What is...(2-)representation theory?
Or: A (fairy) tale of matrices and functors Daniel Tubbenhauer
1 Classical representation theory Main ideas
Some classical results Some examples
2 Categorical representation theory Main ideas
Some categorical results An example
A linearization of group theory
Slogan. Representation theory is group theory in vector spaces.
symmetries ofn-gons⊂ Aut(R2)
−e1
e1 e2
−e2 e1−e2 e2−e1
•
•
•
•
•
•
These symmetry groups of the regularn-gons are the so-called dihedral groups D2n=hs,t|s2=t2= 1, . . .| {z }tsts
n
=w0=. . .| {z }stst
n
i which are the easiest examples of Coxeter groups.
Examplen= 4; its Coxeter complex.
•
•
•
•
• •
• •
1 t
s st ts tst
sts w0
A linearization of group theory
Slogan. Representation theory is group theory in vector spaces.
symmetries ofn-gons⊂ Aut(R2)
−e1
e1 e2
−e2 e1−e2 e2−e1
•
•
•
•
•
• 1
1 0 0 1
,
1
These symmetry groups of the regularn-gons are the so-called dihedral groups D2n=hs,t|s2=t2= 1, . . .| {z }tsts
n
=w0=. . .| {z }stst
n
i which are the easiest examples of Coxeter groups.
Examplen= 4; its Coxeter complex.
•
•
•
•
• •
• •
1 t
s st ts tst
sts w0
A linearization of group theory
Slogan. Representation theory is group theory in vector spaces.
symmetries ofn-gons⊂ Aut(R2)
−e1
e1 e2
−e2 e1−e2 e2−e1
•
•
•
•
•
• 1 s
1 0 0 1
, 1 1
0 −1
,
1 s
These symmetry groups of the regularn-gons are the so-called dihedral groups D2n=hs,t|s2=t2= 1, . . .| {z }tsts
n
=w0=. . .| {z }stst
n
i which are the easiest examples of Coxeter groups.
Examplen= 4; its Coxeter complex.
•
•
•
•
• •
• •
1 t
s st ts tst
sts w0
A linearization of group theory
Slogan. Representation theory is group theory in vector spaces.
symmetries ofn-gons⊂ Aut(R2)
−e1
e1 e2
−e2 e1−e2 e2−e1
•
•
•
•
•
• 1 s t
1 0 0 1
, 1 1
0 −1
,
−1 0
1 1
,
1 s t
These symmetry groups of the regularn-gons are the so-called dihedral groups D2n=hs,t|s2=t2= 1, . . .| {z }tsts
n
=w0=. . .| {z }stst
n
i which are the easiest examples of Coxeter groups.
Examplen= 4; its Coxeter complex.
•
•
•
•
• •
• •
1 t
s st ts tst
sts w0
A linearization of group theory
Slogan. Representation theory is group theory in vector spaces.
symmetries ofn-gons⊂ Aut(R2)
−e1
e1 e2
−e2 e1−e2 e2−e1
•
•
•
•
•
• 1 s ts t
1 0 0 1
, 1 1
0 −1
,
−1 0
1 1
,
−1 −1
1 0
,
1 s t ts
These symmetry groups of the regularn-gons are the so-called dihedral groups D2n=hs,t|s2=t2= 1, . . .| {z }tsts
n
=w0=. . .| {z }stst
n
i which are the easiest examples of Coxeter groups.
Examplen= 4; its Coxeter complex.
•
•
•
•
• •
• •
1 t
s st ts tst
sts w0
A linearization of group theory
Slogan. Representation theory is group theory in vector spaces.
symmetries ofn-gons⊂ Aut(R2)
−e1
e1 e2
−e2 e1−e2 e2−e1
•
•
•
•
•
• 1 s ts t
st
1 0 0 1
, 1 1
0 −1
,
−1 0
1 1
,
−1 −1
1 0
,
0 1
−1 −1
,
1 s t ts st
These symmetry groups of the regularn-gons are the so-called dihedral groups D2n=hs,t|s2=t2= 1, . . .| {z }tsts
n
=w0=. . .| {z }stst
n
i which are the easiest examples of Coxeter groups.
Examplen= 4; its Coxeter complex.
•
•
•
•
• •
• •
1 t
s st ts tst
sts w0
A linearization of group theory
Slogan. Representation theory is group theory in vector spaces.
symmetries ofn-gons⊂ Aut(R2)
−e1
e1 e2
−e2 e1−e2 e2−e1
•
•
•
•
•
• 1 s ts t w0 st
1 0 0 1
, 1 1
0 −1
,
−1 0
1 1
,
−1 −1
1 0
,
0 1
−1 −1
,
0 −1
−1 0
1 s t ts st sts=tst
w0
These symmetry groups of the regularn-gons are the so-called dihedral groups D2n=hs,t|s2=t2= 1, . . .| {z }tsts
n
=w0=. . .| {z }stst
n
i which are the easiest examples of Coxeter groups.
Examplen= 4; its Coxeter complex.
•
•
•
•
• •
• •
1 t
s st ts tst
sts w0
A linearization of group theory
Slogan. Representation theory is group theory in vector spaces.
symmetries ofn-gons⊂ Aut(R2)
−e1
e1 e2
−e2 e1−e2 e2−e1
•
•
•
•
•
• 1 s ts t w0 st
1 0 0 1
, 1 1
0 −1
,
−1 0
1 1
,
−1 −1
1 0
,
0 1
−1 −1
,
0 −1
−1 0
1 s t ts st sts=tst
w0
These symmetry groups of the regularn-gons are the so-called dihedral groups D2n=hs,t|s2=t2= 1, . . .| {z }tsts
n
=w0=. . .| {z }stst
n
i which are the easiest examples of Coxeter groups.
