1. Pick a point (a, b) ∈ Z2 on the

hyperplane x + y = r, where r is the rank of the matroid.

Yvonne Kemper PoEaRoMSCaG

slide-34
SLIDE 34

Example: A Multicomplex in Two Variables

xy 2 (a, b) corresponds to the monomial xay b

  • 1. Pick a point (a, b) ∈ Z2 on the

hyperplane x + y = r, where r is the rank of the matroid.

Yvonne Kemper PoEaRoMSCaG

slide-35
SLIDE 35

Example: A Multicomplex in Two Variables

y 2 y 1 xy x xy 2 M = {1, x, y, xy, y 2, xy 2}

  • 1. Pick a point (a, b) ∈ Z2 on the

hyperplane x + y = r, where r is the rank of the matroid.

  • 2. Add all points in the shadow of

(a, b) to the multicomplex.

Yvonne Kemper PoEaRoMSCaG

slide-36
SLIDE 36

Example: A Multicomplex in Two Variables

y 2 y 1 xy x xy 2 M = {1, x, y, xy, y 2, xy 2}

  • 1. Pick a point (a, b) ∈ Z2 on the

hyperplane x + y = r, where r is the rank of the matroid.

  • 2. Add all points in the shadow of

(a, b) to the multicomplex.

  • 3. Repeat as desired to generate all

possible O-sequences (of rank 3 and corank 2).

Yvonne Kemper PoEaRoMSCaG

slide-37
SLIDE 37

Example: A Multicomplex in Two Variables

y 2 y 1 xy x x2y xy 2 M = {1, x, y, xy, y 2, xy 2}

  • 1. Pick a point (a, b) ∈ Z2 on the

hyperplane x + y = r, where r is the rank of the matroid.

  • 2. Add all points in the shadow of

(a, b) to the multicomplex.

  • 3. Repeat as desired to generate all

possible O-sequences (of rank 3 and corank 2).

Yvonne Kemper PoEaRoMSCaG

slide-38
SLIDE 38

Example: A Multicomplex in Two Variables

x2 y 2 y 1 xy x x2y xy 2 M = {1, x, y, xy, y 2, x2, x2y, y 2}

  • 1. Pick a point (a, b) ∈ Z2 on the

hyperplane x + y = r, where r is the rank of the matroid.

  • 2. Add all points in the shadow of

(a, b) to the multicomplex.

  • 3. Repeat as desired to generate all

possible O-sequences (of rank 3 and corank 2).

Yvonne Kemper PoEaRoMSCaG

slide-39
SLIDE 39

Example: A Multicomplex in Two Variables

x2 y 2 y 1 xy x x2y xy 2 M = {1, x, y, xy, y 2, x2, x2y, y 2}

  • 1. Pick a point (a, b) ∈ Z2 on the

hyperplane x + y = r, where r is the rank of the matroid.

  • 2. Add all points in the shadow of

(a, b) to the multicomplex.

  • 3. Repeat as desired to generate all

possible O-sequences (of rank 3 and corank 2). The corresponding O-sequence is: (1, 2, 3, 2).

Yvonne Kemper PoEaRoMSCaG

slide-40
SLIDE 40

Example: A Multicomplex in Two Variables

x2 y 2 y 1 xy x x2y xy 2 M = {1, x, y, xy, y 2, x2, x2y, y 2}

  • 1. Pick a point (a, b) ∈ Z2 on the

hyperplane x + y = r, where r is the rank of the matroid.

  • 2. Add all points in the shadow of

(a, b) to the multicomplex.

  • 3. Repeat as desired to generate all

possible O-sequences (of rank 3 and corank 2). The corresponding O-sequence is: (1, 2, 3, 2).

3 1 4 5 2

:D

Yvonne Kemper PoEaRoMSCaG

slide-41
SLIDE 41

Future Questions and Directions

◮ Cannot extend results directly, but can we use the geometric

viewpoint to verify the conjecture for further classes of matroid complexes?

◮ Use PS-ear decomposability of matroid complexes? ◮ Characterize f - and h-vectors for further classes of simplicial

complexes, such as matroid polytopes?

◮ There is a list of all matroids on at most 10 elements – can

we find a counterexample here?

Yvonne Kemper PoEaRoMSCaG

slide-42
SLIDE 42

Flows on Simplicial Complexes

Yvonne Kemper PoEaRoMSCaG

slide-43
SLIDE 43

Flows on Graphs

Definition

A Zq-flow on an oriented graph G is a vector x ∈ ZE

q such that

  • h(e)=v

xe ≡

  • t(e)=v

xe mod q, for all v ∈ V . A Zq-flow is nowhere-zero if it is fully supported.

