TENSOR PRODUCT IN CATEGORY O κ.

Similar documents
New tools of set-theoretic homological algebra and their applications to modules

Gödel algebras free over finite distributive lattices

DAY I, TALK 5: GLOBAL QUANTUM THEORY (GL-3)

Laurence Boxer and Ismet KARACA

Rational points of Abelian varieties in \Gamma-extension

DENSITY OF PERIODIC GEODESICS IN THE UNIT TANGENT BUNDLE OF A COMPACT HYPERBOLIC SURFACE

CARDINALITIES OF RESIDUE FIELDS OF NOETHERIAN INTEGRAL DOMAINS

Covering properties of derived models

CATEGORIES OF MODULES OVER AN AFFINE KAC MOODY ALGEBRA AND FINITENESS OF THE KAZHDAN LUSZTIG TENSOR PRODUCT

Fair semigroups. Valdis Laan. University of Tartu, Estonia. (Joint research with László Márki) 1/19

THE NUMBER OF UNARY CLONES CONTAINING THE PERMUTATIONS ON AN INFINITE SET

Laurence Boxer and Ismet KARACA

EISENSTEIN SERIES AND QUANTUM GROUPS

PURITY IN IDEAL LATTICES. Abstract.

Notes on the symmetric group

ON THE QUOTIENT SHAPES OF VECTORIAL SPACES. Nikica Uglešić

Algebra homework 8 Homomorphisms, isomorphisms

Characterizing large cardinals in terms of layered partial orders

4: SINGLE-PERIOD MARKET MODELS

SEMICENTRAL IDEMPOTENTS IN A RING

Chain conditions, layered partial orders and weak compactness

REMARKS ON K3 SURFACES WITH NON-SYMPLECTIC AUTOMORPHISMS OF ORDER 7

COMBINATORICS OF REDUCTIONS BETWEEN EQUIVALENCE RELATIONS

Building Infinite Processes from Regular Conditional Probability Distributions

Orthogonality to the value group is the same as generic stability in C-minimal expansions of ACVF

On axiomatisablity questions about monoid acts

Some Remarks on Finitely Quasi-injective Modules

Transcendental lattices of complex algebraic surfaces

The (λ, κ)-fn and the order theory of bases in boolean algebras

Generalising the weak compactness of ω

The illustrated zoo of order-preserving functions

GUESSING MODELS IMPLY THE SINGULAR CARDINAL HYPOTHESIS arxiv: v1 [math.lo] 25 Mar 2019

CATEGORICAL SKEW LATTICES

Non replication of options

The finite lattice representation problem and intervals in subgroup lattices of finite groups

Introduction to Priestley duality 1 / 24

The reverse self-dual serial cost-sharing rule M. Josune Albizuri, Henar Díez and Amaia de Sarachu. April 17, 2012

Chapter 4. Cardinal Arithmetic.

Fractional Graphs. Figure 1

Equivalence between Semimartingales and Itô Processes

A Property Equivalent to n-permutability for Infinite Groups

The Semi-Weak Square Principle

Ideals and involutive filters in residuated lattices

AN INFINITE CARDINAL-VALUED KRULL DIMENSION FOR RINGS

DEPTH OF BOOLEAN ALGEBRAS SHIMON GARTI AND SAHARON SHELAH

Translates of (Anti) Fuzzy Submodules

KAPLANSKY'S PROBLEM ON VALUATION RINGS

A relative of the approachability ideal, diamond and non-saturation

CHOICE THEORY, UTILITY FUNCTIONS AND RISK AVERSION

Continuous images of closed sets in generalized Baire spaces ESI Workshop: Forcing and Large Cardinals

arxiv: v1 [math.lo] 24 Feb 2014

arxiv:math/ v1 [math.lo] 15 Jan 1991

This is an author version of the contribution published in Topology and its Applications

INSURANCE VALUATION: A COMPUTABLE MULTI-PERIOD COST-OF-CAPITAL APPROACH

Sy D. Friedman. August 28, 2001

PSEUDO SEMI B-FREDHOLM AND GENERALIZED DRAZIN INVERTIBLE OPERATORS THROUGH LOCALIZED SVEP

VAN KAMPEN COLIMITS AS BICOLIMITS IN SPAN. Tobias Heindel and Paweł Sobociński CALCO 10/09/09 Udine