Examplen= 4; its Coxeter complex.
•
•
•
•
• •
• •
1 t
s st ts tst
sts w0
Pioneers of representation theory
LetGbe a finite group.
Frobenius∼1895++, Burnside∼1900++. Representation theory is the useful?
study of linear group actions
M: G−→ Aut(V),
withVbeing some vector space. (Called modules or representations.)
The “atoms” of such an action are called simple. A module is called semisimple if it is a direct sum of simples.
Maschke∼1899. All modules are built out of simples (“Jordan–H¨older”
filtration).
“M(g) = a matrix inAut(V)”
We want to have a categorical version of this!
“M(a) = a matrix inEnd(V)”
We want to have a categorical version of this.
I am going to explain what we can do at present.
Pioneers of representation theory
LetGbe a finite group.
Frobenius∼1895++, Burnside∼1900++. Representation theory is the useful?
study of linear group actions
M: G−→ Aut(V),
withVbeing some vector space. (Called modules or representations.)
The “atoms” of such an action are called simple. A module is called semisimple if it is a direct sum of simples.
Maschke∼1899. All modules are built out of simples (“Jordan–H¨older”
filtration).
“M(g) = a matrix inAut(V)”
We want to have a categorical version of this!
“M(a) = a matrix inEnd(V)”
We want to have a categorical version of this.
I am going to explain what we can do at present.
Pioneers of representation theory
LetAbe a finite-dimensional algebra.
Noether∼1928++. Representation theory is the useful? study of algebra actions M:A−→ End(V),
withVbeing some vector space. (Called modules or representations.)
The “atoms” of such an action are called simple. A module is called semisimple if it is a direct sum of simples.
Noether, Schreier∼1928. All modules are built out of simples (“Jordan–H¨older” filtration).
“M(g) = a matrix inAut(V)”
We want to have a categorical version of this!
“M(a) = a matrix inEnd(V)”
We want to have a categorical version of this.
I am going to explain what we can do at present.
Pioneers of representation theory
LetAbe a finite-dimensional algebra.
Noether∼1928++. Representation theory is the useful? study of algebra actions M:A−→ End(V),
withVbeing some vector space. (Called modules or representations.)
The “atoms” of such an action are called simple. A module is called semisimple if it is a direct sum of simples.
Noether, Schreier∼1928. All modules are built out of simples (“Jordan–H¨older” filtration).
“M(g) = a matrix inAut(V)”
We want to have a categorical version of this!
“M(a) = a matrix inEnd(V)”
We want to have a categorical version of this.
I am going to explain what we can do at present.
Life is non-semisimple
collection(“category”)of modules!the world modules!chemical compounds
simples!elements
semisimple!only trivial compounds non-semisimple!non-trivial compounds
Main goal of representation theory. Find the periodic table of simples.
Example.
Back to the dihedral group, an invariant of the module is the character χwhich only remembers the
traces of the acting matrices: 1 0
0 1
, 1 1
0 −1
,
−1 0
1 1
,
−1 −1
1 0
,
0 1
−1 −1
,
0 −1
−1 0
1 s t ts st sts=tst
w0
χ= 2 χ= 0 χ= 0 χ=−1 χ=−1 χ= 0
Fact. Semisimple case:
the character determines the module
!
mass determines the chemical compound. Example.
Z/2Z→ Aut(C2), 07→ 1 0
0 1
& 17→ 0 1
1 0
Common eigenvectors: (1,1) and (1,−1) and base change gives 07→
1 0 0 1
& 17→
1 0 0 −1
and the module decomposes.
Example. Z/2Z→ Aut(f2
2), 07→ 1 0
0 1
& 17→ 0 1
1 0
Common eigenvector: (1,1) and base change gives 07→
1 0 0 1
& 17→ 1 1
0 1
and the module is non-simple, yet does not decompose. Morally: representation theory overZis never semisimple.
Life is non-semisimple
collection(“category”)of modules!the world modules!chemical compounds
simples!elements
semisimple!only trivial compounds non-semisimple!non-trivial compounds
Main goal of representation theory. Find the periodic table of simples.
Example.
Back to the dihedral group, an invariant of the module is the character χwhich only remembers the
traces of the acting matrices:
1 0 0 1
, 1 1
0 −1
,
−1 0
1 1
,
−1 −1
1 0
,
0 1
−1 −1
,
0 −1
−1 0
1 s t ts st sts=tst
w0
χ= 2 χ= 0 χ= 0 χ=−1 χ=−1 χ= 0
Fact. Semisimple case:
the character determines the module
!
mass determines the chemical compound. Example.
Z/2Z→ Aut(C2), 07→ 1 0
0 1
& 17→ 0 1
1 0
Common eigenvectors: (1,1) and (1,−1) and base change gives 07→
1 0 0 1
& 17→
1 0 0 −1
and the module decomposes.
Example. Z/2Z→ Aut(f2
2), 07→ 1 0
0 1
& 17→ 0 1
1 0
Common eigenvector: (1,1) and base change gives 07→
1 0 0 1
& 17→ 1 1
0 1
and the module is non-simple, yet does not decompose. Morally: representation theory overZis never semisimple.
