categorification---the sl2 case - functor valued invariants of knots, …stovicek/math/... ·...

Categorification—the sl2 caseFunctor valued invariants of knots, links and tangles

Jan Št’ovícekDepartment of Mathematical Sciences

October 22nd , 2009

www.ntnu.no Jan Št’ovícek, Categorification—the sl2 case

2

Outline

1. Knots and links

2. The tangle category and quantum groups

3. Categorification

4. Lie algebras and the category O


3

Outline

1. Knots and links


3. Categorification



4

Knots and links— A knot is a smooth, closed and oriented curve in R3.

Unknot Trefoil Knot 41

— A link is a finite disjoint union of knots.

Hopf link Solomon’s knot


5

Isotopy classes of knots and links— We usually work with 2-dimensional projections of knots

,where we remember for each crossing which part is above andwhich below. Such projections are called knot or link diagrams.

— How do we recognize knots which are actually “the same”(= isotopic)?

and

— Various approaches:• describing changes in the projection which do not change the

isotopy class of the knot—Reidemeister moves.• algebraic invariants—Jones polynomial, functors.


5

Isotopy classes of knots and links— We usually work with 2-dimensional projections of knots ,

where we remember for each crossing which part is above andwhich below.

Such projections are called knot or link diagrams.— How do we recognize knots which are actually “the same”

(= isotopic)?

and




5


where we remember for each crossing which part is above andwhich below. Such projections are called knot or link diagrams.


and




5




and

— Various approaches:

• describing changes in the projection which do not change theisotopy class of the knot—Reidemeister moves.

• algebraic invariants—Jones polynomial, functors.


5




and


isotopy class of the knot

—Reidemeister moves.• algebraic invariants—Jones polynomial, functors.


5




and


isotopy class of the knot—Reidemeister moves.

• algebraic invariants—Jones polynomial, functors.


5




and


isotopy class of the knot—Reidemeister moves.• algebraic invariants

—Jones polynomial, functors.


5




and




6

Reidemeister moves

Reidemeister I

Reidemeister II Reidemeister III

Theorem (Reidemeister 1926, Alexander and Briggs 1927)

Two knots are isotopic

if and only if their diagrams can betransferred to each other using planar isotopy

and theReidemeister moves I–III above.


6

Reidemeister moves

Reidemeister I Reidemeister II

Reidemeister III



if and only if their diagrams can betransferred to each other using planar isotopy and theReidemeister moves I–III above.


6

Reidemeister moves

Reidemeister I Reidemeister II Reidemeister III



if and only if their diagrams can betransferred to each other using planar isotopy and theReidemeister moves I–III above.


6

Reidemeister moves



Two knots are isotopic if and only if their diagrams can betransferred to each other using planar isotopy

and theReidemeister moves I–III above.


6

Reidemeister moves



Two knots are isotopic if and only if their diagrams can betransferred to each other using planar isotopy and theReidemeister moves I–III above.


7

Jones polynomial— Using only the Reidemeister moves, still not very easy to

decide whether two knots (or links) are isotopic.

— There is a simple algorithic way to assign to each linkdiagram L a Laurent polynomial PL ∈ Z[q,q−1] such thatPL = PL′ whenever L and L′ are diagrams of isotopic links.

— PL is called the Jones polynomial and it is fully determined bythe so called skein relation:

q−2 ·⟨ ⟩

+ q ·⟨ ⟩

= (q − q−1) ·⟨ ⟩

and its value on the unknot ( ).— Discovered by Vaughan Jones in 1983, based on work of

Alexander and Conway.— Not a complete knot invariant, but quite powerful in

distinguishing knots.


