Computing the Cohomology Algebra of a Polyhedral Complex Joint work - - PowerPoint PPT Presentation

computing the cohomology algebra of a polyhedral complex
SMART_READER_LITE
LIVE PREVIEW

Computing the Cohomology Algebra of a Polyhedral Complex Joint work - - PowerPoint PPT Presentation

Computing the Cohomology Algebra of a Polyhedral Complex Joint work with R. Gonzalez-Diaz & J. Lamar Ron Umble Millersville Univiversity Escuela de Ingeniera Informatica 27 March 2018 Polyhedral Complexes A polyhedral complex X is a


slide-1
SLIDE 1

Computing the Cohomology Algebra of a Polyhedral Complex

Joint work with R. Gonzalez-Diaz & J. Lamar Ron Umble Millersville Univiversity

Escuela de Ingeniería Informatica

27 March 2018

slide-2
SLIDE 2

Polyhedral Complexes

A polyhedral complex X is a regular cell complex whose

k-cells are k-dim’l polytopes

slide-3
SLIDE 3

Polyhedral Complexes

A polyhedral complex X is a regular cell complex whose

k-cells are k-dim’l polytopes

X is simplicial if its k-cells are k-simplices

slide-4
SLIDE 4

Polyhedral Complexes

A polyhedral complex X is a regular cell complex whose

k-cells are k-dim’l polytopes

X is simplicial if its k-cells are k-simplices X is cubical if its k-cells are k-cubes

slide-5
SLIDE 5

Diagonal Approximations

The cellular chains of X, denoted C∗ (X) , is the graded

Z2-vector space generated by the k-cells of X

slide-6
SLIDE 6

Diagonal Approximations

The cellular chains of X, denoted C∗ (X) , is the graded

Z2-vector space generated by the k-cells of X

Example: If P is a polygon, C∗ (P) is generated by the

vertices, edges, and region of P

slide-7
SLIDE 7

Diagonal Approximations

The cellular chains of X, denoted C∗ (X) , is the graded

Z2-vector space generated by the k-cells of X

Example: If P is a polygon, C∗ (P) is generated by the

vertices, edges, and region of P

The geometric boundary induces a differential operator

∂ : C∗ (X) → C∗−1 (X) such that ∂ ◦ ∂ = 0

slide-8
SLIDE 8

Diagonal Approximations

The cellular chains of X, denoted C∗ (X) , is the graded

Z2-vector space generated by the k-cells of X

Example: If P is a polygon, C∗ (P) is generated by the

vertices, edges, and region of P

The geometric boundary induces a differential operator

∂ : C∗ (X) → C∗−1 (X) such that ∂ ◦ ∂ = 0

A diagonal approximation on X is a map ∆X : X → X × X

that

slide-9
SLIDE 9

Diagonal Approximations

The cellular chains of X, denoted C∗ (X) , is the graded

Z2-vector space generated by the k-cells of X

Example: If P is a polygon, C∗ (P) is generated by the

vertices, edges, and region of P

The geometric boundary induces a differential operator

∂ : C∗ (X) → C∗−1 (X) such that ∂ ◦ ∂ = 0

A diagonal approximation on X is a map ∆X : X → X × X

that

Is homotopic to the geometric diagonal ∆ : x → (x, x)

slide-10
SLIDE 10

Diagonal Approximations

The cellular chains of X, denoted C∗ (X) , is the graded

Z2-vector space generated by the k-cells of X

Example: If P is a polygon, C∗ (P) is generated by the

vertices, edges, and region of P

The geometric boundary induces a differential operator

∂ : C∗ (X) → C∗−1 (X) such that ∂ ◦ ∂ = 0

A diagonal approximation on X is a map ∆X : X → X × X

that

Is homotopic to the geometric diagonal ∆ : x → (x, x) Commutes with the boundary operator

