sl
n-link homologies using ˙ U
q( sl
m)-highest weight theory
Daniel Tubbenhauer
Themis not a typo!
April 2014
FtF4F3F2F3Fbvh{5}
˜Γ( ):F2F3→F3F2
❚❚
❚❚
))
❚❚
❚❚ FtF3F4F2F3Fbvh{4} L
Γ(˜ ):F3F4→F4F3
❥❥
❥❥
55
❥❥
❥❥
Γ(˜❚❚❚ ):F2F3→F2F3
❚
))
❚❚
❚❚
FtF3F4F2F3Fbvh{6}
FtF4F3F2F3Fbvh{5}
−˜Γ( ):F3F4→F4F3
❥❥
❥❥
55
❥❥
❥❥
Daniel Tubbenhauer April 2014 1 / 36
1 A diagrammatic presentation sl2-webs
Connection toRep(Uq(sl2))
How can one prove the graphical representation?
2 Connection to thesln-link polynomials The Jones polynomial
Links as F’s
It’s Reshetikhin-Turaev’s coloredsln-link polynomial
3 Its categorification!
“Higher” representation theory Categorifiedq-skew Howe duality The Khovanov homology
Daniel Tubbenhauer April 2014 2 / 36
An old story: Rumer, Teller and Weyl (1932)
Daniel Tubbenhauer sl2-webs April 2014 3 / 36
Think topologically but write algebraically
Think:
Write:
Advantage: Decomposition `a la Morse intobasic pieces.
Ignore dotted red lines: We used them to solvesign issues(functoriality of Khovanov homology for example). Theyencodethe fact for quantum groups the antipode (dual representations) comes with asign.
Daniel Tubbenhauer sl2-webs April 2014 4 / 36
The (rigid) sl
2-webs - the objects
Definition - Part I
The(rigid) sl2-web spiderSp(Uq(sl2))is the monoidal, ¯Q(q)-linear 1-category consisting of the following.
Theobjectsare ordered compositions~k ofd∈Nwith only 0,1,2 as entries.
Stated otherwise: The objects arem-tuples
~k= (k1, . . . ,km) such that Xm
j=1
kj =d,kj ∈ {0,1,2}.
Example:
d= 10 : ~k1= (2,2,0,1,2,0,1,2,0,0) and ~k2= (2,2,2,2,2,0,0,0,0,0) We call an object ahighest weightobject ifkj ∈ {0,2} andkj≥kj+1, e.g.~k2.
Daniel Tubbenhauer sl2-webs April 2014 5 / 36
The (rigid) sl
2-webs - the generating 1-morphisms
Definition - Part II
Thegenerating 1-morphismsarew:~k →~k′ areedge-labeled graphssuch that The vertices areeither1-valent and part of the bottom (where we place~k) or top (where we place~k′) boundary or3-valent. The labels are from the set {0,1,2}and edges that end in a 1-valent vertex kj should have labelkj. Wedo notpicture edges labeled 0 and picture the edges labeled 2dotted.
The generators are eitheridentities
k1 k2 k3 k4
k1 k2 k3 k4
k1 k2 k3 k4 For example:
1 0 1 2
1 0 1 2
1 0 1 2
Orladders
k1 k2
k1±k k2∓k k
k=0,1,2
For example:
0 2
2 0
2 or
2 0
1 1
1
Daniel Tubbenhauer sl2-webs April 2014 6 / 36
The (rigid) sl
2-webs - and all the rest
Definition - Part III
The ¯Q(q)-linear composition◦ isstacking(see below).
The monoidal structure⊗is given byjuxtaposition, e.g.
1 0 1 2
1 0 1 2
1 0 1 2 ⊗
1 2
2 1
=
1 0 1 2
1 0 1 2
1 0 1 2
1 2
2 1
All 1-morphisms should begeneratedby identities and ladders by◦ and⊗.