7


decide whether two knots (or links) are isotopic.— There is a simple algorithic way

to assign to each linkdiagram L a Laurent polynomial PL ∈ Z[q,q−1] such thatPL = PL′ whenever L and L′ are diagrams of isotopic links.


q−2 ·⟨ ⟩

+ q ·⟨ ⟩

= (q − q−1) ·⟨ ⟩





7


decide whether two knots (or links) are isotopic.— There is a simple algorithic way to assign to each link

diagram L

a Laurent polynomial PL ∈ Z[q,q−1] such thatPL = PL′ whenever L and L′ are diagrams of isotopic links.


q−2 ·⟨ ⟩

+ q ·⟨ ⟩

= (q − q−1) ·⟨ ⟩





7



diagram L a Laurent polynomial PL ∈ Z[q,q−1]

such thatPL = PL′ whenever L and L′ are diagrams of isotopic links.


q−2 ·⟨ ⟩

+ q ·⟨ ⟩

= (q − q−1) ·⟨ ⟩





7



diagram L a Laurent polynomial PL ∈ Z[q,q−1] such thatPL = PL′ whenever L and L′ are diagrams of isotopic links.


q−2 ·⟨ ⟩

+ q ·⟨ ⟩

= (q − q−1) ·⟨ ⟩





7




— PL is called the Jones polynomial

and it is fully determined bythe so called skein relation:

q−2 ·⟨ ⟩

+ q ·⟨ ⟩

= (q − q−1) ·⟨ ⟩





7





q−2 ·⟨ ⟩

+ q ·⟨ ⟩

= (q − q−1) ·⟨ ⟩





7





q−2 ·⟨ ⟩

+ q ·⟨ ⟩

= (q − q−1) ·⟨ ⟩

and its value on the unknot ( ).

— Discovered by Vaughan Jones in 1983, based on work ofAlexander and Conway.

— Not a complete knot invariant, but quite powerful indistinguishing knots.


7





q−2 ·⟨ ⟩

+ q ·⟨ ⟩

= (q − q−1) ·⟨ ⟩


Alexander and Conway.

— Not a complete knot invariant, but quite powerful indistinguishing knots.


7





q−2 ·⟨ ⟩

+ q ·⟨ ⟩

= (q − q−1) ·⟨ ⟩


Alexander and Conway.— Not a complete knot invariant,

but quite powerful indistinguishing knots.


7





q−2 ·⟨ ⟩

+ q ·⟨ ⟩

= (q − q−1) ·⟨ ⟩



distinguishing knots.www.ntnu.no Jan Št’ovícek, Categorification—the sl2 case

8

Outline

1. Knots and links


3. Categorification



9

Tangles

— A tangle is a finite collection of smooth oriented curves in R3.

1

— We will also assume that our tangles are inside a stripe ofheight 1 and all end points of non-closed curves are in thedelimiting planes.


9

Tangles


1

— We will also assume that our tangles are inside a stripe ofheight 1

and all end points of non-closed curves are in thedelimiting planes.


9

Tangles


1

— We will also assume that our tangles are inside a stripe ofheight 1 and all end points of non-closed curves are in thedelimiting planes.


10

Operations on tangles— Given two tangles with compatible ends, we can form a

composition tangle:

T1

,T2 T1 ◦ T2

— Given any pair of tangles, we can form a tensor product:

T ′1

,

T ′2 T ′1 ⊗ T ′2

— Important: These operations are compatible with isotopies!


11

Tensor category of tangles— We form a category of tangles, denoted by T , as follows:

• Objects: finite sequences of signs + and −, and the emptyset ∅.

• Morphisms: isotopy classes of tangles with orientationprescribed by the signs:

+ + + −

+ + + + − −∈ HomT

((+ + +−), (+ + + +−−)

)• Composition: As in the slide before—well defined.

— We also have a tensor product:(X ,Y ) 7→ X ⊗ Y (concatenating, eg. (+−)⊗ (+) = (+−+)),(f ,g) 7→ f ⊗ g (as in the slide before on morphisms).

— Associativity: (X ⊗ Y )⊗ Z ∼= X ⊗ (Y ⊗ Z ).— T is a tensor category, HomT (∅, ∅) = {isotopy classes of links}.


11




+ + + −

+ + + + − −∈ HomT

((+ + +−), (+ + + +−−)

)

• Composition: As in the slide before—well defined.




11




+ + + −

+ + + + − −∈ HomT

((+ + +−), (+ + + +−−)

)• Composition: As in the slide before

—well defined.