∆X ∂ = (∂ ⊗ Id + Id ⊗ ∂) ∆X

slide-11
SLIDE 11

Goals of the Talk

  • 1. Transform a simplicial or cubical complex X into a polyhedral

complex P

slide-12
SLIDE 12

Goals of the Talk

  • 1. Transform a simplicial or cubical complex X into a polyhedral

complex P

  • 2. Given a diagonal on C∗ (X) , induce a diagonal on C∗ (P)
slide-13
SLIDE 13

Alexander-Whitney Diagonal on the Simplex

∆s (012 · · · n) =

n

i=0

012 · · · i ⊗ i · · · n ∆s (012) = 0 ⊗ 012 + 01 ⊗ 12 + 012 ⊗ 2

slide-14
SLIDE 14

Serre Diagonal on the Cube

∆I (In) =

(u1,...,un)∈{0,I}×n

u1 · · · un ⊗ u

1 · · · u n

(0 = I and I = 1) ∆I

  • I2 = 00 ⊗ II + 0I ⊗ I1 + I0 ⊗ 1I + II ⊗ 11
slide-15
SLIDE 15

S-U Diagonal on the Associahedron

slide-16
SLIDE 16

A Diagonal on an n-gon P

Vertices labeled v1, v2, . . . , vn

slide-17
SLIDE 17

A Diagonal on an n-gon P

Vertices labeled v1, v2, . . . , vn Edges with endpoints vi and vi+1 labeled ei for i < n

slide-18
SLIDE 18

A Diagonal on an n-gon P

Vertices labeled v1, v2, . . . , vn Edges with endpoints vi and vi+1 labeled ei for i < n Edge with endpoints vn and v1 labeled en

slide-19
SLIDE 19

A Diagonal on an n-gon P

Vertices labeled v1, v2, . . . , vn Edges with endpoints vi and vi+1 labeled ei for i < n Edge with endpoints vn and v1 labeled en v1 is the initial vertex; vn is the terminal vertex

slide-20
SLIDE 20

A Diagonal on an n-gon P

Vertices labeled v1, v2, . . . , vn Edges with endpoints vi and vi+1 labeled ei for i < n Edge with endpoints vn and v1 labeled en v1 is the initial vertex; vn is the terminal vertex Edges are directed from v1 to vn

slide-21
SLIDE 21

A Diagonal on an n-gon P

Theorem (D. Kravatz, 2008 thesis) There is a diagonal

approximation on C∗ (P) defined by ∆P(vi) = vi ⊗ vi ∆P(ei) = vi ⊗ ei + ei ⊗ vi+1 if i < n ∆P(en) = v1 ⊗ en + en ⊗ vn ∆P(P) = v1 ⊗ P + P ⊗ vn +

0<i1<i2<n

ei1 ⊗ ei2

slide-22
SLIDE 22

A General Diagonal on an n-gon P

Let vt be the terminal vertex

slide-23
SLIDE 23

A General Diagonal on an n-gon P

Let vt be the terminal vertex Introduce a new edge e 0 from v1 to vt

slide-24
SLIDE 24

A General Diagonal on an n-gon P

Let vt be the terminal vertex Introduce a new edge e 0 from v1 to vt Let P1 be the subpolygon with vertices v1, v2, . . . , vt

slide-25
SLIDE 25

A General Diagonal on an n-gon P

Let vt be the terminal vertex Introduce a new edge e 0 from v1 to vt Let P1 be the subpolygon with vertices v1, v2, . . . , vt Let P2 be the subpolygon with vertices v1, vt, vt+1, . . . , vn

slide-26
SLIDE 26

A General Diagonal on an n-gon P

Let vt be the terminal vertex Introduce a new edge e 0 from v1 to vt Let P1 be the subpolygon with vertices v1, v2, . . . , vt Let P2 be the subpolygon with vertices v1, vt, vt+1, . . . , vn Edges are directed from v1 to vt

slide-27
SLIDE 27

A General Diagonal on an n-gon P

Corollary Let P be an n-gon with initial vertex v1 and terminal vertex vt. Then ∆

P(P) = v1 ⊗ P + P ⊗ vt +

0<i1<i2<t

ei1 ⊗ ei2 +

n≥i1>i2≥t

ei1 ⊗ ei2 is a diagonal approximation on C∗ (P)

slide-28
SLIDE 28

Application to Closed Compact Surfaces

Let Xg be a closed compact surface of genus g. The celebrated Classification of Closed Compact Surfaces states that Xg is homeomorphic to a