Yvonne Kemper PoEaRoMSCaG

slide-44
SLIDE 44

Flows on Graphs

Definition

A Zq-flow on an oriented graph G is a vector x ∈ ZE

q such that

  • h(e)=v

xe ≡

  • t(e)=v

xe mod q, for all v ∈ V . A Zq-flow is nowhere-zero if it is fully supported. Equivalently, a nowhere-zero Zq-flow is a fully-supported element

  • f the kernel mod q of the signed incidence matrix of the graph.

Yvonne Kemper PoEaRoMSCaG

slide-45
SLIDE 45

An Example

1 2 2 3 3 1 2 3 4 12 13 14 23 24 1 −1 −1 −1 2 1 −1 −1 3 1 1 4 1 1 (1, 2, 2, 3, 3) is a nowhere-zero Z5-flow and an element of the kernel (mod 5) of the incidence matrix.

Yvonne Kemper PoEaRoMSCaG

slide-46
SLIDE 46

Graph Flows: Origins and Open Questions

◮ Flows originally defined in the context of electric circuits and

networks

◮ Some Previous Work:

◮ The number of nowhere-zero Zq-flows on a graph is a

polynomial in q.

◮ For a planar graph G, χG(k) = kc(G)φG ∗(k). ◮ The Max-Flow/Min-Cut problem of optimization.

◮ Open Questions:

◮ 5-flow conjecture ◮ Volumes of flow polytopes Yvonne Kemper PoEaRoMSCaG

slide-47
SLIDE 47

Boundary/Incidence Matrices

Definition

Let ∂ be a boundary map on a (d − 1)-dimensional complex ∆ given by: ∂[vi0 · · · vir ] =

r

  • j=0

(−1)j[vi0 · · · vij · · · vir ], where 0 ≤ r ≤ d. The boundary matrix of ∆ is given by the signs

  • f the ridges in the boundary maps of the facets. We denote this

matrix ∂∆.

Yvonne Kemper PoEaRoMSCaG

slide-48
SLIDE 48

Boundary/Incidence Matrices

Definition

Let ∂ be a boundary map on a (d − 1)-dimensional complex ∆ given by: ∂[vi0 · · · vir ] =

r

  • j=0

(−1)j[vi0 · · · vij · · · vir ], where 0 ≤ r ≤ d. The boundary matrix of ∆ is given by the signs

  • f the ridges in the boundary maps of the facets. We denote this

matrix ∂∆. The boundary matrix of an oriented graph is equal to its signed incidence matrix.

Yvonne Kemper PoEaRoMSCaG

slide-49
SLIDE 49

Boundary/Incidence Matrices

Definition

Let ∂ be a boundary map on a (d − 1)-dimensional complex ∆ given by: ∂[vi0 · · · vir ] =

r

  • j=0

(−1)j[vi0 · · · vij · · · vir ], where 0 ≤ r ≤ d. The boundary matrix of ∆ is given by the signs

  • f the ridges in the boundary maps of the facets. We denote this

matrix ∂∆. The boundary matrix of an oriented graph is equal to its signed incidence matrix.

Definition

A Zq-flow on a pure simplicial complex ∆ is an element of the kernel mod q of the boundary matrix of ∆.

Yvonne Kemper PoEaRoMSCaG

slide-50
SLIDE 50

Flows on Simplicial Complexes

Example

The surface of a tetrahedron. 1 2 3 4 123 124 134 234 12 1 1 13 −1 1 14 −1 −1 23 1 1 24 1 −1 34 1 1 An example of a Z4-flow is: (1, 3, 1, 3).

Yvonne Kemper PoEaRoMSCaG

slide-51
SLIDE 51

Results

Proposition (Beck, K.)

Let ∆ be a triangulation of a manifold, and let φ∆(q) be the number of nowhere-zero Zq-flows on ∆. Then, φ∆(q) =            if ∆ has boundary; q − 1 if ∆ is without boundary, Z-orientable; if ∆ is without boundary, non-Z-orientable, q odd; 1 if ∆ is without boundary, non-Z-orientable, q even.

Yvonne Kemper PoEaRoMSCaG

slide-52
SLIDE 52

Results

Definition

A function φ in an integer variable t is a quasipolynomial if there exists an integer k > 0 and polynomials p0(t), . . . , pk−1(t) such that φ(t) = pj(t) if t ≡ j mod k. The minimal such k is the period of ϕ.