Life is non-semisimple
collection(“category”)of modules!the world modules!chemical compounds
simples!elements
semisimple!only trivial compounds non-semisimple!non-trivial compounds
Main goal of representation theory. Find the periodic table of simples.
Example.
Back to the dihedral group, an invariant of the module is the character χwhich only remembers the
traces of the acting matrices:
1 0 0 1
, 1 1
0 −1
,
−1 0
1 1
,
−1 −1
1 0
,
0 1
−1 −1
,
0 −1
−1 0
1 s t ts st sts=tst
w0
χ= 2 χ= 0 χ= 0 χ=−1 χ=−1 χ= 0
Fact.
Semisimple case:
the character determines the module
!
mass determines the chemical compound.
Example. Z/2Z→ Aut(C2), 07→
1 0 0 1
& 17→ 0 1
1 0
Common eigenvectors: (1,1) and (1,−1) and base change gives 07→
1 0 0 1
& 17→
1 0 0 −1
and the module decomposes.
Example. Z/2Z→ Aut(f2
2), 07→ 1 0
0 1
& 17→ 0 1
1 0
Common eigenvector: (1,1) and base change gives 07→
1 0 0 1
& 17→ 1 1
0 1
and the module is non-simple, yet does not decompose. Morally: representation theory overZis never semisimple.
Life is non-semisimple
collection(“category”)of modules!the world modules!chemical compounds
simples!elements
semisimple!only trivial compounds non-semisimple!non-trivial compounds
Main goal of representation theory. Find the periodic table of simples.
Example.
Back to the dihedral group, an invariant of the module is the character χwhich only remembers the
traces of the acting matrices: 1 0
0 1
, 1 1
0 −1
,
−1 0
1 1
,
−1 −1
1 0
,
0 1
−1 −1
,
0 −1
−1 0
1 s t ts st sts=tst
w0
χ= 2 χ= 0 χ= 0 χ=−1 χ=−1 χ= 0
Fact. Semisimple case:
the character determines the module
!
mass determines the chemical compound.
Example.
Z/2Z→ Aut(C2), 07→
1 0 0 1
& 17→
0 1 1 0
Common eigenvectors: (1,1) and (1,−1) and base change gives 07→
1 0 0 1
& 17→
1 0 0 −1
and the module decomposes.
Example. Z/2Z→ Aut(f2
2), 07→ 1 0
0 1
& 17→ 0 1
1 0
Common eigenvector: (1,1) and base change gives 07→
1 0 0 1
& 17→ 1 1
0 1
and the module is non-simple, yet does not decompose. Morally: representation theory overZis never semisimple.
Life is non-semisimple
collection(“category”)of modules!the world modules!chemical compounds
simples!elements
semisimple!only trivial compounds non-semisimple!non-trivial compounds
Main goal of representation theory. Find the periodic table of simples.
Example.
Back to the dihedral group, an invariant of the module is the character χwhich only remembers the
traces of the acting matrices: 1 0
0 1
, 1 1
0 −1
,
−1 0
1 1
,
−1 −1
1 0
,
0 1
−1 −1
,
0 −1
−1 0
1 s t ts st sts=tst
w0
χ= 2 χ= 0 χ= 0 χ=−1 χ=−1 χ= 0
Fact. Semisimple case:
the character determines the module
!
mass determines the chemical compound.
Example.
Z/2Z→ Aut(C2), 07→
1 0 0 1
& 17→
0 1 1 0
Common eigenvectors: (1,1) and (1,−1) and base change gives 07→
1 0 0 1
& 17→
1 0 0 −1
and the module decomposes.
Example.
Z/2Z→ Aut(f2
2), 07→
1 0 0 1
& 17→
0 1 1 0
Common eigenvector: (1,1) and base change gives 07→
1 0 0 1
& 17→
1 1 0 1
and the module is non-simple, yet does not decompose.
Morally: representation theory overZis never semisimple.
Life is non-semisimple
collection(“category”)of modules!the world modules!chemical compounds
simples!elements
semisimple!only trivial compounds non-semisimple!non-trivial compounds
Main goal of representation theory. Find the periodic table of simples.
Example.
Back to the dihedral group, an invariant of the module is the character χwhich only remembers the
traces of the acting matrices: 1 0
0 1
, 1 1
0 −1
,
−1 0
1 1
,
−1 −1
1 0
,
0 1
−1 −1
,
0 −1
−1 0
1 s t ts st sts=tst
w0
χ= 2 χ= 0 χ= 0 χ=−1 χ=−1 χ= 0
Fact. Semisimple case:
the character determines the module
!
mass determines the chemical compound.
Example.
Z/2Z→ Aut(C2), 07→
1 0 0 1
& 17→
0 1 1 0
Common eigenvectors: (1,1) and (1,−1) and base change gives 07→
1 0 0 1
& 17→
1 0 0 −1
and the module decomposes.
Example.
Z/2Z→ Aut(f2
2), 07→
1 0 0 1
& 17→
0 1 1 0
Common eigenvector: (1,1) and base change gives 07→
1 0 0 1
& 17→
1 1 0 1
and the module is non-simple, yet does not decompose.
Morally: representation theory overZis never semisimple.
The strategy
“Groups, as men, will be known by their actions.” – Guillermo Moreno The study of group actions is of fundamental importance in mathematics and related field. Sadly, it is also very hard.