THE IRREDUCIBILITY OF CERTAIN PURE-CYCLE HURWITZ SPACES

ADDING A LOT OF COHEN REALS BY ADDING A FEW II. 1. Introduction

Log-linear Dynamics and Local Potential

Variations on a theme by Weetman

INTERIM CORRELATED RATIONALIZABILITY IN INFINITE GAMES

GPD-POT and GEV block maxima

On the h-vector of a Lattice Path Matroid

CONGRUENCES AND IDEALS IN A DISTRIBUTIVE LATTICE WITH RESPECT TO A DERIVATION

Epimorphisms and Ideals of Distributive Nearlattices

Interpolation of κ-compactness and PCF

Double Ore Extensions versus Iterated Ore Extensions

UPWARD STABILITY TRANSFER FOR TAME ABSTRACT ELEMENTARY CLASSES

Ratio Mathematica 20, Gamma Modules. R. Ameri, R. Sadeghi. Department of Mathematics, Faculty of Basic Science

being saturated Lemma 0.2 Suppose V = L[E]. Every Woodin cardinal is Woodin with.

Filters - Part II. Quotient Lattices Modulo Filters and Direct Product of Two Lattices

Polynomial processes in stochastic portofolio theory

Lie Algebras and Representation Theory Homework 7

Skew lattices of matrices in rings

Semantics with Applications 2b. Structural Operational Semantics

MAT 4250: Lecture 1 Eric Chung

MITCHELL S THEOREM REVISITED. Contents

Introduction Random Walk One-Period Option Pricing Binomial Option Pricing Nice Math. Binomial Models. Christopher Ting.

3 The Model Existence Theorem

In this lecture, we will use the semantics of our simple language of arithmetic expressions,

The mod 2 Adams Spectral Sequence for tmf

THE GENUS FIELD AND GENUS NUMBER IN ALGEBRAIC NUMBER FIELDS

A precipitous club guessing ideal on ω 1

Optimal Allocation of Policy Limits and Deductibles

Square-Root Measurement for Ternary Coherent State Signal

PERFECT TREE FORCINGS FOR SINGULAR CARDINALS

arxiv:physics/ v2 [math-ph] 13 Jan 1997

PARTITIONS OF 2 ω AND COMPLETELY ULTRAMETRIZABLE SPACES

Forecast Horizons for Production Planning with Stochastic Demand

ORDERED SEMIGROUPS HAVING THE P -PROPERTY. Niovi Kehayopulu, Michael Tsingelis

LIBOR models, multi-curve extensions, and the pricing of callable structured derivatives

Fuzzy L-Quotient Ideals

SHIMON GARTI AND SAHARON SHELAH

Congruence lattices of finite intransitive group acts

A CATEGORICAL FOUNDATION FOR STRUCTURED REVERSIBLE FLOWCHART LANGUAGES: SOUNDNESS AND ADEQUACY

RUDIN-KEISLER POSETS OF COMPLETE BOOLEAN ALGEBRAS

Large Cardinals with Few Measures

Attempt QUESTIONS 1 and 2, and THREE other questions. Do not turn over until you are told to do so by the Invigilator.

5.1 Gauss Remarkable Theorem

Transcription:

TENSOR PRODUCT IN CATEGORY O κ. GIORGIA FORTUNA Let V 1,..., V n be ĝ κ -modules. Today we will construct a new object V 1 V n in O κ that plays the role of the usual tensor product. Unfortunately in fact V 1 V n is not in O κ (has already the wrong central charge nκ and moreover does not have reasonable finiteness properties. Before starting we will recall the main objects we are going to use. 1. Recap of definitions If S = {x 0,..., x n } is a finite set, we denote by ĝ κ S the Lie algebra fitting into the exact sequence 0 C ĝ κ S n i=0g((t i 0, corresponding to the quotient of ĝ κ ĝ }{{ κ by Ker(C } n+1 C. Note (n+1 times that for any collection of ĝ κ -modules V 1,..., V n, the tensor product V 1 V n carries a natural ĝ S κ -action (with central charge κ coming from the natural action of (ĝ κ S on it (in other words this action factors through the kernel of the map from C n+1 to C. Let now C be a connected projective curve isomorphic to P 1, and take a collection of n + 1 points S = {x 0,..., x n } on it. Assume that for every point x i in S, we are given charts γ i : P 1 C, such that γ i (0 = x i. If Γ(C {x 0,..., x n } is the algebra of regular functions on C {x 0,..., x n }, for any f Γ(C {x 0,..., x n } we denote by x i f the expansion at 0 of the rational function f γ i. From now on we will however focus on the case x 0 =, i.e. S = {, x 1,..., x n }. We will denote by ĝ κ,out the Lie algebra ĝ κ,out := Γ(P 1 {, x 1,..., x n } g. Remark 1. For f g ĝ κ,out, g g, the series Taylor expansion at each point in S gives a well defined map ĝ κ,out n i=0g((t i, Date: October 4, 2011. 1