Yvonne Kemper PoEaRoMSCaG

slide-53
SLIDE 53

Results

Definition

A function φ in an integer variable t is a quasipolynomial if there exists an integer k > 0 and polynomials p0(t), . . . , pk−1(t) such that φ(t) = pj(t) if t ≡ j mod k. The minimal such k is the period of ϕ.

Example

Let φ(t) be defined for t ∈ Z as follows: φ(t) =        t2 + 1 if t ≡ 0 mod 5 t − 4 if t ≡ 1, 3 mod 5 3t3 + 1

2t

if t ≡ 2 mod 5 if t ≡ 4 mod 5. Then φ(t) is a quasipolynomial with period 5.

Yvonne Kemper PoEaRoMSCaG

slide-54
SLIDE 54

Results

Theorem (Beck, K.)

The number φ∆(q) of nowhere-zero Zq-flows on ∆ is a quasipolynomial in q. Furthermore, there exists a polynomial p(x) such that φ∆(k) = p(k) for all integers k that are relatively prime to the period of φ∆(q).

Theorem (Beck, K.)

◮ Let q be a sufficiently large prime number, and let ∆ be a

simplicial complex of dimension d. Then the number φ∆(q)

  • f nowhere-zero Zq-flows on ∆ is a polynomial in q of degree

dimQ( Hd(∆; Q)).

◮ In particular, φ∆(q) = (−1)|E(M)|−rk(M)TM(0, 1 − q), where

M is the matroid given by the columns of ∂∆.

Yvonne Kemper PoEaRoMSCaG

slide-55
SLIDE 55

Definitely Quasipolynomials: The Klein Bottle

Example

Let K be the Klein bottle. Then: H2(K; Zq) =

  • Z2

q even q odd. Therefore: φK(q) =

  • 1

q even q odd; φK(q) is a quasipolynomial with period 2.

Yvonne Kemper PoEaRoMSCaG

slide-56
SLIDE 56

The Period of the Flow Quasipolynomial

Definition

A matrix is totally unimodular (TU) iff every subdeterminant is 0, 1, or −1.

Fact

If the boundary matrix of a simplicial complex ∆ is TU, then φ∆(q) has period 1.

Yvonne Kemper PoEaRoMSCaG

slide-57
SLIDE 57

The Period of the Flow Quasipolynomial

Definition

A matrix is totally unimodular (TU) iff every subdeterminant is 0, 1, or −1.

Fact

If the boundary matrix of a simplicial complex ∆ is TU, then φ∆(q) has period 1.

Theorem (Dey, Hirani, Krishnamoorthy)

For a finite simplicial complex ∆ of dimension greater than d − 1, the boundary matrix [∂d] is totally unimodular if and only if Hd−1(L, L0) is torsion-free for all pure subcomplexes L0, L in ∆ of dimensions d − 1 and d respectively, where L0 ⊂ L.

Yvonne Kemper PoEaRoMSCaG

slide-58
SLIDE 58

Convex Ear Decomposable Simplicial Complexes

Definition

A convex ear decomposition of a pure rank-d simplicial complex ∆ is an ordered sequence Σ1, Σ2, . . . , Σn (the ears) of pure rank-d subcomplexes of ∆ such that (1) Σ1 is the boundary complex of a simplicial d-polytope, while for each i = 2, . . . , n, Σi is a (d − 1)-ball which is a (proper) sub-complex of the boundary complex of a simplicial d-polytope, and (2) For i ≥ 2, Σi ∩ i−1

j=1 Σj

  • = ∂Σi.

1 2 3 4 1 2 3 Σ1 1 2 3 4 Σ2

Yvonne Kemper PoEaRoMSCaG

slide-59
SLIDE 59

CED But Not TU

1 2 3 4 5 6

Yvonne Kemper PoEaRoMSCaG

slide-60
SLIDE 60

CED But Not TU

1 2 3 4 5 6 φ∆(q) = (q − 1)(q − 2) BUT the period is still equal to 1 – perhaps there is hope...?

Yvonne Kemper PoEaRoMSCaG

slide-61
SLIDE 61

CED, Not TU, and p > 1

1 2 3 4 5 6 1 2 4 5 7 8 2 6 5 3 4 7 1 φ∆(q) = q3 − 7q2 + 15q − 8 − gcd(2, q)

Yvonne Kemper PoEaRoMSCaG

slide-62
SLIDE 62

Open Questions and Future Directions

◮ Topological Conditions

◮ Necessary and/or sufficient topological conditions for period

equal to 1?

◮ Necessary and/or sufficient topological conditions for period

greater than 1?