Representation theory approach. The analog linear problem of classifying G-modules has a satisfactory answer for many groups.
Problem involving a group action
G X
Philosophy. Turn problems into linear algebra.
The strategy
“Groups, as men, will be known by their actions.” – Guillermo Moreno The study of group actions is of fundamental importance in mathematics and related field. Sadly, it is also very hard.
Representation theory approach. The analog linear problem of classifying G-modules has a satisfactory answer for many groups.
Problem involving a group action
G X
Problem involving a linear group action
k[G] kX
Philosophy. Turn problems into linear algebra.
The strategy
“Groups, as men, will be known by their actions.” – Guillermo Moreno The study of group actions is of fundamental importance in mathematics and related field. Sadly, it is also very hard.
Representation theory approach. The analog linear problem of classifying G-modules has a satisfactory answer for many groups.
Problem involving a group action
G X
Problem involving a linear group action
k[G] kX
“Decomposition of the problem”
k[G] L Vi
Philosophy. Turn problems into linear algebra.
The strategy
“Groups, as men, will be known by their actions.” – Guillermo Moreno The study of group actions is of fundamental importance in mathematics and related field. Sadly, it is also very hard.
Representation theory approach. The analog linear problem of classifying G-modules has a satisfactory answer for many groups.
Problem involving a group action
G X
Problem involving a linear group action
k[G] kX
“Decomposition of the problem”
k[G] L Vi new
insights?
Philosophy. Turn problems into linear algebra.
Some theorems in classical representation theory
B AllG-modules are built out of simples.
B The character of a simpleG-module is an invariant.
B There is an injection
{simpleG-modules}/iso ,→
{conjugacy classes inG}, which is 1 : 1 in the semisimple case.
B All simples can be constructed intrinsically using the regularG-module.
“RegularG-module
=Gacting on itself.”
Find categorical versions of these facts.
Some theorems in classical representation theory
B AllG-modules are built out of simples.
B The character of a simpleG-module is an invariant.
B There is an injection
{simpleG-modules}/iso ,→
{conjugacy classes inG}, which is 1 : 1 in the semisimple case.
B All simples can be constructed intrinsically using the regularG-module.
“RegularG-module
=Gacting on itself.”
Find categorical versions of these facts.
Some theorems in classical representation theory
B AllG-modules are built out of simples.
B The character of a simpleG-module is an invariant.
B There is an injection
{simpleG-modules}/iso ,→
{conjugacy classes inG}, which is 1 : 1 in the semisimple case.
B All simples can be constructed intrinsically using the regularG-module.
“RegularG-module
=Gacting on itself.”
Find categorical versions of these facts.
Dihedral representation theory on one slide
One-dimensional modules. Mλs,λt, λs, λt∈C,s7→λs,t7→λt.
e≡0 mod 2 e6≡0 mod 2
M−1,−1,M1,−1,M−1,1,M1,1 M−1,−1,M1,1
Two-dimensional modules. Mz,z ∈C,s7→ 10−1z
,t7→ −1 0z 1
.
n≡0 mod 2 n6≡0 mod 2
Mz,z ∈V(n)−{0} Mz,z ∈V(n) V(n) ={2 cos(πk/n−1)|k = 1, . . . ,n−2}.
Proposition (Lusztig?).
The list of one- and two-dimensionalD2n-modules is a complete, irredundant list of simples.
I learned this construction from Mackaay in 2017.
Note that this requires complex parameters. In particular, this does not work overZ.
Dihedral representation theory on one slide
One-dimensional modules. Mλs,λt, λs, λt∈C,s7→λs,t7→λt.
e≡0 mod 2 e6≡0 mod 2
M−1,−1,M1,−1,M−1,1,M1,1 M−1,−1,M1,1
Two-dimensional modules. Mz,z ∈C,s7→ 10−1z
,t7→ −1 0z 1
.
n≡0 mod 2 n6≡0 mod 2
Mz,z ∈V(n)−{0} Mz,z ∈V(n) V(n) ={2 cos(πk/n−1)|k = 1, . . . ,n−2}.
Proposition (Lusztig?).
The list of one- and two-dimensionalD2n-modules is a complete, irredundant list of simples.
I learned this construction from Mackaay in 2017.
Note that this requires complex parameters. In particular, this does not work overZ.
Dihedral representation theory on one slide
One-dimensional modules. Mλs,λt, λs, λt∈C,s7→λs,t7→λt.
e≡0 mod 2 e6≡0 mod 2
M−1,−1,M1,−1,M−1,1,M1,1 M−1,−1,M1,1
Two-dimensional modules. Mz,z ∈C,s7→ 10−1z
,t7→ −1 0z 1
.
n≡0 mod 2 n6≡0 mod 2
Mz,z ∈V(n)−{0} Mz,z ∈V(n) V(n) ={2 cos(πk/n−1)|k = 1, . . . ,n−2}.
Proposition (Lusztig?).
The list of one- and two-dimensionalD2n-modules is a complete, irredundant list of simples.
I learned this construction from Mackaay in 2017.
Note that this requires complex parameters.
In particular, this does not work overZ.
Beware of infinite dimensions
Take the infinite-dimensional Weyl algebraW=Chx, δ|δx= 1 +xδi. It has a very nice infinite-dimensional module
W→ End(C[X]), x7→ ·X, δ7→d/dX, andδx= 1 +xδjust becomes Leibniz’ product rule.