Sphere with g ≥ 0 handles when orientable

slide-29
SLIDE 29

Application to Closed Compact Surfaces

Let Xg be a closed compact surface of genus g. The celebrated Classification of Closed Compact Surfaces states that Xg is homeomorphic to a

Sphere with g ≥ 0 handles when orientable Connected sum of g ≥ 1 real projective planes when

unorientable

slide-30
SLIDE 30

Application to Closed Compact Surfaces

Let Xg be a closed compact surface of genus g. The celebrated Classification of Closed Compact Surfaces states that Xg is homeomorphic to a

Sphere with g ≥ 0 handles when orientable Connected sum of g ≥ 1 real projective planes when

unorientable

When g ≥ 1, Xg is the quotient of a

slide-31
SLIDE 31

Application to Closed Compact Surfaces

Let Xg be a closed compact surface of genus g. The celebrated Classification of Closed Compact Surfaces states that Xg is homeomorphic to a

Sphere with g ≥ 0 handles when orientable Connected sum of g ≥ 1 real projective planes when

unorientable

When g ≥ 1, Xg is the quotient of a

4g-gon when orientable

slide-32
SLIDE 32

Application to Closed Compact Surfaces

Let Xg be a closed compact surface of genus g. The celebrated Classification of Closed Compact Surfaces states that Xg is homeomorphic to a

Sphere with g ≥ 0 handles when orientable Connected sum of g ≥ 1 real projective planes when

unorientable

When g ≥ 1, Xg is the quotient of a

4g-gon when orientable 2g-gon when unorientable

slide-33
SLIDE 33

Polygonal Decomposition of a Torus

= ⇒

slide-34
SLIDE 34

Polygonal Decomposition of Real Projective Plane

= ⇒

slide-35
SLIDE 35

Polygonal Decomposition of a Klein Bottle

= ⇒

slide-36
SLIDE 36

Connected Sums

To obtain the connected sum X#Y of two surfaces, remove the interior of a disk from X and from Y then glue the two surfaces together along their boundaries Connected sums of four real projective planes and three tori

slide-37
SLIDE 37

The g-fold Torus as a Quotient of a 4g-gon

An g-fold torus as a quotient of a 4g-gon has

slide-38
SLIDE 38

The g-fold Torus as a Quotient of a 4g-gon

An g-fold torus as a quotient of a 4g-gon has

  • ne 0-cell v
slide-39
SLIDE 39

The g-fold Torus as a Quotient of a 4g-gon

An g-fold torus as a quotient of a 4g-gon has

  • ne 0-cell v

2g 1-cells (e1, e2) , . . . ,

  • e2g −1, e2g
slide-40
SLIDE 40

The g-fold Torus as a Quotient of a 4g-gon

An g-fold torus as a quotient of a 4g-gon has

  • ne 0-cell v

2g 1-cells (e1, e2) , . . . ,

  • e2g −1, e2g
  • ne 2-cell Tg
slide-41
SLIDE 41

A Diagonal on the g-fold Torus

A diagonal on Tg is defined by

∆Tg (Tg ) = v ⊗ Tg + Tg ⊗ v +

g

i=1

e2i−1 ⊗ e2i + e2i ⊗ e2i−1

slide-42
SLIDE 42

A Diagonal on the g-fold Projective Plane

A diagonal on RPg is defined by

∆RPg (RPg ) = v ⊗ RPg + RPg ⊗ v +

g

i=1

ei ⊗ ei

slide-43
SLIDE 43

A Diagonal on the g-fold Projective Plane

A diagonal on RPg is defined by

∆RPg (RPg ) = v ⊗ RPg + RPg ⊗ v +

g

i=1

ei ⊗ ei

∆Tg and ∆RPg are strikingly different and determine the

homeomorphism type of the surface

slide-44
SLIDE 44

Cohomology of a Closed Compact Surface

Choose a polygonal decomposition of Xg

slide-45
SLIDE 45

Cohomology of a Closed Compact Surface

Choose a polygonal decomposition of Xg ∂ = 0 on C∗ (Xg ) implies Hk (Xg ) = Ck (Xg )

slide-46
SLIDE 46

Cohomology of a Closed Compact Surface

Choose a polygonal decomposition of Xg ∂ = 0 on C∗ (Xg ) implies Hk (Xg ) = Ck (Xg ) Cohomology is the linear dual of homology