Relations are thecircle removalsandisotopies, e.g. ([2] =q+q−1)
2 0
1 1
2 0
= [2]·
2 0
2 0
2 0
and
1 2
2 ◦ 1
2 1
1 2
=
1 2
2 1
1 2
=
1 2
1 2
1 2
=
1 2
1 ◦ 2
1 2
1 2
Daniel Tubbenhauer sl2-webs April 2014 7 / 36
The quantum algebra U
q( sl
m)
Definition
Form∈N>1thequantum special linear algebraUq(slm) is the associative, unital Q(q)-algebra¯ generated byKi±1 andEi andFi, fori= 1, . . . ,m−1 subject the followingrelations.
KiKj =KjKi, KiKi−1=Ki−1Ki = 1, EiFj−FjEi=δi,j
KiKi+1−1−Ki−1Ki+1
q−q−1 , KiEj =q(ǫi,αj)EjKi,
KiFj =q−(ǫi,αj)FjKi,
Ei2Ej−[2]EiEjEi+EjEi2= 0, if |i−j|= 1, EiEj−EjEi = 0, else,
Fi2Fj−[2]FiFjFi+FjFi2= 0, if |i−j|= 1, FiFj−FjFi= 0, else.
Daniel Tubbenhauer Connection toRep(Uq(sl2)) April 2014 8 / 36
The idempotented version
Definition(Beilinson-Lusztig-MacPherson)
For each~k ∈Zm−1 adjoin anidempotent1~k (think: projection to the~k-weight space!) toUq(slm) and add some relations, e.g.
1~k1~k′ =δ~k,~k′1~k and Fi1~k = 1~k−αFi and K±i1~k =q±~ki1~k (noK′s anymore!).
Theidempotented quantum special linear algebrais defined by U˙q(slm) = M
~k,~k′∈Zm−1
1~kUq(slm)1~k′.
Itslower part ˙U−q(slm)is the subalgebra ofonlyF’s.
An important fact: The ˙Uq(slm) has the“same”representation theory asUq(slm) and ˙U−q(slm) suffices to describe it.
Daniel Tubbenhauer Connection toRep(Uq(sl2)) April 2014 9 / 36
The category Rep(U
q( sl
2))
Definition
Therepresentation categoryRep(Uq(sl2))is the monoidal, ¯Q(q)-linear 1-category consisting of:
Theobjectsare finite tensor products of theUq(sl2)-representations ΛkQ¯2. Denote them by~k = (k1, . . . ,km) withki ∈ {0,1,2}.
The1-cellsw:~k→~k′ areUq(sl2)-intertwiners.
Composition◦of 1-cells iscomposition of intertwinersand⊗is theordered tensor product.
It is worth noting that Λ0Q¯2= ¯Qis the trivialUq(sl2)-representation, Λ2Q¯2∼= ¯Q its dual and Λ1Q¯2= ¯Q2 is the (self-dual)Uq(sl2)-vector representation.
Example:~k = (2,0),~k= (1,1) and ¯Q2=hx+1,x−1i. Then
cup : ¯Q∼= Λ2Q¯2⊗Q¯ →Λ1Q¯2⊗Λ1Q¯2∼= ¯Q2⊗Q¯2,17→x+1⊗x−1−q−1·x−1⊗x+1
forms a basis ofMor(~k, ~k′).
Daniel Tubbenhauer Connection toRep(Uq(sl2)) April 2014 10 / 36
Intertwiner are pictures
Theorem(Kuperberg 1997, n > 3: Cautis-Kamnitzer-Morrison 2012)
The 1-categoriesRep(Uq(sl2)) andSp(Uq(sl2)) areequivalent.
Example: cup = cup, i.e.
cup : Λ2Q¯2⊗Q¯ →Λ1Q¯2⊗Λ1Q¯2 7→
2 0
1 1
1
Question
How can one prove such a statement?
Finding the generators forRep(Uq(sl2)) isdoable, but...
Finding acompleteset of relations isvery hard!