◮ The Period of the Flow Quasipolynomial

◮ Is there a bound for quasipolynomials from modular flows? ◮ Can we find a subcomplex that guarantees a period greater

than 1 – or is there always the possibility of period collapse?

◮ Constructions preserving/leading to polynomiality ◮ Families of simplicial complexes with p = 1? ◮ Relationship between φG(q) of a graph G and φ∆(G)(q)?

Yvonne Kemper PoEaRoMSCaG

slide-63
SLIDE 63

Polytopal Embeddings of Cayley Graphs

Yvonne Kemper PoEaRoMSCaG

slide-64
SLIDE 64

Cayley Graphs

Definition

Let Γ be a group, and ∆ a set of generators of Γ. The Cayley color graph, C(Γ, ∆), of (Γ, ∆) is a directed, edge-colored graph such that:

◮ its vertices are the elements of Γ, and ◮ there is directed edge colored h from g1 to g2 if there exists a

generator h ∈ ∆ such that g1h = g2. If we forget the colors and directions of the edges of C(Γ, ∆), we have the Cayley graph, G(Γ, ∆).

Remark

A group will have many Cayley graphs, which depend on the representation that is used.

Yvonne Kemper PoEaRoMSCaG

slide-65
SLIDE 65

An Example

Say we have a representation of a group Γ = x, y | xy = yx, x3 = y2 = 1. The Cayley color graph is: 1 x x2 y yx yx2

Yvonne Kemper PoEaRoMSCaG

slide-66
SLIDE 66

An Example

Say we have a representation of a graph Γ = x, y | xy = yx, x3 = y2 = 1. The Cayley color graph is: 1 x x2 y yx yx2

Yvonne Kemper PoEaRoMSCaG

slide-67
SLIDE 67

An Example

Say we have a representation of a graph Γ = x, y | xy = yx, x3 = y2 = 1. The Cayley graph is: 1 x x2 y yx yx2

Yvonne Kemper PoEaRoMSCaG

slide-68
SLIDE 68

The Genus of a Cayley Graph

Definition

The genus of a graph is the minimal genus of all orientable surfaces in which G can be embedded.

Definition

The genus of a group Γ, γ(Γ), is the minimal genus among the genera of all possible Cayley graphs of Γ.

Yvonne Kemper PoEaRoMSCaG

slide-69
SLIDE 69

The Genus of a Cayley Graph

Definition

The genus of a graph is the minimal genus of all orientable surfaces in which G can be embedded.

Definition

The genus of a group Γ, γ(Γ), is the minimal genus among the genera of all possible Cayley graphs of Γ.

Open Problem!

Classify all finite groups of a particular genus γ, for all γ > 2.

Yvonne Kemper PoEaRoMSCaG

slide-70
SLIDE 70

Finite Groups of Genus 0, 1, and 2

◮ Genus 0: Classified by Maschke (1896) ◮ Genus 1: Classified by Proulx in her thesis (1978) ◮ Genus 2: Just one of them, found by Tucker (1984)

x, y, z | x2 = y2 = z2 = 1, (xy)2 = (yz)3 = (xz)8 = 1, y(xz)4y(xz)4 = 1

Yvonne Kemper PoEaRoMSCaG

slide-71
SLIDE 71

Finite Groups of Genus 0, 1, and 2

◮ Genus 0: Classified by Maschke (1896) ◮ Genus 1: Classified by Proulx in her thesis (1978) ◮ Genus 2: Just one of them, found by Tucker (1984)

x, y, z | x2 = y2 = z2 = 1, (xy)2 = (yz)3 = (xz)8 = 1, y(xz)4y(xz)4 = 1

Yvonne Kemper PoEaRoMSCaG

slide-72
SLIDE 72

Finite Groups of Genus 0

Question

When is there a polyhedral embedding of a planar group?

Yvonne Kemper PoEaRoMSCaG

slide-73
SLIDE 73

Hold Up: Connectivity? and Other Definitions

Definition

◮ A separator S of a graph G is a subset of the vertices V such

that V \ S has at least two components.

◮ A k-separator is a separator of cardinality k. ◮ A graph is k-connected if there exist no separators of

cardinality ≤ k − 1.

Yvonne Kemper PoEaRoMSCaG

slide-74
SLIDE 74

Finite Groups of Genus 0

Question

When is there a polyhedral embedding of a planar group?

Yvonne Kemper PoEaRoMSCaG

slide-75
SLIDE 75

Finite Groups of Genus 0

Question

When is there a polyhedral embedding of a planar group?

◮ Fact 1: (Steinitz, 1922) A graph is the 1-skeleton of a

polyhedron iff it is 3-connected and planar.