Hk (Xg ) = Hom (Hk (Xg ) , Z2)

slide-47
SLIDE 47

Cohomology of a Closed Compact Surface

Choose a polygonal decomposition of Xg ∂ = 0 on C∗ (Xg ) implies Hk (Xg ) = Ck (Xg ) Cohomology is the linear dual of homology

Hk (Xg ) = Hom (Hk (Xg ) , Z2)

If x ∈ Hk (Xg ) , define x∗ (e) =

1, if e = x 0,

  • therwise
slide-48
SLIDE 48

Cohomology of a Closed Compact Surface

Choose a polygonal decomposition of Xg ∂ = 0 on C∗ (Xg ) implies Hk (Xg ) = Ck (Xg ) Cohomology is the linear dual of homology

Hk (Xg ) = Hom (Hk (Xg ) , Z2)

If x ∈ Hk (Xg ) , define x∗ (e) =

1, if e = x 0,

  • therwise

Hk (Xg ) = {x∗ : x ∈ Hk (Xg )}

slide-49
SLIDE 49

Cohomology Algebra of a Closed Compact Surface

Given x∗, y ∗ ∈ H∗ (X) , define x∗ y ∗ = m (x∗ ⊗ y ∗) ∆X

slide-50
SLIDE 50

Cohomology Algebra of a Closed Compact Surface

Given x∗, y ∗ ∈ H∗ (X) , define x∗ y ∗ = m (x∗ ⊗ y ∗) ∆X The terms of ∆X (x) determine the factors of x∗

slide-51
SLIDE 51

Cohomology Algebra of a Closed Compact Surface

Given x∗, y ∗ ∈ H∗ (X) , define x∗ y ∗ = m (x∗ ⊗ y ∗) ∆X The terms of ∆X (x) determine the factors of x∗ Example

∆Tg (Tg ) = v ⊗ Tg + Tg ⊗ v +

g

i=1

e2i−1 ⊗ e2i + e2i ⊗ e2i−1 (e∗

2i−1 e∗ 2i) (Tg )

= m (e∗

2i−1 ⊗ e∗ 2i) ∆Tg (Tg )

= m (e∗

2i−1 ⊗ e∗ 2i) (e2i−1 ⊗ e2i)

= m (1 ⊗ 1) = 1

slide-52
SLIDE 52

Cohomology Algebra of a Closed Compact Surface

Given x∗, y ∗ ∈ H∗ (X) , define x∗ y ∗ = m (x∗ ⊗ y ∗) ∆X The terms of ∆X (x) determine the factors of x∗ Example

∆Tg (Tg ) = v ⊗ Tg + Tg ⊗ v +

g

i=1

e2i−1 ⊗ e2i + e2i ⊗ e2i−1 (e∗

2i−1 e∗ 2i) (Tg )

= m (e∗

2i−1 ⊗ e∗ 2i) ∆Tg (Tg )

= m (e∗

2i−1 ⊗ e∗ 2i) (e2i−1 ⊗ e2i)