Daniel Tubbenhauer Connection toRep(Uq(sl2)) April 2014 11 / 36
An instance of q -skew Howe duality
The commuting actions of ˙Uq(slm) and ˙Uq(sl2) on M
a1+···+am=N
(Λa1Q¯2⊗ · · · ⊗ΛamQ¯2)∼= ΛN( ¯Qm⊗Q¯2)∼= M
a1+a2=N
(Λa1Q¯m⊗Λa2Q¯m)
introducea ˙Uq(slm)-action on the left side and a ˙Uq(sl2)-action on the right side.
The left and right side are ˙Uq(slm)- and ˙Uq(sl2)-weight spaceswith weights
~kU˙q(slm) = (a1−a2, . . . ,am−1−am) and ~kU˙q(sl2) = (a1−a2) respectively.
Here the ΛkQ¯lqare irreducible ˙Uq(sll)-representations (l∈ {2,m}).
Daniel Tubbenhauer How can one prove the graphical representation? April 2014 12 / 36
Graphical quantum skew Howe duality
Theorem
There is anU˙q(slm)-actiononSp(Uq(sl2))m (objects of lengthm)!
1~k 7→
... ...
k1 ki−1 ki ki+1 ki+2 km
k1 ki−1 ki ki+1 ki+2 km
Ei1~k, Fi1~k 7→
... ...
k1 ki−1 ki ki+1 ki+2 km k1 ki−1 ki±1 ki+1∓1 ki+2 km
That is, we stack these pictures ontopof a givensl2-web.
Thus,Sp(Uq(sl2))mis a U˙q(slm)-module andnot just aUq(sl2)-module.
Daniel Tubbenhauer How can one prove the graphical representation? April 2014 13 / 36
Graphical quantum skew Howe duality - even better
Theorem
The ˙Uq(slm)-submodule W2((2ℓ)) = M
~k∈Λ(m,2ℓ)
W2(~k) = M
~k∈Λ(m,2ℓ)
MorSp(Uq(sl2))((2ℓ), ~k),
called thesl2-web space, is a ˙Uq(slm)-module of highest weight (2ℓ). Thus, it is generated by ˙U−q(slm) (akaF’s suffice).
↑
U˙−q(slm)−action
↓
Uq(sl2)−web →
←2 2 0 0
F2
2 1 1 0
F3
2 1 0 1
F1
1 2 0 1
F2
1 1 1 1
Λ2Q¯2⊗Λ2Q¯2⊗Λ0Q¯2⊗Λ0Q¯2
↑F2
Λ2Q¯2⊗Λ1Q¯2⊗Λ1Q¯2⊗Λ0Q¯2
↑F3
Λ2Q¯2⊗Λ1Q¯2⊗Λ0Q¯2⊗Λ1Q¯2
↑F1
Λ1Q¯2⊗Λ2Q¯2⊗Λ0Q¯2⊗Λ1Q¯2
↑F2
Λ1Q¯2⊗Λ1Q¯2⊗Λ1Q¯2⊗Λ1Q¯2
Daniel Tubbenhauer How can one prove the graphical representation? April 2014 14 / 36
An instance of ˙ U
q( sl
m)-highest weight theory
What is theupshotof this?
“Explains”theUq(sl2)-intertwiner as instances of the (well developed) U˙q(slm)-highest weight theory.
The action of theF’s isexplicit and inductive- a powerful tool to prove statements.
Allthe relations follow from the well-known ones from ˙Uq(slm), e.g.
E1F1v20−F1E1v20
| {z }
=0
=K1K2−1−K1−1K2
q−q−1
| {z }
=[2]120in ˙Uq(slm)
v20⇒
2 0
1 1
2 0
F1
E1
= [2]·
2 0
2 0
2 0
120
Even better: ˙U−q(slm)sufficesfor everything!