Yvonne Kemper PoEaRoMSCaG

slide-76
SLIDE 76

Finite Groups of Genus 0

Question

When is there a polyhedral embedding of a planar group?

◮ Fact 1: (Steinitz, 1922) A graph is the 1-skeleton of a

polyhedron iff it is 3-connected and planar.

◮ Fact 1+: (Mani, 1971) Every 3-connected, planar graph G is

the 1-skeleton of a polyhedron P such that every automorphism of G is induced by a symmetry of P.

Yvonne Kemper PoEaRoMSCaG

slide-77
SLIDE 77

Finite Groups of Genus 0

Question

When is there a polyhedral embedding of a planar group?

◮ Fact 1: (Steinitz, 1922) A graph is the 1-skeleton of a

polyhedron iff it is 3-connected and planar.

◮ Fact 1+: (Mani, 1971) Every 3-connected, planar graph G is

the 1-skeleton of a polyhedron P such that every automorphism of G is induced by a symmetry of P.

Proposition (De Loera)

Let G(Γ, ∆) be a planar Cayley graph for the group Γ. Then G(Γ, ∆) can be embedded as the 1-skeleton of a polytope.

Yvonne Kemper PoEaRoMSCaG

slide-78
SLIDE 78

An Example

Say we have a representation of a graph Γ = x, y | xy = yx, x3 = y2 = 1. The polytonal embedding of the Cayley color graph is...? 1 x x2 y yx yx2 It is 3-connected, and it is clearly planar...

Yvonne Kemper PoEaRoMSCaG

slide-79
SLIDE 79

An Example

Say we have a representation of a graph Γ = x, y | xy = yx, x3 = y2 = 1. The polytonal embedding of the Cayley color graph is: 1 x x2 y yx yx2

Yvonne Kemper PoEaRoMSCaG

slide-80
SLIDE 80

A Natural Question

Question

For any group Γ, and any representation ∆, can we always find a convex polytope such that G(Γ, ∆) is its 1-skeleton?

Yvonne Kemper PoEaRoMSCaG

slide-81
SLIDE 81

A Natural Question

Question

For any group Γ, and any representation ∆, can we always find a convex polytope such that G(Γ, ∆) is its 1-skeleton? Let’s find out...

Yvonne Kemper PoEaRoMSCaG

slide-82
SLIDE 82

The Quaternions: Q8

One presentation of Q8 is: ∆ = i, j | i4 = 1, i2 = j2, j−1ij = i−1.

Yvonne Kemper PoEaRoMSCaG

slide-83
SLIDE 83

The Quaternions: Q8

One presentation of Q8 is: ∆ = i, j | i4 = 1, i2 = j2, j−1ij = i−1. This has the corresponding Cayley color graph:

ij j −ij −j 1 −i −1 i multiplication by i multiplication by j

Yvonne Kemper PoEaRoMSCaG

slide-84
SLIDE 84

Q8 has Genus 1

We can embed this Cayley color graph on a Torus:

ij j −ij −j 1 −i −1 i

Yvonne Kemper PoEaRoMSCaG

slide-85
SLIDE 85

A Natural Question: Let’s be specific, here.

Question

Can Q8 be embedded as the 1-skeleton of some convex polytope?

Yvonne Kemper PoEaRoMSCaG

slide-86
SLIDE 86

A Natural Question: Let’s be specific, here.

Question

Can Q8 be embedded as the 1-skeleton of some convex polytope?

Theorem

Nope!

Yvonne Kemper PoEaRoMSCaG

slide-87
SLIDE 87

A Natural Question: Let’s be specific, here.

Question

Can Q8 be embedded as the 1-skeleton of some convex polytope?

Theorem

Nope!

A More Honest Theorem

There exists no convex polytope P with G(P) equal to the Cayley graph of a minimal presentation of the quaternion group.

Yvonne Kemper PoEaRoMSCaG

slide-88
SLIDE 88

What’s next?

◮ Are there (infinite) families of groups the minimal

presentations of which cannot be embedded as the graphs of convex d-polytopes?

◮ Can we use group theory to characterize the embeddability of

Cayley graphs?

◮ Characterize subgroups that “block” the embedding of the

Cayley graphs

◮ Show that there exist no such subgroups

◮ Are there forbidden minor characterizations for the

embeddability of Cayley graphs?

◮ Can we develop constructions that give d-polytopes with

graphs equal to Cayley graphs?

Yvonne Kemper PoEaRoMSCaG

slide-89
SLIDE 89

Thank you!

Yvonne Kemper PoEaRoMSCaG