= m (1 ⊗ 1) = 1

e∗ 2i−1 e∗ 2i acting non-trivially on Tg implies

e∗

2i−1 e∗ 2i = T ∗ g

slide-53
SLIDE 53

Cohomology Algebra of a g-Fold Torus

∆Tg (Tg ) = v ⊗ Tg + Tg ⊗ v + g

i=1

e2i−1 ⊗ e2i + e2i ⊗ e2i−1 ⇒ v ∗ T ∗

g = T ∗ g v ∗ = e∗ 1 e∗ 2 = e∗ 2 e∗ 1 =

· · · = e∗

2g−1 e∗ 2g = e∗ 2g e∗ 2g−1 = T ∗ g

slide-54
SLIDE 54

Cohomology Algebra of a g-Fold Torus

∆Tg (Tg ) = v ⊗ Tg + Tg ⊗ v + g

i=1

e2i−1 ⊗ e2i + e2i ⊗ e2i−1 ⇒ v ∗ T ∗

g = T ∗ g v ∗ = e∗ 1 e∗ 2 = e∗ 2 e∗ 1 =

· · · = e∗

2g−1 e∗ 2g = e∗ 2g e∗ 2g−1 = T ∗ g ∆Tg (e2i) = v ⊗ e2i + e2i ⊗ v ⇒

v ∗ e∗

2i = e∗ 2i v ∗ = e∗ 2i

slide-55
SLIDE 55

Cohomology Algebra of a g-Fold Torus

∆Tg (Tg ) = v ⊗ Tg + Tg ⊗ v + g

i=1

e2i−1 ⊗ e2i + e2i ⊗ e2i−1 ⇒ v ∗ T ∗

g = T ∗ g v ∗ = e∗ 1 e∗ 2 = e∗ 2 e∗ 1 =

· · · = e∗

2g−1 e∗ 2g = e∗ 2g e∗ 2g−1 = T ∗ g ∆Tg (e2i) = v ⊗ e2i + e2i ⊗ v ⇒

v ∗ e∗

2i = e∗ 2i v ∗ = e∗ 2i Factors in a non-vanishing cup product of 1-dim’l classes are

dual to 1-cells in the same component of the connected sum

slide-56
SLIDE 56

Cohomology Algebra of a g-Fold Torus

∆Tg (Tg ) = v ⊗ Tg + Tg ⊗ v + g

i=1

e2i−1 ⊗ e2i + e2i ⊗ e2i−1 ⇒ v ∗ T ∗

g = T ∗ g v ∗ = e∗ 1 e∗ 2 = e∗ 2 e∗ 1 =

· · · = e∗

2g−1 e∗ 2g = e∗ 2g e∗ 2g−1 = T ∗ g ∆Tg (e2i) = v ⊗ e2i + e2i ⊗ v ⇒

v ∗ e∗

2i = e∗ 2i v ∗ = e∗ 2i Factors in a non-vanishing cup product of 1-dim’l classes are

dual to 1-cells in the same component of the connected sum

All 1-dim’l cup squares vanish

slide-57
SLIDE 57

Cohomology Algebra of a g-Fold Torus

∆Tg (Tg ) = v ⊗ Tg + Tg ⊗ v + g

i=1

e2i−1 ⊗ e2i + e2i ⊗ e2i−1 ⇒ v ∗ T ∗

g = T ∗ g v ∗ = e∗ 1 e∗ 2 = e∗ 2 e∗ 1 =

· · · = e∗

2g−1 e∗ 2g = e∗ 2g e∗ 2g−1 = T ∗ g ∆Tg (e2i) = v ⊗ e2i + e2i ⊗ v ⇒

v ∗ e∗

2i = e∗ 2i v ∗ = e∗ 2i Factors in a non-vanishing cup product of 1-dim’l classes are

dual to 1-cells in the same component of the connected sum

All 1-dim’l cup squares vanish H∗ (Tg ) is a graded commutative algebra with identity v ∗

slide-58
SLIDE 58

Cohomology Algebra of a g-Fold Projective Plane

∆RPg (RPg ) = v ⊗ RPg + RPg ⊗ v + g

i=1

ei ⊗ ei ⇒ v ∗ RP∗

g = RP∗ g v ∗ = e∗ 1 e∗ 1 =

· · · = e∗

g e∗ g = RP∗ g

slide-59
SLIDE 59

Cohomology Algebra of a g-Fold Projective Plane

∆RPg (RPg ) = v ⊗ RPg + RPg ⊗ v + g

i=1

ei ⊗ ei ⇒ v ∗ RP∗

g = RP∗ g v ∗ = e∗ 1 e∗ 1 =

· · · = e∗

g e∗ g = RP∗ g e∗ i e∗ j = 0 for all i = j

slide-60
SLIDE 60

Cohomology Algebra of a g-Fold Projective Plane

∆RPg (RPg ) = v ⊗ RPg + RPg ⊗ v + g

i=1

ei ⊗ ei ⇒ v ∗ RP∗

g = RP∗ g v ∗ = e∗ 1 e∗ 1 =

· · · = e∗

g e∗ g = RP∗ g e∗ i e∗ j = 0 for all i = j All non-vanishing cup products are squares