Daniel Tubbenhauer How can one prove the graphical representation? April 2014 15 / 36
The famous Jones polynomial
LetLD be a diagram of an oriented link. Set [2] =q+q−1 and n+= number of crossings n−= number of crossings
Definition/Theorem(Jones 1984, Kauffman 1987)
Thebracket polynomialof the diagramLD (without orientations) is a polynomial hLDi ∈Z[q,q−1] given by the following rules.
h∅i= 1 (normalization).
h i=h i −qh i (recursion step 1).
h ∐LDi= [2]· hLDi(recursion step 2).
[2]J(LD) = (−1)n−qn+−2n−hLDi(Re-normalization).
The polynomialJ(·)∈Z[q,q−1] is an invariantof oriented links.
Daniel Tubbenhauer The Jones polynomial April 2014 16 / 36
Exempli gratia
* +
:
■■
■■
▲▲
▲▲
▲ rr rr r
✉✉
✉✉
[2]2 −2q·[2] +q2·[2]2
Thus,J(Hopf) =q5+q, i.e the Hopf link isnot trivial!
Daniel Tubbenhauer The Jones polynomial April 2014 17 / 36
Crossings measure the difference between F
iF
i+1and F
i+1F
iDefine anUq(sl2)-intertwiner called positive crossingT1+as follows.
T1+
=
1 1
1 1
1 1
111
0-resolution
−q·
1 1
2 0
1 1
E1
F1
1-resolution
= 111v11−q·F1E1v11.
Wait: It is a ˙Uq(slm)-highest weight module:NoE’sare needed!
T1+
=
1 1 0
1 0 1
0 1 1
F2 F1
0-resolution
−q·
1 1 0
0 2 0
0 1 1
F2
F1
1-resolution
= F1F2v110−q·F2F1v110.
Exercise: Do .
Daniel Tubbenhauer Links asF’s April 2014 18 / 36
U ˙
−q( sl
m) knows link diagrams
Using theseTk+andTk− together with theF’s we can write link diagrams as
↑
U˙−q(slm)−action
↓
Uq(sl2)−“web”→
2 ←
1 1 1 1 0 0 0 0 0 0 0
0 1 2 2 1 1 0 0 0 0 0
0 0 0 0 0 1 1 1 1 0 0
0 0 0 0 0 0 0 1 0 0 1 2
2 2 1 0 1 1 1 1 1 2 1 1 0
0 0 0 1 1 1 1 1 1 1 2 2 2
F1
F2 F3 F4
F5
F2 F3 F4
T1+ T2+
F4
F5
F3
F4
qH(Hopf) =F4(2)F4F3F5F4T2+T1+F4F3F2F5F4F3F2F1F4(2)F3(2)F2(2)v220000.
Daniel Tubbenhauer Links asF’s April 2014 19 / 36
Jumping from a highest to a lowest weight
↑
U˙−q(slm)−action
↓
Uq(sln)−web →
2 ←
1 1 1 1 0 0 0 0 0 0 0 0
0 1 2 2 1 2 1 1 0 0 0 0 0
0 0 0 0 0 0 1 0 1 1 1 0 0
0 0 0 0 0 0 0 1 1 0 0 1 2
2 2 1 0 1 1 1 1 1 2 1 1 0
0 0 0 1 1 1 1 1 1 1 2 2 2
F1
F2 F3 F4
F5
F2 F3 F4
F1
F2
F3
F2
F4
F5
F3
F4
Resolutions are strings ofF’s jumping from a highest to a lowest ˙Uq(slm)-weight space. Both are1-dimensional, thus, this gives aquantum number!
Daniel Tubbenhauer Links asF’s April 2014 20 / 36
It works fine!
Definition
Given a link diagramLD. Put it in a position `a la Morse and obtain qH(LD).
DefinePRT2 (LD)∈Z[q,q−1] as theq-weighed, alternating sum over all resolutions multiplied by (−1)n−qn+−2n− (re-normalization).
Exercise: Check that this doesnotdepend on the choice of the position `a la Morse by using the relations from ˙Uq(slm).
There is asln-variantof this, denoted byPRTn (LD), that can also becolored with different fundamentalUq(sln)-representations ΛkQ¯n.