11




+ + + −

+ + + + − −∈ HomT

((+ + +−), (+ + + +−−)





11




+ + + −

+ + + + − −∈ HomT

((+ + +−), (+ + + +−−)


— We also have a tensor product:

(X ,Y ) 7→ X ⊗ Y (concatenating, eg. (+−)⊗ (+) = (+−+)),(f ,g) 7→ f ⊗ g (as in the slide before on morphisms).



11




+ + + −

+ + + + − −∈ HomT

((+ + +−), (+ + + +−−)


— We also have a tensor product:(X ,Y ) 7→ X ⊗ Y (concatenating, eg. (+−)⊗ (+) = (+−+)),

(f ,g) 7→ f ⊗ g (as in the slide before on morphisms).— Associativity: (X ⊗ Y )⊗ Z ∼= X ⊗ (Y ⊗ Z ).— T is a tensor category, HomT (∅, ∅) = {isotopy classes of links}.


11




+ + + −

+ + + + − −∈ HomT

((+ + +−), (+ + + +−−)





11




+ + + −

+ + + + − −∈ HomT

((+ + +−), (+ + + +−−)



— Associativity: (X ⊗ Y )⊗ Z ∼= X ⊗ (Y ⊗ Z ).

— T is a tensor category, HomT (∅, ∅) = {isotopy classes of links}.


11




+ + + −

+ + + + − −∈ HomT

((+ + +−), (+ + + +−−)



— Associativity: (X ⊗ Y )⊗ Z ∼= X ⊗ (Y ⊗ Z ).— T is a tensor category,

HomT (∅, ∅) = {isotopy classes of links}.


11




+ + + −

+ + + + − −∈ HomT

((+ + +−), (+ + + +−−)





12

Invariants of tangles

— Let R = C[q,q−1] and V = Rk be a free R-module, k ≥ 2.

— One can construct a functor F : T → modR of tensorcategories such that F (X ) = V ⊗R V ⊗R · · · ⊗R V︸︷︷︸

length(X)

(Reshetikhin and Turaev 1990).— If T is a tangle, then F ([T ]) : V⊗m → V⊗n is an isotopy

invariant.— In particular for a link L, we have F ([L]) : R → R. So F ([L])

acts as multiplication by a Laurent polynomial PL(q) ∈ R.— Indeed, we can reconstruct the Jones polynomial by making a

particular choice of F for k = 2!


12


— Let R = C[q,q−1] and V = Rk be a free R-module, k ≥ 2.— One can construct a functor F : T → modR of tensor

categories such that F (X ) = V ⊗R V ⊗R · · · ⊗R V︸︷︷︸length(X)






12




(Reshetikhin and Turaev 1990).

— If T is a tangle, then F ([T ]) : V⊗m → V⊗n is an isotopyinvariant.

— In particular for a link L, we have F ([L]) : R → R. So F ([L])acts as multiplication by a Laurent polynomial PL(q) ∈ R.

— Indeed, we can reconstruct the Jones polynomial by making aparticular choice of F for k = 2!


12





invariant.

— In particular for a link L, we have F ([L]) : R → R. So F ([L])acts as multiplication by a Laurent polynomial PL(q) ∈ R.



12





invariant.— In particular for a link L, we have F ([L]) : R → R.

So F ([L])acts as multiplication by a Laurent polynomial PL(q) ∈ R.



12






acts as multiplication by a Laurent polynomial PL(q) ∈ R.



12









13

Construction of the invariants— All morphisms in T are generated, using composition and ⊗,

by six elementary ones:

— As for links, two tangles are isotopic if and only if theirdiagrams can be transformed to each other using planarisotopy and the Reidemeister moves.

— This can be translated into a finite list of relations for thegenerators above, for example:

◦ ◦ = ◦ ◦— It is enough to define six R-linear maps V⊗2 → V⊗2, V⊗2 → R

and R → V⊗2 satisfying these relations and this uniquelydefines the functor F : T → modR.

— Based on work of Turaev and Yetter, around 1988.


13



— As for links, two tangles are isotopic

if and only if theirdiagrams can be transformed to each other using planarisotopy and the Reidemeister moves.