2 GIORGIA FORTUNA moreover, because of the residue theorem, the above map lifts to a maps 0 C ĝ κ S n i=0 g((t i 0 ĝ κ,out hence the ĝ κ S -module structure on V 1 V n we introduced before, naturally gives a ĝ κ,out -action on this tensor product. 2. Existence of the tensor product Let S be the collection of points {, x 1,..., x n } and let V 1,..., V n be ĝ κ -modules. We will now construct the tensor product V 1 V n of these modules. One main property that we would like V 1 V n to satisfy is the following: < X, V 1 V n >= (X V 1 V n ĝκ,out, where the action of ĝ κ,out on the tensor product is the one coming from the ĝ κ S -action that we discussed before. By dualizing and rephrasing the expression above (using the fact that Hom Oκ (V, D(V =< V, V > this is equivalent to requiring Hom Oκ (X, D(V 1 V n = ((X V 1 V n ĝκ,out, hence to find such an object, it is enough to show that the functor is representable. F V1,...,V n : O op κ Vect X ((X V 1 V n ĝκ,out, Vect defined above is repre- Proposition 1. The functor F V1,...,V n : O op κ sentable by an element O κ. Proof. First of all we will show that F is left exact. Lemma 1. The functor defined above preserves finite limits (i.e. exact. is left Proof. First of all since finite limits in O op κ are finite colimits in O κ what we want to show is that if X = lim X I i, for a finite I, then ((lim X I i V 1 V n ĝκ,out = lim((x I i V 1 V n ĝκ,out. However, since right (resp. left exact functors commutes with finite colimits (resp. limits, and since tensor product and taking invariants are right exact,

we immediately get the equality: ( (lim I (lim I TENSOR PRODUCT IN CATEGORY O κ. 3 X i V 1 V n = lim(x I i V 1 V n (X i V 1 V n ĝκ,out = lim ((X I i V 1 V n ĝκ,out ( = lim (Xi V 1 V n ĝκ,out. I lim ((X i V 1 V n ĝκ,out I By a general theorem in category theory (see [B] 6.1.2 and by Proposition 1, we know that F V1,...,V n is the inductive limit of representable functors, in other words it has the form: (1 F V1,...,V n ( = lim Hom I Oκ (, i, where i O κ. Moreover, since we are working in category O κ, we can rewrite (1 as In fact the following is true: Hom KLκ (, lim I i. Lemma 2. The category of KL κ -modules is the ind-completion of the category O κ. In other words (1 Objects in O κ are compact in KL κ. (2 Every object of KL κ is a filtred colimit of objects of O κ. Proof. Recall that we defined the category of KL κ -modules as the category of those ĝ κ -modules V such that: (1 V is smooth. (2 The action of tg[[t]] is locally nilpotent. (3 The action of g is locally finite. This is an abelian category and contains the abelian category O κ consisting of those modules V that are finitely generated. For part (1, if X is in O κ and lim A I i is a filtered colimit of objects A i KL κ, then the natural map lim Hom Oκ (X, A i Homĝκ mod(x, lim A I i I is an isomorphism. This follows from a general fact in category theory that asserts that a module is finitely presented if and only if Hom(X, commutes with filtered colimits. Since objects in O κ are finitely generated, and the category is noetherian, in particular they are finitely presented. Now if V

4 GIORGIA FORTUNA is a KL κ module, to prove part (2, we only have to show that V can be written as V = i V i, V i O κ. However every element v V we can simply consider the submodule W generated by this vector. By definition W is finitely generated, hence is in O κ. From the previous Lemma we know that F V1,...,V n is hence represented by a KL κ -module := lim I i. Now we want to show that is in O κ. To do this it is sufficient to show that Since (1 = ( tg[[t]] is finite dimensional. (1 = λ Λ + Homĝκ mod(v κ λ, M V1,...,Vn, (where Vλ κ denotes the Weyl module it is enough to show that only for finitely many λ s, Homĝκ mod(vλ κ, is non zero and moreover is finite dimensional. However, by definition we have Homĝκ mod(vλ κ, = ((Vλ κ V 1 V n ĝκ,out. Hence we want to show that ((Vλ κ V 1 V n ĝκ,out is finite dimensional. Because of the right exactness of tensor product and coinvariants, it λ Λ + is enough to show it in the case V i = Ni κ, where N i κ = Indĝκ g[[t]]+c N i, with dim(n i < ( every v O κ admits a surjection from such a module. In this case we have: (2 ((V κ λ N κ 2 N κ n ĝκ,out ((V λ N 2 N n g, and the last is by definition Hom g (V λ N 2 N n which is finite dimensional. Moreover, since only finitely many λ s can appear in the composition series of (N 2 N n we have that the sum over all dominant λ s is indeed a finite sum and we are done. Note that in the proof above we have used the following fact (proved in Sasha s notes: Proposition 2. For N κ i = Indĝκ g[[t]]+c N i, i = 1,..., n we have (N κ 1 N κ 2 N κ n ĝκ,out (N 1 N 2 N n g.