Theorem
The (colored) polynomialPRTn (LD) is an invariant of links. Moreover, itis the (colored) Reshetikhin-Turaevsln-link polynomial.
In particular: The polynomialPRT2 (LD) colored with theUq(sl2)-tensor representations Λ1Q¯2 givesthe Jones polynomialJ(LD).
Daniel Tubbenhauer It’s Reshetikhin-Turaev’s coloredsln-link polynomial April 2014 21 / 36
The sl
n-link polynomials using sl
m-symmetries
Let ussummarizethe connection between (colored)sln-link polynomials and the U˙q(slm)- ˙Uq(sln)-skew Howe duality.
Reshetikhin-Turaev: Thesln-link polynomialsPRTn (·)areUq(sln)-intertwiner.
Uq(sln)-intertwinerarevectors in hom’s between ˙Uq(slm)-weight spaces.
OnlyF’s: ˙U−q(slm)suffices. Conclusion: The (colored)sln-link polynomials PRTn (·) areinstances of ˙Uq(slm)-highest weight theory!
Even better: There exists a fixedmfor each linkLsuch that ˙Uq(slm)-highest weight theorygovernsall thesln-polynomials ofL.
IfLD is a link diagram, thenPRTn (LD) is obtained byjumping viaF’sfrom a highest ˙Uq(slm)-weightvh to a lowest ˙Uq(slm)-weightvl!
PRTn (LD) “measure” thedifferencebetween different “ways” from vh tovl.
Daniel Tubbenhauer It’s Reshetikhin-Turaev’s coloredsln-link polynomial April 2014 22 / 36
Please, fasten your seat belts!
Let’scategorifyeverything!
Daniel Tubbenhauer It’s Reshetikhin-Turaev’s coloredsln-link polynomial April 2014 23 / 36
Categorified symmetries
LetAbe some algebra,M be aA-module andCbe asuitablecategory. Denote by a,a1,a2some words in some generating set.
“Usual” /o /o /o /o /o /o /o /o /o /o /o /o /o /o /o /o //“Higher”
a7→fa∈End(M) /o /o /o /o /o /o /o /o /o /o /o //a7→ Fa∈End(C)
(fa1·fa2)(m) =fa1a2(m) /o /o /o /o /o /o /o //(Fa1◦ Fa2) Xϕ∼=Fa1a2
X ϕ
a1∼a2 !
⇒fa1 =fa2 /o /o /o /o /o /o /o /o /o //a1∼a2 !
⇒ Fa1∼=Fa2
X ϕ
Moral:Liftmodules to categories, actions to functors, = to natural isomorphisms.
Daniel Tubbenhauer “Higher” representation theory April 2014 24 / 36
There is no direct minus
We haveseveralupshots.
The natural transformations between functors give informationinvisiblein
“classical” representation theory.
A categorical representation containsmoreinformation about the symmetries (or representations) ofA.
IfCis suitable, e.g. module categories over an algebra, then its
indecomposable objectsX gives a basis [X] ofM withpositivity properties.
In particular, considerAas aA-module. Then [X] gives a basis ofAwith positivestructure coefficientsckij via
Xai⊗Xaj ∼=M
k
Xc
ij
akk aiaj=X
k
ckijak, ckij ∈N.
Daniel Tubbenhauer “Higher” representation theory April 2014 25 / 36
How to get our hand on the natural transformations?
Reformulate: Let us seeAas acategoryAwith one object∗and a morphisma for eacha∈A. Then a categorical action can be seen as afunctor
R:A →End(C),∗ 7→ C anda7→ Fa.
But since End(C) is a2-category (2-morphisms are the natural transformations) one can expect that thereshouldbe a 2-categoryAthatcategorifiesAand a 2-functorR:A→End(C) thatcategorifiesR.
A 2-actionR
This should exist!
//
forget 2-structur
End(C)
forget 2-structur
A 1-actionR
This is given
//End(C)
Daniel Tubbenhauer “Higher” representation theory April 2014 26 / 36
The overview
U(slm)
How it should be!