13



— As for links, two tangles are isotopic if and only if theirdiagrams can be transformed to each other using planarisotopy

and the Reidemeister moves.— This can be translated into a finite list of relations for the

generators above, for example:◦ ◦ = ◦ ◦

— It is enough to define six R-linear maps V⊗2 → V⊗2, V⊗2 → Rand R → V⊗2 satisfying these relations and this uniquelydefines the functor F : T → modR.



13









13




— This can be translated into a finite list of relations for thegenerators above,

for example:◦ ◦ = ◦ ◦




13





◦ ◦ = ◦ ◦




13






and R → V⊗2 satisfying these relations

and this uniquelydefines the functor F : T → modR.



13









14

YB-equation and quantum groups

— Put Φ = F ( ), recall Reidemeister III:

— This forces Φ to satisfy

(Φ⊗IdV )◦(IdV⊗Φ)◦(Φ⊗IdV ) = (IdV⊗Φ)◦(Φ⊗IdV )◦(IdV⊗Φ),

a so called quantum Yang-Baxter equation.— Solutions to this equation are called R-matrices.— In order to construct them, Drinfeld and Jimbo independently

introduced quantum groups (around 1985)—a class ofassociative non-commutative algebras.


14


— Put Φ = F ( ), recall Reidemeister III:— This forces Φ to satisfy





14




a so called quantum Yang-Baxter equation.

— Solutions to this equation are called R-matrices.— In order to construct them, Drinfeld and Jimbo independently



14




a so called quantum Yang-Baxter equation.— Solutions to this equation are called R-matrices.

— In order to construct them, Drinfeld and Jimbo independentlyintroduced quantum groups (around 1985)—a class ofassociative non-commutative algebras.


14





introduced quantum groups (around 1985)—

a class ofassociative non-commutative algebras.


14







15

The sl2-case— One representative of quantum groups is Uq(sl2)

defined asthe associative non-commutative algebra over R = C[q,q−1]with generators E ,F ,K ,K−1 and relations:• KK−1 = 1 = K−1K ,• KEK−1 = q2E ,• KFK−1 = q−2F ,• EF − FE = K 2−K−2

q−q−1 .

— In fact, we can equip V = R2 with a structure of a leftUq(sl2)-module such that the functor F sending X toV⊗length(X) is actually a functor

F : T → modUq(sl2).— This refers to the sl2-case from the title. One can also consider

other quantum groups.


15

The sl2-case— One representative of quantum groups is Uq(sl2) defined as

the associative non-commutative algebra over R = C[q,q−1]

with generators E ,F ,K ,K−1 and relations:• KK−1 = 1 = K−1K ,• KEK−1 = q2E ,• KFK−1 = q−2F ,• EF − FE = K 2−K−2

q−q−1 .





15


the associative non-commutative algebra over R = C[q,q−1]with generators E ,F ,K ,K−1

and relations:• KK−1 = 1 = K−1K ,• KEK−1 = q2E ,• KFK−1 = q−2F ,• EF − FE = K 2−K−2

q−q−1 .





15


the associative non-commutative algebra over R = C[q,q−1]with generators E ,F ,K ,K−1 and relations:• KK−1 = 1 = K−1K ,• KEK−1 = q2E ,• KFK−1 = q−2F ,• EF − FE = K 2−K−2

q−q−1 .





15



q−q−1 .

— In fact, we can equip V = R2 with a structure of a leftUq(sl2)-module

such that the functor F sending X toV⊗length(X) is actually a functor




15



q−q−1 .

— In fact, we can equip V = R2 with a structure of a leftUq(sl2)-module such that the functor F sending X toV⊗length(X)

is actually a functor




15



q−q−1 .


F : T → modUq(sl2).

— This refers to the sl2-case from the title. One can also considerother quantum groups.


15



q−q−1 .


F : T → modUq(sl2).— This refers to the sl2-case from the title.

One can also considerother quantum groups.


15



q−q−1 .





16

Outline

1. Knots and links


3. Categorification



17

First stepsTopological objects Alg. invariants Categorification(±,±, . . . ,±) V⊗m

a category Cm

tangle (up to isotopy) V⊗m → V⊗n

a functor Cm → Cn

link (up to isotopy) f (x) ∈ C[q,q−1]

a complex(Khovanov homology)

— We require that:• the functor be an isotopy invariant (more precisely, isotopic

tangles yield isomorphic functors),• composition of tangles correspond to composition of functors.