However, the classification of simples is not so easy. For example,Wdoes not have any finite-dimensional module.
Why? Assume it has andx 7→some matrixM; δ7→some matrixN. Then:
tr(MN) =tr(NM) = 1 +tr(MN) ⇒ 0 = 1.
But even there representation theory help
Take the infinite Artin–Tits groupB(C) =hbi|. . .bjbibj
| {z }
mij
=bibjbi
| {z }
mij
i. Example
One can easily cook-up finite-dimensional modules which help to distinguish the elements ofB(C).
However, it is very hard and not known in general how to find faithful (“injective”) finite-dimensional modules.
Categorification in a nutshell
setN0
categoryVect
The point.
The categoryVecthas the whole power of linear algebra at hand! There is nothing comparable forN0:
N0is just a shadow ofVect.
Categorification in a nutshell
setN0
categoryVect
1 2
3 6
n
The point.
The categoryVecthas the whole power of linear algebra at hand! There is nothing comparable forN0:
N0is just a shadow ofVect.
Categorification in a nutshell
setN0
categoryVect
1 2
3 6
n k
k2 k3
k6 kn
The point.
The categoryVecthas the whole power of linear algebra at hand! There is nothing comparable forN0:
N0is just a shadow ofVect.
Categorification in a nutshell
setN0
categoryVect
1 2
3 6
n k
k2 k3
k6 kn
dim.
The point.
The categoryVecthas the whole power of linear algebra at hand! There is nothing comparable forN0:
N0is just a shadow ofVect.
Categorification in a nutshell
setN0
categoryVect
1 2
3 6
n k
k2 k3
k6 kn
dim.
+
The point.
The categoryVecthas the whole power of linear algebra at hand! There is nothing comparable forN0:
N0is just a shadow ofVect.
Categorification in a nutshell
setN0
categoryVect
1 2
3 6
n k
k2 k3
k6 kn
dim.
+
⊕
The point.
The categoryVecthas the whole power of linear algebra at hand! There is nothing comparable forN0:
N0is just a shadow ofVect.
Categorification in a nutshell
setN0
categoryVect
1 2
3 6
n k
k2 k3
k6 kn
dim.
+
⊕
·
The point.
The categoryVecthas the whole power of linear algebra at hand! There is nothing comparable forN0:
N0is just a shadow ofVect.
Categorification in a nutshell
setN0
categoryVect
1 2
3 6
n k
k2 k3
k6 kn
dim.
+
⊕
·
⊗
The point.
The categoryVecthas the whole power of linear algebra at hand! There is nothing comparable forN0:
N0is just a shadow ofVect.
Categorification in a nutshell
setN0
categoryVect
1 2
3 6
n k
k2 k3
k6 kn
dim.
+
⊕
·
⊗
>
<
The point.
The categoryVecthas the whole power of linear algebra at hand! There is nothing comparable forN0:
N0is just a shadow ofVect.
Categorification in a nutshell
setN0
categoryVect
1 2
3 6
n k
k2 k3
k6 kn
dim.
+
⊕
·
⊗
>
<
sur.
inj.
The point.
The categoryVecthas the whole power of linear algebra at hand! There is nothing comparable forN0:
N0is just a shadow ofVect.
Categorification in a nutshell
setN0
categoryVect
1 2
3 6
n k
k2 k3
k6 kn
dim.
+
⊕
·
⊗
>
<
sur.
inj.
alln-dim.
vector spaces
V f W A universe itself!
The point.
The categoryVecthas the whole power of linear algebra at hand! There is nothing comparable forN0:
N0is just a shadow ofVect.
Categorification in a nutshell
setN0
categoryVect
1 2
3 6
n k
k2 k3
k6 kn
dim.
+
⊕
·
⊗
>
<
sur.
inj.
alln-dim.
vector spaces
V f W A universe itself!
all rank 1 2−1-matrices
k2 k g
A universe itself!
The point.
The categoryVecthas the whole power of linear algebra at hand! There is nothing comparable forN0:
N0is just a shadow ofVect.
Categorification in a nutshell
setN0
categoryVect
1 2
3 6
n k
k2 k3
k6 kn
dim.
+
⊕
·
⊗
>
<
sur.
inj.
alln-dim.
vector spaces
V f W A universe itself!
all rank 1 2−1-matrices
k2 k g
A universe itself!
The point.
The categoryVecthas the whole power of linear algebra at hand!
There is nothing comparable forN0: N0is just a shadow ofVect.
2-representation theory in a nutshell
2-category categories functors nat. trafos
1-category vector spaces linear maps
0-category numbers
relate relate
relate categorify
categorify
categorify forms
forms
forms categorifies
categorifies
The ladder of categorification: in each step there is a new layer of structure which is invisible on the ladder rung below.
Goal.
Categorify the theory “representation theory” itself.
Observation.
A groupGcan be viewed as a single-object categoryG, and a module as a functor fromG
into the single-object categoryAut(V), i.e. M:G −→ Aut(V).
What one can hope for.
Problem involving a group action
G X
Problem involving a categorical group action
Decomposition of the problem into 2-simples
“lift”
new insights?
2-representation theory in a nutshell
2-category categories functors nat. trafos
1-category vector spaces linear maps
0-category numbers
relate relate
relate categorify
categorify
categorify forms
forms
forms categorifies
categorifies
The ladder of categorification: in each step there is a new layer of structure which is invisible on the ladder rung below.
Goal.
Categorify the theory “representation theory” itself.