Categorifiedq-skew Howe U(slm) acts
//
/o
/o
/o
/o
/o
/o
/o
/o
/o
/o
/o
/o
/o
/o
/o
/o
/o
/o
K0⊕
????
K0⊕
U˙q(slm) q-skew Howe U˙q(slm) acts
//
/o
/o
/o
/o
/o
/o
/o
/o
/o
/o
/o
/o
/o
/o
/o Sp(Uq(sln))m
This is how it should be: There is anU˙q(slm)-actionon thesln-web spiders (for us it was mostly the casen= 2)
On the left side: There isKhovanov-Lauda’s categorificationof ˙Uq(slm) denoted byU(slm) (which Ibrieflyexplain soon).
Conclusion: Thereshouldbe a 2-action of U(slm) on the top right - a suitable 2-category of “natural transformations” betweenUq(sln)-intertwiners!
Daniel Tubbenhauer Categorifiedq-skew Howe duality April 2014 27 / 36
Rigid sl
2-foams: Hopefully illustrating examples
Instead of giving theformaldefinition of the rigidsl2-foam 2-categoryFoam2
(that fills the top right from before) let me just give someexamples.
Think←
→ Write
∈hom
,
∈hom
,
Think←
→ Write
∈hom ,
!
= −
Daniel Tubbenhauer Categorifiedq-skew Howe duality April 2014 28 / 36
Khovanov-Lauda’s 2-category U ( sl
m)
Definition/Theorem (Khovanov-Lauda 2008)
The 2-categoryU(slm) is defined by (everything suitablyZ-graded and ¯Q-linear):
The objects ofU(slm) are the weights~k ∈Zm−1.
The 1-morphisms are finite formal sums of the formEi1~k{t} andFi1~k{t}.
2-cells are graded, ¯Q-vector spacesgeneratedby compositions of diagrams (additional ones with reversed arrows) as illustrated belowplus relations.
i
~k
~k+αi
i
~k
~k−αi
i
~k
~k−αi
i j
~k
i i
~k
i i
~k
We have
U˙q(slm)∼=K0⊕(U(slm))⊗Z[q,q−1]Q(q).¯ Roughly:Readthis as follows.
i i
~k ϕ:EiFi1~k →1~k
i i
~k ψ: 1~k →EiFi1~k
Daniel Tubbenhauer Categorifiedq-skew Howe duality April 2014 29 / 36
The KL-R algebra
Definition/Theorem(Khovanov-Lauda, Rouquier 2008/2009)
LetRmbe acertain direct sum of subalgebras of homU(slm)(Fi1~k{t},Fj1~k′{t}).
Thusonly downwardspointing arrows - akaonlyF’s. That is, working withRm
enables us to ignore orientations and consider only diagrams of the form
or or or
The KL-R algebra has the structure of aZ-graded, ¯Q-algebra. We have U˙−q(slm)∼=K0⊕(Rm)⊗Z[q,q−1]Q(q).¯
NOT allowed: But = 0 is the Nil-Hecke relation
Daniel Tubbenhauer Categorifiedq-skew Howe duality April 2014 30 / 36
The cyclotomic quotient
Definition(Khovanov-Lauda, Rouquier 2008/2009)
Fix a dominantslm-weight Λ. Thecyclotomic KL-R algebraRΛis the subquotient ofU(slm) defined by the subalgebra ofonly downward (onlyF’s!)pointing arrows and rightmost region labeled Λ modulo the so-calledcyclotomic relation
id i3 i2 i1
~kd
Λ
Theorem(Brundan-Kleshchev, Lauda-Vazirani, Webster, Kang-Kashiwara,... > 2008)
LetVΛbe the ˙Uq(slm)-module of highest weight Λ. We have VΛ∼=K0⊕(RΛ)⊗Z[q,q−1]Q(q)¯ as ˙Uq(slm)-modules (note that this works for moregeneralg).