— We also wish to reconstruct V⊗m → V⊗n from Cm → Cn. Thisis done by passing to the Grothendieck groups. Thus, we infact categorify the algebraic invariant (vector spaces 7→categories, linear maps 7→ functors).

— In our case, C0 = Db(grmod C), F : C0 → C0 determined by FC.


17

First stepsTopological objects Alg. invariants Categorification(±,±, . . . ,±) V⊗m a category Cmtangle (up to isotopy) V⊗m → V⊗n a functor Cm → Cnlink (up to isotopy) f (x) ∈ C[q,q−1] a complex

(Khovanov homology)— We require that:

• the functor be an isotopy invariant (more precisely, isotopictangles yield isomorphic functors),

• composition of tangles correspond to composition of functors.




17


(Khovanov homology)

— We require that:• the functor be an isotopy invariant (more precisely, isotopic

tangles yield isomorphic functors),• composition of tangles correspond to composition of functors.




17



• the functor be an isotopy invariant

(more precisely, isotopictangles yield isomorphic functors),





17








17





— We also wish to reconstruct V⊗m → V⊗n from Cm → Cn.

Thisis done by passing to the Grothendieck groups. Thus, we infact categorify the algebraic invariant (vector spaces 7→categories, linear maps 7→ functors).



17





— We also wish to reconstruct V⊗m → V⊗n from Cm → Cn. Thisis done by passing to the Grothendieck groups.

Thus, we infact categorify the algebraic invariant (vector spaces 7→categories, linear maps 7→ functors).



17





— We also wish to reconstruct V⊗m → V⊗n from Cm → Cn. Thisis done by passing to the Grothendieck groups. Thus, we infact categorify the algebraic invariant

(vector spaces 7→categories, linear maps 7→ functors).



17








17






— In our case, C0 = Db(grmod C),

F : C0 → C0 determined by FC.


17








18

Grothendieck groups— If C is an abelian category,

then the (complexified)Grothendieck group of C, denoted by K0(C), is a C-vectorspace defined by• basis {[X ] | X ∈ C},• relations: [Y ] = [X ] + [Z ] for each exact sequence

0→ X → Y → Z → 0.

— If C is a triangulated category, we can do the same trick withtriangles.

— Well known: if C is abelian, then K0(C) ∼= K0(Db(C)) viaC → Db(C).

— Note:• K0(C0) ∼= C[q,q−1] for C0 = Db(grmod C). The action of q

corresponds to the shift in grading.• We want K0(Cm) ∼= V⊗m ∼= R2m

(where R = C[q,q−1]).


18

Grothendieck groups— If C is an abelian category, then the (complexified)

Grothendieck group of C, denoted by K0(C), is a C-vectorspace defined by• basis {[X ] | X ∈ C},

• relations: [Y ] = [X ] + [Z ] for each exact sequence0→ X → Y → Z → 0.







18


Grothendieck group of C, denoted by K0(C), is a C-vectorspace defined by• basis {[X ] | X ∈ C},• relations: [Y ] = [X ] + [Z ] for each exact sequence

0→ X → Y → Z → 0.







18



0→ X → Y → Z → 0.



— Note:• K0(C0) ∼= C[q,q−1] for C0 = Db(grmod C).

The action of qcorresponds to the shift in grading.

• We want K0(Cm) ∼= V⊗m ∼= R2m(where R = C[q,q−1]).


18



0→ X → Y → Z → 0.




corresponds to the shift in grading.

• We want K0(Cm) ∼= V⊗m ∼= R2m(where R = C[q,q−1]).


18



0→ X → Y → Z → 0.







19

What are the categories and functors?

— Bernstein, Frenkel and Khovanov 1999; Stroppel 2005:• Cm =

⊕mi=0Oi,m−i

where each Oi,m−i is a graded version of aparabolic subcategory of the principal block of the category Ofor slm.

• Cm → Cn are so-called projective functors.— Khovanov 2002; Khovanov and Chen 2006:

• Cm = Db(grmodA), where A is a combinatorially defined gradedalgebra.

