SLIDE 1
Connected gradings and fundamental groups
Mar´ ıa Julia Redondo
Joint work with Claude Cibils and Andrea Solotar
ICTP, Trieste - February 1-5, 2010
SLIDE 2 The intrinsic fundamental group of a linear category,
- C. Cibils, M. J. Redondo, A. Solotar,
arXiv:0706.2491 π1(A) = automorphism group of the fibre functor Main goal: compute explicitely π1(A) for several algebras A Main tool: the relation between gradings and Galois coverings of the algebra (considered as a k-linear category with one object).
SLIDE 3
Galois coverings A k-category B is a small category such that
◮ each set of morphisms yBx is a k-module, and ◮ composition of morphisms is k-bilinear.
Each xBx is a k-algebra and yBx is a yBy-xBx-bimodule. A k-algebra A can be viewed as a k-category with one object.
SLIDE 4 Let C and B be k-categories. A k-functor F : C → B is a covering if it is surjective on objects and if for each b ∈ B0 and each x in F −1(b), the map F x
b : StxC → StbB,
provided by F, is a k-isomorphism, where the star StbB at an
- bject b of a k-category B is the direct sum of all k-modules of
morphisms with source or target b.
SLIDE 5 A morphism from a covering F : C → B to a covering G : D → B is a pair of k-linear functors (H, J) where C
H
G
J ∼ =
B
conmutes, with J an isomorphism. The morphism (H, J) induces a unique group epimorphism λH : Aut1 F → Aut1 G verifying Hf = λH(f )H, for all f ∈ Aut1 F. C
f
H
G
G
B
J ∼ =
B
B
SLIDE 6
Aut1F = {(H, id) : H invertible} Let b ∈ B0 and let F −1(b) be the corresponding fibre.
◮ F −1(b) = ∅ by definition of covering. ◮ Aut1F acts freely on F −1(b).
Definition
A covering F : C → B of k-categories is a Galois covering if C is connected and if Aut1F acts transitively on some fibre. One can prove that for a Galois covering F, the group Aut1F acts transitively on any fibre.
SLIDE 7
The fundamental group Given a k-category B and a fixed object b0 of B, Gal(B, b0) denotes the category of Galois coverings of B with morphisms (H, J) such that J(b) = b, for any b in B0. Consider Φ : Gal(B, b0) → Sets given by Φ(F) = F −1(b0)
Definition
π1(B, b0) = Aut Φ
SLIDE 8 A universal covering U : U → B is an object in Gal(B) such that for any Galois covering F : C → B, and for any u0 ∈ U0, c0 ∈ C0 with U(u0) = F(c0), there exists a unique morphism (H, id) from U to F verifying H(u0) = c0. U
H
F
B U(u0) F(c0)
Theorem
If a connected k-category B admits a universal covering U then π1(B, b0) ≃ Aut1U.
SLIDE 9 Examples of Galois coverings A grading X of a k-category B by a group Γ is given by a direct sum decomposition of each k-module of morphisms
yBx =
X s (yBx) such that X t (zBy) X s (yBx) ⊂ X ts (zBx). The homogeneous component of degree s from x to y is the k-module X s (yBx). A grading is said to be connected if given any two objects in B, and any element g ∈ Γ, they can be joined by a non-zero homogeneous walk of degree g.
SLIDE 10
Let X be a Γ-grading of the k-category B. The smash product category B#Γ is given by:
◮ (B#Γ)0 = B0 × Γ, ◮ the module of morphisms from (b, g) to (c, h) is X h−1g cBb.
Morphisms are provided by homogeneous components, and composition in B#Γ is given by the original composition in B.
SLIDE 11
The smash product construction provides examples of Galois coverings. If X is a Γ-grading of the k-category B, the functor B#Γ
B
(b, g)
b
(c,h)B#Γ(b,g) = X h−1g cBb
cBb
is a Galois covering with Γ as group of automorphisms.
SLIDE 12
The action of Γ on the smash product category B#Γ is given as follows. The action on objects is given by the left action of Γ on itself: s(b, g) = (b, sg). A morphism (b, g) → (c, h) is a homogeneous morphism from b to c of degree h−1g. So it is also a morphism from (b, sg) to (c, sh) since (sh)−1sg = h−1g.
SLIDE 13
Let F : B#Γ1 → B and G : B#Γ2 → B be Galois coverings associated to connected gradings X1 and X2 of B with groups Γ1 and Γ2, and let b0 ∈ B. Let (H, J) : F → G be a morphism of coverings in Gal(B, b0). Then there exists a map h : Γ1 → Γ2 such that H(b0, g) = (b0, h(g)) for all g ∈ Γ1. Moreover, h(g) = λH(g)h(1), where λH : Γ1 → Γ2 is the group morphism associated to H. Given σ ∈ Aut Φ and F : B#Γ → B, then the corresponding map σF : Γ → Γ is given by σF(g) = gσF(1).