Daniel Tubbenhauer Categorifiedq-skew Howe duality April 2014 31 / 36
sl
2-foamation (works for all n > 1!)
We define a 2-functor
Γ :U(slm)→Foamm2 calledsl2-foamation, roughly in the following way.
On 2-cells:We define
i,~k
7→
~ki
~ki+1
i,~k
7→
~ki
~ki+1
i,i,~k
7→
~ki
~ki+1
And some others (that are not important today).
Theorem
The 2-functor Γ :U(slm)→Foamm2 categorifiesq-skew Howe duality.
It descents down to a 2-functor ˜Γ :RΛ(~k)- (p)Modgr →Foamm2.
Daniel Tubbenhauer Categorifiedq-skew Howe duality April 2014 32 / 36
Khovanov’s categorification of the Jones polynomial
Recall the rules for the Jones polynomial.
h∅i= 1 (normalization).
h i=h i −qh i (recursion step 1).
h ∐LDi= [2]· hLDi(recursion step 2).
[2]J(LD) = (−1)n−qn+−2n−hLDi(Re-normalization).
Definition/Theorem(Khovanov 1999)
LetLD be a diagram of an oriented link. Denote byA= ¯Q[X]/X2the dual numbers with qdeg(1) = 1 and qdeg(X) =−1 - this is a Frobenius algebra with a given comultiplication ∆. We assign to it a chain complexJLDKofZ-graded Q¯-vector spaces using thecategorified rules:
J∅K= 0→Q¯ →0 (normalization).
J K= Γ
0→J K→d J K→0
withd=m,∆ (recursion step 1).
J ∐LDK=A⊗Q¯JLDK(recursion step 2).
Kh(LD) =JLDK[−n−]{n+−2n−}(Re-normalization).
ThenKh(·) is aninvariantof oriented links whose graded Euler characteristic givesχq(Kh(LD)) = [2]J(LD).
Daniel Tubbenhauer The Khovanov homology April 2014 33 / 36
Link diagrams are F ’s and differentials are KL-R crossings
Very roughly: Usecategorifiedq-skew Howe duality to express a link diagramLD
as a certain string ofonlyFi(j)’s. Obtain a complex as FtF4F3F2F3Fbvh{5}
˜Γ( ):F2F3→F3F2
❚❚
❚❚
))
❚❚
❚❚ FtF3F4F2F3Fbvh{4} L
Γ(˜ ):F3F4→F4F3
❥❥
❥❥
55
❥❥
❥❥
Γ(˜ ):F2F3→F2F3
❚❚
❚❚
))
❚❚
❚❚
FtF3F4F2F3Fbvh{6}
FtF4F3F2F3Fbvh{5}
−˜Γ( ):F3F4→F4F3
❥❥
❥❥
55
❥❥
❥❥
Theorem
This, under categorifiedq-skew Howe duality,givesthesln-link homology (because the “are”the “saddles”).
Daniel Tubbenhauer The Khovanov homology April 2014 34 / 36
The sl
n-homologies using sl
m-symmetries
Let ussummarizethe connection between sln-homologies and the higherq-skew Howe duality.
Khovanov, Khovanov-Rozansky and others: Thesln-link homology can be obtainedusing certain “sln-foams”.
OnlyF’s: The (cyclotomic) KL-Rsuffices.
Conclusion: Thesln-link homologies areinstances of highestU(slm)-weight representation theory!
IfLD is a link diagram, then they are obtained byjumping viaF’sfrom a highestU(slm)-weightVh object to a lowestU(slm)-weight objectVl! Missing: Connection to Webster’s categorification of the RT-polynomials!
Missing: Is the module category of the cyclotomic KL-R algebra braided?
Missing: Details about coloredsln-homologies have to be worked out!
Daniel Tubbenhauer The Khovanov homology April 2014 35 / 36
There is stillmuchto do...
Daniel Tubbenhauer The Khovanov homology April 2014 36 / 36
Thanks for your attention!
Daniel Tubbenhauer The Khovanov homology April 2014 36 / 36