slide-61
SLIDE 61

Cohomology Algebra of a g-Fold Projective Plane

∆RPg (RPg ) = v ⊗ RPg + RPg ⊗ v + g

i=1

ei ⊗ ei ⇒ v ∗ RP∗

g = RP∗ g v ∗ = e∗ 1 e∗ 1 =

· · · = e∗

g e∗ g = RP∗ g e∗ i e∗ j = 0 for all i = j All non-vanishing cup products are squares H∗ (RPg ) is a graded commutative algebra with identity v ∗

slide-62
SLIDE 62

Important Lessons

The algebra structure of H∗ (Xg ) is a complete topological

invariant

slide-63
SLIDE 63

Important Lessons

The algebra structure of H∗ (Xg ) is a complete topological

invariant

Cup squares of 1-dim’l classes vanish iff Xg is orientabile

slide-64
SLIDE 64

Important Lessons

The algebra structure of H∗ (Xg ) is a complete topological

invariant

Cup squares of 1-dim’l classes vanish iff Xg is orientabile The number of 1-dim’l generators determines the genus g

slide-65
SLIDE 65

Important Lessons

The algebra structure of H∗ (Xg ) is a complete topological

invariant

Cup squares of 1-dim’l classes vanish iff Xg is orientabile The number of 1-dim’l generators determines the genus g 2g distinct 1-dim’l generators if orientable and g otherwise

slide-66
SLIDE 66

Important Lessons

The algebra structure of H∗ (Xg ) is a complete topological

invariant

Cup squares of 1-dim’l classes vanish iff Xg is orientabile The number of 1-dim’l generators determines the genus g 2g distinct 1-dim’l generators if orientable and g otherwise

A polygonal cell decomposition of Xg produces a diagonal ∆Xg

slide-67
SLIDE 67

Important Lessons

The algebra structure of H∗ (Xg ) is a complete topological

invariant

Cup squares of 1-dim’l classes vanish iff Xg is orientabile The number of 1-dim’l generators determines the genus g 2g distinct 1-dim’l generators if orientable and g otherwise

A polygonal cell decomposition of Xg produces a diagonal ∆Xg Algebra structure of H∗ (Xg ) follows immediately from ∆Xg

slide-68
SLIDE 68

Computing Cup Products on a Polyhedral Complex

STRATEGY:

Given a simplicial complex X with its A-W diagonal

slide-69
SLIDE 69

Computing Cup Products on a Polyhedral Complex

STRATEGY:

Given a simplicial complex X with its A-W diagonal Iteratively apply a chain contraction to

slide-70
SLIDE 70

Computing Cup Products on a Polyhedral Complex

STRATEGY:

Given a simplicial complex X with its A-W diagonal Iteratively apply a chain contraction to

merge adjacent cells and

slide-71
SLIDE 71

Computing Cup Products on a Polyhedral Complex

STRATEGY:

Given a simplicial complex X with its A-W diagonal Iteratively apply a chain contraction to

merge adjacent cells and induce a diagonal on the resulting polyhedral complex

slide-72
SLIDE 72

Computing Cup Products on a Polyhedral Complex

STRATEGY:

Given a simplicial complex X with its A-W diagonal Iteratively apply a chain contraction to

merge adjacent cells and induce a diagonal on the resulting polyhedral complex

Compute the cohomology algebra of the polyhedral complex

slide-73
SLIDE 73

Merging Adjacent Cells

Let (X, ∂) be a regular cell complex

slide-74
SLIDE 74

Merging Adjacent Cells

Let (X, ∂) be a regular cell complex Assume a k-cell e is the intersection of exactly two (k + 1)-

cells a and b

slide-75
SLIDE 75

Merging Adjacent Cells

Let (X, ∂) be a regular cell complex Assume a k-cell e is the intersection of exactly two (k + 1)-

cells a and b

Remove int (a ∪ b) and attach a (k + 1)-cell c along ∂ (a ∪ b)

slide-76
SLIDE 76

Merging Adjacent Cells

Let (X, ∂) be a regular cell complex Assume a k-cell e is the intersection of exactly two (k + 1)-

cells a and b

Remove int (a ∪ b) and attach a (k + 1)-cell c along ∂ (a ∪ b) Obtain the cell complex (X , ∂) with fewer cells