— Cautis and Kamnitzer• Cm are derived categories of equivariant sheaves on certain

smooth projective varieties.

— Beyond the sl2-case: Cautis and Kamnitzer, Sussan,Mazorchuk and Stroppel and others.


19



⊕mi=0Oi,m−i where each Oi,m−i is a graded version of a

parabolic subcategory of the principal block of the category Ofor slm.







19





• Cm → Cn are so-called projective functors.

— Khovanov 2002; Khovanov and Chen 2006:• Cm = Db(grmodA), where A is a combinatorially defined graded

algebra.— Cautis and Kamnitzer

• Cm are derived categories of equivariant sheaves on certainsmooth projective varieties.



19











20

Outline

1. Knots and links


3. Categorification



21

Representations of slk

— Recall: slk = {A ∈ Mk×k (C) | tr M = 0}.

— Example: sl2 = C · E ,F ,H where E = (00

10), F = (0

100),

H = (10

0−1).

— If A,B ∈ slk , so is [A,B] := AB − BA. slk is a Lie algebra.— Important: We have the triangular decomposition

slk = n− ⊕ h⊕ n+.— An slk -module is complex vector space W together with a

linear action w 7→ A · w for each A ∈ slk such that[A,B]w = A · (B · w)− B · (A · w) for all A,B ∈ slk and w ∈W .

— A weight space Wα corresponding to α : h→ C is defined asWα = {w ∈W | H · w = α(H)w for each H ∈ h}.


21


— Recall: slk = {A ∈ Mk×k (C) | tr M = 0}.— Example: sl2 = C · E ,F ,H where E = (0

010), F = (0

100),

H = (10

0−1).






21



010), F = (0

100),

H = (10

0−1).

— If A,B ∈ slk , so is [A,B] := AB − BA.

slk is a Lie algebra.— Important: We have the triangular decomposition





21



010), F = (0

100),

H = (10

0−1).

— If A,B ∈ slk , so is [A,B] := AB − BA. slk is a Lie algebra.

— Important: We have the triangular decompositionslk = n− ⊕ h⊕ n+.

— An slk -module is complex vector space W together with alinear action w 7→ A · w for each A ∈ slk such that[A,B]w = A · (B · w)− B · (A · w) for all A,B ∈ slk and w ∈W .



21



010), F = (0

100),

H = (10

0−1).


slk = n− ⊕ h⊕ n+.

— An slk -module is complex vector space W together with alinear action w 7→ A · w for each A ∈ slk such that[A,B]w = A · (B · w)− B · (A · w) for all A,B ∈ slk and w ∈W .



21



010), F = (0

100),

H = (10

0−1).


slk = n− ⊕ h⊕ n+.— An slk -module is complex vector space W

together with alinear action w 7→ A · w for each A ∈ slk such that[A,B]w = A · (B · w)− B · (A · w) for all A,B ∈ slk and w ∈W .



21



010), F = (0

100),

H = (10

0−1).



linear action w 7→ A · w for each A ∈ slk

such that[A,B]w = A · (B · w)− B · (A · w) for all A,B ∈ slk and w ∈W .



21



010), F = (0

100),

H = (10

0−1).






22

Weight lattices and root systems

sl2:

−4 −3 −2 −1 0 1 2 3 4

EF

sl3:

E1F1

E2

F2

E12

F12

Legend:fundamental weights

rootsother weights walls

Components delimited by walls are called chambers.


22


sl2:−4 −3 −2 −1 0 1 2 3 4

EF

sl3:

E1F1

E2

F2

E12

F12

Legend:fundamental weights

roots

other weights

wallsComponents delimited by walls are called chambers.


22


sl2:−4 −3 −2 −1 0 1 2 3 4

EF

sl3: E1F1

E2

F2

E12

F12

Legend:fundamental weights rootsother weights

wallsComponents delimited by walls are called chambers.


22


sl2:−4 −3 −2 −1 0 1 2 3 4

EF

sl3: E1F1

E2

F2

E12

F12

Legend:fundamental weights rootsother weights walls

Components delimited by walls are called chambers.