Observation.
A groupGcan be viewed as a single-object categoryG, and a module as a functor fromG
into the single-object categoryAut(V), i.e. M:G −→ Aut(V).
What one can hope for.
Problem involving a group action
G X
Problem involving a categorical group action
Decomposition of the problem into 2-simples
“lift”
new insights?
2-representation theory in a nutshell
2-category categories functors nat. trafos
1-category vector spaces linear maps
0-category numbers
relate relate
relate categorify
categorify
categorify forms
forms
forms categorifies
categorifies
The ladder of categorification: in each step there is a new layer of structure which is invisible on the ladder rung below.
Goal.
Categorify the theory “representation theory” itself.
Observation.
A groupGcan be viewed as a single-object categoryG, and a module as a functor fromG
into the single-object categoryAut(V), i.e.
M:G −→ Aut(V).
What one can hope for.
Problem involving a group action
G X
Problem involving a categorical group action
Decomposition of the problem into 2-simples
“lift”
new insights?
2-representation theory in a nutshell
2-category categories functors nat. trafos
1-category vector spaces linear maps
0-category numbers
relate relate
relate categorify
categorify
categorify forms
forms
forms categorifies
categorifies
Classical representation theory lives here
The ladder of categorification: in each step there is a new layer of structure which is invisible on the ladder rung below.
Goal.
Categorify the theory “representation theory” itself.
Observation.
A groupGcan be viewed as a single-object categoryG, and a module as a functor fromG
into the single-object categoryAut(V), i.e. M:G −→ Aut(V).
What one can hope for.
Problem involving a group action
G X
Problem involving a categorical group action
Decomposition of the problem into 2-simples
“lift”
new insights?
2-representation theory in a nutshell
2-category categories functors nat. trafos
1-category vector spaces linear maps
0-category numbers
relate relate
relate categorify
categorify
categorify forms
forms
forms categorifies
categorifies
Classical representation theory lives here
2-representation theory should live here
The ladder of categorification: in each step there is a new layer of structure which is invisible on the ladder rung below.
Goal.
Categorify the theory “representation theory” itself.
Observation.
A groupGcan be viewed as a single-object categoryG, and a module as a functor fromG
into the single-object categoryAut(V), i.e. M:G −→ Aut(V).
What one can hope for.
Problem involving a group action
G X
Problem involving a categorical group action
Decomposition of the problem into 2-simples
“lift”
new insights?
2-representation theory in a nutshell
2-moduleM i7→M(i)
category F7→M(F)
functor α7→ M(α)
nat. trafo
1-moduleM i7→ M(i)
vector space F7→M(F)
linear map
0-modulem i7→m(i)
number
categorifies
categorifies
categorifies
categorifies
categorifies
categorical module
The ladder of categorification: in each step there is a new layer of structure which is invisible on the ladder rung below.
Goal.
Categorify the theory “representation theory” itself.
Observation.
A groupGcan be viewed as a single-object categoryG, and a module as a functor fromG
into the single-object categoryAut(V), i.e. M:G −→ Aut(V).
What one can hope for.
Problem involving a group action
G X
Problem involving a categorical group action
Decomposition of the problem into 2-simples
“lift”
new insights?
2-representation theory in a nutshell
2-moduleM i7→M(i)
category F7→M(F)
functor α7→ M(α)
nat. trafo
1-moduleM i7→ M(i)
vector space F7→M(F)
linear map
0-modulem i7→m(i)
number
categorical module
categorifies
categorifies
categorifies
categorifies
categorifies
The ladder of categorification: in each step there is a new layer of structure which is invisible on the ladder rung below.
Goal.
Categorify the theory “representation theory” itself.
Observation.
A groupGcan be viewed as a single-object categoryG, and a module as a functor fromG
into the single-object categoryAut(V), i.e. M:G −→ Aut(V).
What one can hope for.
Problem involving a group action
G X
Problem involving a categorical group action
Decomposition of the problem into 2-simples
“lift”
new insights?
The next ladder rung
Slogan. 2-representation theory is group theory in categories.
W=Chx, δ|δx= 1 +xδi
W→ End(C[X]) x7→ ·X δ7→d/dX
Step 1.
Replace the vector spacesC{Xi}by appropriate categoriesNi-Mod.
HereNi are certain algebras(“Nil Coxeter”)which embed into each otherNi ,→Ni+1, of which we think about as liftingC{Xi}−→·X C{Xi+1}.
Step 2.
Replace the linear operators·X:C{Xi} →C{Xi+1}by appropriate(“induction”)functorsIndi+1i :Ni-Mod→Ni+1-Mod.
Step 3.
Replace the linear operatorsd/dX:C{Xi+1} →C{Xi}by appropriate(“restriction”)functorsResii+1:Ni-Mod→Ni+1-Mod.
Step 4.
Check that everything works.
In particular, the reciprocityResii+1Indi+1i ∼=Id⊕ Indi−1i Resii−1 categorifies Leibniz’ product rule.
The next ladder rung
Slogan. 2-representation theory is group theory in categories.
W=Chx, δ|δx= 1 +xδi
W→ End(L
i∈N0C{Xi}) x7→L
i∈N0·X δ7→L
i∈N0d/dX
Step 1.
Replace the vector spacesC{Xi}by appropriate categoriesNi-Mod.
HereNi are certain algebras(“Nil Coxeter”)which embed into each otherNi ,→Ni+1, of which we think about as liftingC{Xi}−→·X C{Xi+1}.