TENSOR PRODUCT IN CATEGORY O κ. 5 3. explicit construction of the tensor product Unfortunately what we have constructed is not quite the object V 1 V n that we wanted, but it is its dual. Moreover we don t have an explicit description of that allows us to take the dual. Hence we will now recover from the representability property. Proposition 3. The tensor product V 1 V n is equal to ( V 1 V n := D( = lim m (V 1 V n # G(m where G(m is the right ideal generated by those elements in U(ĝ κ,out of the form (f 1 c 1 (f m c m, with f i functions with zeros of order at least 1 Proof. To prove the proposition, we will first describe the module using the functor it represents. It would be nice to say that = Hom Oκ (U(ĝ κ, but unfortunately U(ĝ κ is not in our category. However we can take the pro-object lim U(ĝ κ /U(ĝ κ (tg i (tg m + U(ĝ κ J, where J runs over all left ideals in U(g of finite codimension. int, In fact Û(ĝ κ = lim U(ĝ κ and maps from Û(ĝ κ to a module V KL κ are by definition the same as maps from U(ĝ κ to it. Moreover if we denote by U(ĝ κ the term U(ĝ κ := U(ĝ κ /U(ĝ κ (tg i (tg m + U(ĝ κ J, then U(ĝ κ is in our category. Now we can see how looks like: = Homĝκ mod( lim U(ĝ κ, = lim Hom Oκ (U(ĝ κ, = lim((u(ĝ κ V 1 V n ĝκ,out = lim((v 1 V n G(m+J, where G(m is the right ideal generated by those elements in U(ĝ κ,out of the form (f 1 c 1 (f m c m, with f i functions with poles of order at least 1, and where J runs over the right ideals generated by the ideals of finite codimension in U(g. As we said before the module corresponds to the dual of the sought for V 1 V n. Hence to recover the definition of V 1 V n we need to compute D(. Recall that D(V := (V int. When we take

6 GIORGIA FORTUNA the dual we have: ( = Hom(lim((V 1 V n G(m+J, C = = lim((v 1 V n G(m+J. Lemma 3. The vector space (V 1 V n G(m+J is finite dimensional. Proof. This follows from the fact that is in O κ, hence in particular (m is finite dimensional. In fact we have the following injection into (m (m Hom(U(ĝ κ, = ((V 2 V n G(m+J hence ((V 1 V n G(m+J is finite dimensional too, which implies the lemma. Because of Lemma 3 we have ( = lim((v 1 V n G(m+J lim(v 1 V n G(m+J. Note that elements in U(g stabilize the ideal G(m, hence we can consider (V 1 V n G(m as a g-module and compute coinvariants in two steps: lim m+j (V 1 V n G(m+J = lim m lim J ((V 1 V n G(m J. The last term is equal to lim m (V 1 V n G(m since by the proposition below, also (V 1 V n G(m happens to be finite dimensional. Hence we have ( = lim m (V 1 V n G(m. applying and taking integrable vectors, now gives us ( V 1 V n = D( = lim(v 1 V n # G(m m as we wanted. Note that this definition coincides with the one that is given in [KL] Section 4. Proposition 4. For every V 2,..., V n in O κ the vector space is finite dimensional. (V 1 V n G(m Proof. Since the object (V 1 V n G(m is acted on by g integrably, it s enough to show that for every λ, the space of coinvariants ((V 1 V n G(m V λ g int,

TENSOR PRODUCT IN CATEGORY O κ. 7 are finite-dimensional, and non-zero for only finitely many λ s. In fact this would imply that ((V 1 V n G(m Hom g ((V 1 V n G(m, V λ Vλ λ is finite dimensional. Let J λ U(g be such that U(g/J λ = V λ. Consider the corresponding object U(ĝ κ λ. Then However ((V 1 V n G(m V λ g (U(ĝ κ λ V 1 V n ĝκ,out. ((U(ĝ κ λ V 1 V n ĝκ,out = Hom Oκ (U(ĝ κ λ, = = Hom(V λ, M(m, which is finite-dimensional and non-zero for only finitely many λ s since is in O κ (hence (m is finite dimensional. References [B] F. Borceux, Handbook of Categorical Algebra: Basic category theory. [KL] D. Kazhdan, G. Lusztig, Tensor Structures Arising from Affine Lie Algebras. I, JAMS, Vol 6, Num. 4, October 1993.