SLIDE 14 Connection between gradings and coverings Let Gal#(B, b0) be the full subcategory of Gal(B, b0) whose
- bjects are the smash product Galois coverings F : B#Γ → B.
Theorem (Cibils, Marcos - 2006)
The categories Gal#(B, b0) and Gal(B, b0) are equivalent. The proof follows from the fact that any Galois covering F : C → B is isomorphic to the Galois covering B#Aut1F → B.
Corollary
Let Φ# : Gal#(B, b0) → Sets be the functor given by Φ#(F : B#Γ → B) = F −1(b0) = Γ. Then π1(B, b0) ∼ = Aut Φ#.
SLIDE 15
The matrix algebra Let k be a field containing a primitive n-th root of unity q. The matrix algebra Mn(k) has a well-known presentation as follows: Mn(k) = k{x, y}/xn = 1, yn = 1, yx = qxy where x = · · · 1 1 · · · 1 · · · . . . ... . . . · · · 1 , y = q · · · q2 · · · q3 · · · . . . ... . . . · · · qn .
SLIDE 16 If we set deg(x) = (t, 1) and deg(y) = (1, t), for t a generator of Cn, we obtain a connected grading of k{x, y} such that the ideal
- f relations is homogeneous.
Theorem
The algebra Mn(k) admits a simply connected grading by the group Cn × Cn.
SLIDE 17 Another presentation of Mn(k) is given by the following quiver with relations (Q, I). Q : 1
α1 2 β1
β2
n − 1
αn−1
n
βn−1
- I =< βiαi − ei, αiβi − ei+1|1 ≤ i < n > .
Then kQ/I is isomorphic to Mn(k).
SLIDE 18
Let Fn−1 be the free group on n − 1 generators s1, . . . , sn−1, and let
◮ deg ei = 1 for any i with 1 ≤ i ≤ n, and ◮ deg αi = si and deg βi = (si)−1 for any i with 1 ≤ i ≤ n − 1.
This provides a well defined grading of kQ/I, hence of Mn(k). Moreover, it is also simply connected.
Corollary
The matrix algebra Mn(k) has no universal covering.
SLIDE 19 Definition
The quotient of a Γ-grading X of a category B by a normal subgroup N of Γ is a Γ/N-grading X/N of B, where the homogeneous component of degree α is (X/N)αcBb =
X g cBb. The corresponding functor between the smash product coverings is precisely the canonical projection obtained through the quotient of B#Γ → B by N: B#Γ
B#Γ/N
SLIDE 20 Proposition
Let k be a field containing a primitive n-th root of unity. The grading by Cn × Cn and the grading by the free group Fn−1 have a maximal common quotient Cn-grading, which is unique. Mn(k)#Cn × Cn
- Mn(k)#Fn−1
- Mn(k)#Cn
- Mn(k)
Next we use the description of gradings of matrix algebras given by several authors and we obtain:
SLIDE 21
Theorem
◮ (Boboc, D˘
asc˘ alescu and Khazal, 2003) If char(k) = 2,then π1M2(k) ≃ Z × C2.
◮ (Boboc and D˘
asc˘ alescu, 2007) If char(k) = 3,then π1M3(k) ≃ F2 × C3.
◮ (Bahturin and Zaicev, 2002)
If k is an algebraically closed field, char(k) = 0 and p a prime then π1Mp(k) ≃ Fp−1 × Cp.
SLIDE 22 Triangular matrices Using a description of gradings of triangular algebras given by Valenti and Zaicev (2007),
Theorem
Let k be a field and let Tn(k) be the algebra of triangular matrices
π1Tn(k) ≃ Fn−1.
SLIDE 23
Truncated polynomial algebra
Theorem
Let k be a field of characteristic p and let A = k[x]/(xp). There are two types of connected gradings of A, with no common quotient except the trivial one:
◮ the natural grading given by Cp since k[x]/(xp) is isomorphic
to the group algebra KCp.
◮ the grading given by Z or any of its quotients.
Corollary
Let k be a field of characteristic p. Then π1k[x]/(xp) = Z × Cp.
SLIDE 24 The diagonal algebra kn Let E be a finite set with n elements and k a field. The diagonal algebra kn is the vector space of maps from E to k with pointwise multiplication.
Proposition
Let E be a finite set with cardinality n and let k be a field with enough n-th roots of unity. Let G be any abelian group of order n. Then there is a simply connected G-grading of kn.
Corollary
Let n be a non-square free positive integer and let k be a field as
- above. The algebra kn does not admit a universal covering.
SLIDE 25 The following theorem is based on results due to D˘ asc˘ alescu.
Theorem
◮ Let k be a field of characteristic different from 2. Then
π1(k × k) = C2.
◮ Let k be a field containing all roots of unity of order 2 and 3.
Then π1(k3) = C3 × C2.
◮ Let k be a field containing all roots of unity of order 2, 3 and
- 4. Then π1(k4) = C4 × (C2 × C2) × C3 × (C2 ∗ C2).