Step 2.
Replace the linear operators·X:C{Xi} →C{Xi+1}by appropriate(“induction”)functorsIndi+1i :Ni-Mod→Ni+1-Mod.
Step 3.
Replace the linear operatorsd/dX:C{Xi+1} →C{Xi}by appropriate(“restriction”)functorsResii+1:Ni-Mod→Ni+1-Mod.
Step 4.
Check that everything works.
In particular, the reciprocityResii+1Indi+1i ∼=Id⊕ Indi−1i Resii−1 categorifies Leibniz’ product rule.
The next ladder rung
Slogan. 2-representation theory is group theory in categories.
W=Chx, δ|δx= 1 +xδi
x7→L
i∈N0·X δ7→L
i∈N0d/dX W→End(L
i∈N0Ni-Mod)
Step 1.
Replace the vector spacesC{Xi}by appropriate categoriesNi-Mod.
HereNi are certain algebras(“Nil Coxeter”)which embed into each otherNi ,→Ni+1, of which we think about as liftingC{Xi}−→·X C{Xi+1}.
Step 2.
Replace the linear operators·X:C{Xi} →C{Xi+1}by appropriate(“induction”)functorsIndi+1i :Ni-Mod→Ni+1-Mod.
Step 3.
Replace the linear operatorsd/dX:C{Xi+1} →C{Xi}by appropriate(“restriction”)functorsResii+1:Ni-Mod→Ni+1-Mod.
Step 4.
Check that everything works.
In particular, the reciprocityResii+1Indi+1i ∼=Id⊕ Indi−1i Resii−1 categorifies Leibniz’ product rule.
The next ladder rung
Slogan. 2-representation theory is group theory in categories.
W=Chx, δ|δx= 1 +xδi
δ7→L
i∈N0d/dX W→End(L
i∈N0Ni-Mod) x7→L
i∈N0Indi+1i
Step 1.
Replace the vector spacesC{Xi}by appropriate categoriesNi-Mod.
HereNi are certain algebras(“Nil Coxeter”)which embed into each otherNi ,→Ni+1, of which we think about as liftingC{Xi}−→·X C{Xi+1}.
Step 2.
Replace the linear operators·X:C{Xi} →C{Xi+1}by appropriate(“induction”)functorsIndi+1i :Ni-Mod→Ni+1-Mod.
Step 3.
Replace the linear operatorsd/dX:C{Xi+1} →C{Xi}by appropriate(“restriction”)functorsResii+1:Ni-Mod→Ni+1-Mod.
Step 4.
Check that everything works.
In particular, the reciprocityResii+1Indi+1i ∼=Id⊕ Indi−1i Resii−1 categorifies Leibniz’ product rule.
The next ladder rung
Slogan. 2-representation theory is group theory in categories.
W=Chx, δ|δx= 1 +xδi
W→End(L
i∈N0Ni-Mod) x7→L
i∈N0Indi+1i δ7→L
i∈N0Resii+1
Step 1.
Replace the vector spacesC{Xi}by appropriate categoriesNi-Mod.
HereNi are certain algebras(“Nil Coxeter”)which embed into each otherNi ,→Ni+1, of which we think about as liftingC{Xi}−→·X C{Xi+1}.
Step 2.
Replace the linear operators·X:C{Xi} →C{Xi+1}by appropriate(“induction”)functorsIndi+1i :Ni-Mod→Ni+1-Mod.
Step 3.
Replace the linear operatorsd/dX:C{Xi+1} →C{Xi}by appropriate(“restriction”)functorsResii+1:Ni-Mod→Ni+1-Mod.
Step 4.
Check that everything works.
In particular, the reciprocityResii+1Indi+1i ∼=Id⊕ Indi−1i Resii−1 categorifies Leibniz’ product rule.
The next ladder rung
Slogan. 2-representation theory is group theory in categories.
W=Chx, δ|δx= 1 +xδi
W→End(L
i∈N0Ni-Mod) x7→L
i∈N0Indi+1i δ7→L
i∈N0Resii+1
Step 1.
Replace the vector spacesC{Xi}by appropriate categoriesNi-Mod.
HereNi are certain algebras(“Nil Coxeter”)which embed into each otherNi ,→Ni+1, of which we think about as liftingC{Xi}−→·X C{Xi+1}.
Step 2.
Replace the linear operators·X:C{Xi} →C{Xi+1}by appropriate(“induction”)functorsIndi+1i :Ni-Mod→Ni+1-Mod.
Step 3.
Replace the linear operatorsd/dX:C{Xi+1} →C{Xi}by appropriate(“restriction”)functorsResii+1:Ni-Mod→Ni+1-Mod.
Step 4.
Check that everything works.
In particular, the reciprocityResii+1Indi+1i ∼=Id⊕ Indi−1i Resii−1 categorifies Leibniz’ product rule.
Pioneers of 2-representation theory
LetGbe a finite group.
Chuang–Rouquier & many others ∼2004++. Higher representation theory is the useful? study of (certain) categorical actions, e.g.
M:G−→Aut(V),
withV being someC-linear category. (Called 2-modules or 2-representations.) The “atoms” of such an action are called 2-simple.
Mazorchuk–Miemietz∼2014. All(suitable)2-modules are built out of 2-simples (“weak2-Jordan–H¨older filtration”).
“M(g) = a functor inAut(V)”
Plus some coherence conditions which I will not explain.