X X

slide-77
SLIDE 77

The Chain Contraction

X X There exist chain maps

f : C∗ (X) → C∗ (X )

slide-78
SLIDE 78

The Chain Contraction

X X There exist chain maps

f : C∗ (X) → C∗ (X ) g : C∗ (X ) → C∗ (X)

slide-79
SLIDE 79

The Chain Contraction

X X There exist chain maps

f : C∗ (X) → C∗ (X ) g : C∗ (X ) → C∗ (X) φ : C∗ (X) → C∗+1 (X) defined on generators by

f (e) = ∂a + e g (c) = a + b f (a) = 0 g (σ) = σ, σ = c f (b) = c φ (e) = a f (σ) = σ, σ = e, a, b φ (σ) = 0, σ = e

slide-80
SLIDE 80

The Chain Contraction

X X

fg = IdC∗(X ) and φ is a chain homotopy from gf to IdC∗(X )

∂φ + φ∂ = IdC∗(X ) + gf

slide-81
SLIDE 81

The Chain Contraction

X X

fg = IdC∗(X ) and φ is a chain homotopy from gf to IdC∗(X )

∂φ + φ∂ = IdC∗(X ) + gf

g is a chain homotopy equivalence

slide-82
SLIDE 82

The Chain Contraction

X X

fg = IdC∗(X ) and φ is a chain homotopy from gf to IdC∗(X )

∂φ + φ∂ = IdC∗(X ) + gf

g is a chain homotopy equivalence (f , g, φ) is called a chain contraction of C∗ (X) onto C∗ (X )

(Introduced by Henri Cartan 1904-2008)

slide-83
SLIDE 83

The Transfer Theorem

Theorem A chain contraction (f , g, φ, C∗ (X) , C∗ (X ))

preserves the algebraic topology of X

slide-84
SLIDE 84

The Transfer Theorem

Theorem A chain contraction (f , g, φ, C∗ (X) , C∗ (X ))

preserves the algebraic topology of X

Given a diagonal

∆X : C∗ (X) → C∗ (X) ⊗ C∗ (X) the composition ∆X = (f ⊗ f ) ◦ ∆X ◦ g

slide-85
SLIDE 85

The Transfer Theorem

Theorem A chain contraction (f , g, φ, C∗ (X) , C∗ (X ))

preserves the algebraic topology of X

Given a diagonal

∆X : C∗ (X) → C∗ (X) ⊗ C∗ (X) the composition ∆X = (f ⊗ f ) ◦ ∆X ◦ g

Is a diagonal ∆X : C∗ (X ) → C∗ (X ) ⊗ C∗ (X )

slide-86
SLIDE 86

The Transfer Theorem

Theorem A chain contraction (f , g, φ, C∗ (X) , C∗ (X ))

preserves the algebraic topology of X

Given a diagonal

∆X : C∗ (X) → C∗ (X) ⊗ C∗ (X) the composition ∆X = (f ⊗ f ) ◦ ∆X ◦ g

Is a diagonal ∆X : C∗ (X ) → C∗ (X ) ⊗ C∗ (X ) If ∆X is homotopy cocommutative, so is ∆X

slide-87
SLIDE 87

The Transfer Theorem

Theorem A chain contraction (f , g, φ, C∗ (X) , C∗ (X ))

preserves the algebraic topology of X

Given a diagonal

∆X : C∗ (X) → C∗ (X) ⊗ C∗ (X) the composition ∆X = (f ⊗ f ) ◦ ∆X ◦ g

Is a diagonal ∆X : C∗ (X ) → C∗ (X ) ⊗ C∗ (X ) If ∆X is homotopy cocommutative, so is ∆X If ∆X is homotopy coassociative, so is ∆X

slide-88
SLIDE 88

Example: Merging Adjacent 2-Simplices

X X f (e) = ∂a + e g (c) = a + b f (a) = 0 g (σ) = σ, σ = c f (b) = c φ (e) = a f (σ) = σ, σ = e, a, b φ (σ) = 0, σ = e