23

The category O

— Introduced by Bernstein, Gelfand and Gelfand, 1976.

— The category O for slk is the category formed by allslk -modules W such that

• W is finitely generated over slk ,• W decomposes into weight spaces,• W is locally n+-finite.

— There is a canonical decomposition O =⊕Oλ, where λ runs

over all weights in a fixed chamber (including walls),a so-called block decomposition.

— For each λ, Oλ is equivalent to modAλ, where Aλ is a gradedfinite dimensional algebra over C (grading due to Beilinson,Ginzburg and Soergel, 1996).


23

The category O

— Introduced by Bernstein, Gelfand and Gelfand, 1976.— The category O for slk is the category formed by all

slk -modules W such that






23

The category O




— There is a canonical decomposition O =⊕Oλ,

where λ runsover all weights in a fixed chamber (including walls),a so-called block decomposition.



23

The category O





over all weights in a fixed chamber (including walls),

a so-called block decomposition.— For each λ, Oλ is equivalent to modAλ, where Aλ is a graded

finite dimensional algebra over C (grading due to Beilinson,Ginzburg and Soergel, 1996).


23

The category O








23

The category O






— For each λ, Oλ is equivalent to modAλ,

where Aλ is a gradedfinite dimensional algebra over C (grading due to Beilinson,Ginzburg and Soergel, 1996).


23

The category O






— For each λ, Oλ is equivalent to modAλ, where Aλ is a gradedfinite dimensional algebra over C

(grading due to Beilinson,Ginzburg and Soergel, 1996).


23

The category O








24

Facts about O, continued— For each pair λ, µ, we have a so-called translation functor

θµλ : Oλ → Oµ

which is exact and preserves projectivity and injectivity.— These functors are often equivalences, unless we translate

onto a wall or out of a wall.— If λ is not on a wall and µ is on a wall, then

θλµ ◦ θ

µλ : Oλ → Oλ

is called a translation through the wall. These are exactly thefunctors Bernstein-Frenkel-Khovanov-Stroppel used for thecategorification.

— To be precise, there are a few more technicalaspects—grading, parabolic subcategories, Enright-Sheltonequivalences and choosing suitable bases for K0(Oλ) (Vermamodules).


24



which is exact and preserves projectivity and injectivity.

— These functors are often equivalences, unless we translateonto a wall or out of a wall.

— If λ is not on a wall and µ is on a wall, thenθλµ ◦ θ

µλ : Oλ → Oλ




24



which is exact and preserves projectivity and injectivity.— These functors are often equivalences,

unless we translateonto a wall or out of a wall.


µλ : Oλ → Oλ




24




onto a wall or out of a wall.


µλ : Oλ → Oλ




24





θλµ ◦ θ

µλ : Oλ → Oλ

is called a translation through the wall.

These are exactly thefunctors Bernstein-Frenkel-Khovanov-Stroppel used for thecategorification.



24





θλµ ◦ θ

µλ : Oλ → Oλ




24





θλµ ◦ θ

µλ : Oλ → Oλ


— To be precise, there are a few more technicalaspects

—grading, parabolic subcategories, Enright-Sheltonequivalences and choosing suitable bases for K0(Oλ) (Vermamodules).


24





θλµ ◦ θ

µλ : Oλ → Oλ


— To be precise, there are a few more technicalaspects—grading,

parabolic subcategories, Enright-Sheltonequivalences and choosing suitable bases for K0(Oλ) (Vermamodules).


24





θλµ ◦ θ

µλ : Oλ → Oλ


— To be precise, there are a few more technicalaspects—grading, parabolic subcategories,

Enright-Sheltonequivalences and choosing suitable bases for K0(Oλ) (Vermamodules).


24





θλµ ◦ θ

µλ : Oλ → Oλ


— To be precise, there are a few more technicalaspects—grading, parabolic subcategories, Enright-Sheltonequivalences

and choosing suitable bases for K0(Oλ) (Vermamodules).


24





θλµ ◦ θ

µλ : Oλ → Oλ




categorification---the sl2 case - functor valued invariants of knots, …stovicek/math/... ·...

Documents