The three goals of2-representation theory. Improve the theory itself.
Discuss examples. Find applications.
Pioneers of 2-representation theory
LetC be a finitary 2-category.
Chuang–Rouquier & many others ∼2004++. Higher representation theory is
the useful? study of actions of 2-categories:
M:C −→End(V),
withV being someC-linear category. (Called 2-modules or 2-representations.) The “atoms” of such an action are called 2-simple.
Mazorchuk–Miemietz∼2014. All(suitable)2-modules are built out of 2-simples (“weak2-Jordan–H¨older filtration”).
“M(g) = a functor inAut(V)”
Plus some coherence conditions which I will not explain.
The three goals of2-representation theory.
Improve the theory itself.
Discuss examples.
Find applications.
“Lifting” classical representation theory
B AllG-modules are built out of simples.
B The character of a simpleG-module is an invariant.
B There is an injection
{simpleG-modules}/iso ,→
{conjugacy classes inG}, which is 1 : 1 in the semisimple case.
B All simples can be constructed intrinsically using the regularG-module.
Note that we have a very particular notion what a “suitable” 2-module is. What characters were for Frobenius
are these matrices for us.
There are some technicalities.
These turned out to be very interesting,
since their importance is only visible via categorification.
“Lifting” classical representation theory
B All(suitable)2-modules are built out of 2-simples.
B The character of a simpleG-module is an invariant.
B There is an injection
{simpleG-modules}/iso ,→
{conjugacy classes inG}, which is 1 : 1 in the semisimple case.
B All simples can be constructed intrinsically using the regularG-module.
Goal 1. Improve the theory itself.
Note that we have a very particular notion what a “suitable” 2-module is.
What characters were for Frobenius are these matrices for us.
There are some technicalities.
These turned out to be very interesting,
since their importance is only visible via categorification.
“Lifting” classical representation theory
B All(suitable)2-modules are built out of 2-simples.
B The decategorified actions (a.k.a. matrices) of the M(F)’s are invariants.
B There is an injection
{simpleG-modules}/iso ,→
{conjugacy classes inG}, which is 1 : 1 in the semisimple case.
B All simples can be constructed intrinsically using the regularG-module.
Note that we have a very particular notion what a “suitable” 2-module is.
What characters were for Frobenius are these matrices for us.
There are some technicalities.
These turned out to be very interesting,
since their importance is only visible via categorification.
“Lifting” classical representation theory
B All(suitable)2-modules are built out of 2-simples.
B The decategorified actions (a.k.a. matrices) of the M(F)’s are invariants.
B There is an injection
{2-simples ofC}/equi.
,→
{certain (co)algebra 1-morphisms}/“2-Morita equi.”, which is 1 : 1 in well-behaved cases.
B All simples can be constructed intrinsically using the regularG-module.
Goal 1. Improve the theory itself.
Note that we have a very particular notion what a “suitable” 2-module is. What characters were for Frobenius
are these matrices for us.
There are some technicalities.
These turned out to be very interesting,
since their importance is only visible via categorification.
“Lifting” classical representation theory
B All(suitable)2-modules are built out of 2-simples.
B The decategorified actions (a.k.a. matrices) of the M(F)’s are invariants.
B There is an injection
{2-simples ofC}/equi.
,→
{certain (co)algebra 1-morphisms}/“2-Morita equi.”, which is 1 : 1 in well-behaved cases.
B There exists principal 2-modules lifting the regular module.
Even in well-behaved cases there are 2-simples which do not arise in this way.
Note that we have a very particular notion what a “suitable” 2-module is. What characters were for Frobenius
are these matrices for us.
There are some technicalities.
These turned out to be very interesting,
since their importance is only visible via categorification.
2-modules of dihedral groups
Consider : θs=s+ 1, θt=t+ 1.
(Motivation. The Kazhdan–Lusztig basis has some neat integral properties.)
These elements generateC[D2n] and their relations are fully understood:
θsθs= 2θs, θtθt= 2θt, a relation for . . .| {z }sts
n
=. . .| {z }tst
n
.
We want a categorical action. So we need:
B A categoryV to act on.
B EndofunctorsΘsandΘtacting onV.
B The relations ofθsandθthave to be satisfied by the functors.
B A coherent choice of natural transformations. (Skipped today.)
Some details.
Theorem∼2016.
There is a one-to-one correspondence {(non-trivial)2-simpleD2n-modules}/2-iso
←→1:1
{bicolored ADE Dynkin diagrams with Coxeter numbern}. Thus, its easy to write down a list .
Goal 2. Discuss examples.
2-modules of dihedral groups
Consider : θs=s+ 1, θt=t+ 1.
(Motivation. The Kazhdan–Lusztig basis has some neat integral properties.)
These elements generateC[D2n] and their relations are fully understood:
θsθs= 2θs, θtθt= 2θt, a relation for . . .| {z }sts
n
=. . .| {z }tst
n
.
We want a categorical action. So we need:
B A categoryV to act on.
B EndofunctorsΘsandΘtacting onV.
B The relations ofθsandθthave to be satisfied by the functors.
B A coherent choice of natural transformations. (Skipped today.)
Some details.
Theorem∼2016.
There is a one-to-one correspondence {(non-trivial)2-simpleD2n-modules}/2-iso
←→1:1
{bicolored ADE Dynkin diagrams with Coxeter numbern}. Thus, its easy to write down a list .
Goal 2. Discuss examples.