∆X (c) = [(f ⊗ f ) ◦ ∆X ◦ g] (c) = (f ⊗ f ) (∆X (a + b))

slide-89
SLIDE 89

Example: Merging Adjacent 2-Simplices

X X f (e) = ∂a + e g (c) = a + b f (a) = 0 g (σ) = σ, σ = c f (b) = c φ (e) = a f (σ) = σ, σ = e, a, b φ (σ) = 0, σ = e

∆X (c) = [(f ⊗ f ) ◦ ∆X ◦ g] (c) = (f ⊗ f ) (∆X (a + b))

  • = (f ⊗ f ) [1 ⊗ a + 14 ⊗ 43 + a ⊗ 3

+ 1 ⊗ b + 12 ⊗ 23 + b ⊗ 3]

slide-90
SLIDE 90

Example: Merging Adjacent 2-Simplices

X X f (e) = ∂a + e g (c) = a + b f (a) = 0 g (σ) = σ, σ = c f (b) = c φ (e) = a f (σ) = σ, σ = e, a, b φ (σ) = 0, σ = e

∆X (c) = [(f ⊗ f ) ◦ ∆X ◦ g] (c) = (f ⊗ f ) (∆X (a + b))

  • = (f ⊗ f ) [1 ⊗ a + 14 ⊗ 43 + a ⊗ 3

+ 1 ⊗ b + 12 ⊗ 23 + b ⊗ 3]

  • = 1 ⊗ c + 12 ⊗ 23 + 14 ⊗ 43 + c ⊗ 3
slide-91
SLIDE 91

The Cohomology Algebra of a Torus

∆T (γ) = v ⊗ γ + α ⊗ β + β ⊗ α + γ ⊗ v α∗ β∗ = β∗ α∗ = γ∗

slide-92
SLIDE 92

The Cohomology Algebra of T#T via Chain Contraction

f (e) = α1 + β1 + α1 + β1 = 0 ¯ ∆

T #T (T#T)

= [(f ⊗ f ) ◦ ¯ ∆X ◦ g] (T#T) = [(f ⊗ f ) ◦ ¯ ∆X ] (T1 + T2) = (f ⊗ f ) [α1 ⊗ β1 + e ⊗ (β1 + α1) + β1 ⊗ α1 +α2 ⊗ β2 + e ⊗ (β2 + α2) + β2 ⊗ α2] = α1 ⊗ β1 + β1 ⊗ α1 + α2 ⊗ β2 + β2 ⊗ α2 α∗

i β∗ i = β∗ i α∗ i = T#T ∗

slide-93
SLIDE 93

Generalization of Kravatz’s Diagonal on an n-gon

Theorem (G-L-U) Let X be a 3D polyhedral complex with vertices numbered arbitrarily from 1 to n. Represent a k-gon P of X as an ordered k-tuple of vertices i1, . . . , ik , where i1 = min {i1, . . . , ik} , i1 is adjacent to ik, and ij is adjacent to ij+1 for 1 < j < k. Then ∆P (P) = i1 ⊗ P + P ⊗ im(k) +

m(k)−1

j=2

(u2 + e2 + · · · + λjej) ⊗ ej +

k−1

j=m(k)

[(1 + λj) ej + ej+1 + · · · + ek−1 + uk] ⊗ ej, where im(k) = max {i2, . . . , ik} , λj = 0 iff ij < ij+1, {uj = i1, ij}2≤j≤k and {ej = ijij+1}2≤j≤k−1

slide-94
SLIDE 94

Computational Considerations

X X Number of 2-cells Cup product computed in X 1,638 28.00 sec X 46 1.04 sec

slide-95
SLIDE 95

Trabecular Bone

Makes up the inner layer of the bone and has a spongy,

honeycomb-like structure.

slide-96
SLIDE 96

Micro-CT Images of a Trabecular Bone

X X

slide-97
SLIDE 97

Representative 1-cycles

Non-vanishing cup products: α∗

2α∗ 4, α∗ 2α∗ 5, α∗ 2α∗ 9, α∗ 3α∗ 8, α∗ 4α∗ 5 Computational methods such as these allow us to identify

diseased tissue

slide-98
SLIDE 98

Thank you!