Lusztig sheaves, decomposition rule and restriction rule

Yixin Lan · 2025

Read Lusztig sheaves, decomposition rule and restriction rule by Yixin Lan (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-14911.txt.

Introduction

Yixin Lan uses, in a 2025 solo-authored paper, Lusztig's perverse sheaves to build canonical bases for certain algebra modules. The modules arise as layered pieces of tensor products and of restricted modules for symmetric Kac-Moody algebras, objects at the heart of modern representation theory.

Realizing the modules geometrically, through sheaves on multi-framed quivers, supplies the bases directly. As an application, Lan proves that the decomposition and restriction numbers of these algebras equal the dimensions of top Borel-Moore homology groups of certain regions inside Nakajima's quiver varieties. Algebra counts become geometry measurements.

Catalog record

CREATORYixin Lan
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE170 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2504-14911

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Lusztig sheaves, decomposition rule and restriction rule Thanks: The author is supported by the National Natural Science Foundation of China [Grant No. 1288201] and [Grant No. 12471030]. Yixin Lan Abstract . In this article, we realize the subquotient based modules of certain tensor products or restricted modules via Lusztig’s perverse sheaves on multi-framed quivers, and provide a construction of their canonical bases. As an application, we prove that the decomposition and restriction coefficients of symmetric Kac-Moody algebras equal to the dimensions of top Borel-Moore homology groups for certain locally closed subsets of Nakajima’s quiver varieties. 2000 Mathematics Subject Classification Primary 17B37; Secondary 14D21, 16G20. 1. Introduction Given a symmetric Cartan matrix C=(a_{ij})_{i,j\in I} , the matrix C defines a bilinear form \langle-,-\rangle:\mathbb{Z}I\times\mathbb{Z}I\rightarrow\mathbb{Z} , and there is a Kac-Moody Lie algebra \mathfrak{g} associated to C . The quantum group \mathbf{U}(\mathfrak{g}) is the \mathbb{Q}(v) -algebra generated by E_{i},F_{i},K_{\nu} for i\in I,\nu\in\mathbb{Z}I , subject to the following relations: • (a) K_{0}=1 and K_{\nu}K_{\nu^{\prime}}=K_{\nu+\nu^{\prime}} for any \nu,\nu^{\prime}\in\mathbb{Z}I ; • (b) K_{\nu}E_{i}=v^{\langle\nu,i\rangle}E_{i}K_{\nu} for any i\in I,\nu\in\mathbb{Z}I ; • (c) K_{\nu}F_{i}=v^{-\langle\nu,i\rangle}F_{i}K_{\nu} for any i\in I,\nu\in\mathbb{Z}I ; • (d) E_{i}F_{j}-F_{j}E_{i}=\delta_{ij}\frac{K_{i}-K_{-i}}{v-v^{-1}} for any i,j\in I ; • (e) \sum_{p+q=1-a_{ij}}(-1)^{p}E_{i}^{(p)}E_{j}E_{i}^{(q)}=0 for any i\not=j ; • (f) \sum_{p+q=1-a_{ij}}(-1)^{p}F_{i}^{(p)}F_{j}F_{i}^{(q)}=0 for any i\not=j . Here E_{i}^{(n)}=E_{i}^{n}/[n]! , F_{i}^{(n)}=F_{i}^{n}/[n]! are the divided power of E_{i},F_{i} , where [n]!:=\prod_{s=1}^{n}\frac{v^{s}-v^{-s}}{v-v^{-1}} . Taking the classical limit v\rightarrow 1 , we obtain the universal enveloping algebra \mathbf{U}_{1}(\mathfrak{g}) of \mathfrak{g} from the quantum group. Given a dominant integral weights \lambda , the module M(\lambda) is realized by \begin{split}M(\lambda)=&\mathbf{U}(\mathfrak{g})/(\sum_{i\in I}\mathbf{U}(\mathfrak{g})E_{i}+\sum_{\nu\in\mathbb{Z}[I]}\mathbf{U}(\mathfrak{g})(K_{\nu}-v^{\langle\nu,\lambda\rangle})).\end{split} The Verma module M(\lambda) has a unique irreducible quotient, (1) \begin{split}L(\lambda)=&\mathbf{U}(\mathfrak{g})/(\sum_{i\in I}\mathbf{U}(\mathfrak{g})E_{i}+\sum_{\nu\in\mathbb{Z}[I]}\mathbf{U}(\mathfrak{g})(K_{\nu}-v^{\langle\nu,\lambda\rangle})+\sum_{i\in I}\mathbf{U}(\mathfrak{g})F_{i}^{\langle i,\lambda\rangle+1})\\ \cong&\mathbf{U}^{-}(\mathfrak{g})/(\sum_{i\in I}\mathbf{U}^{-}(\mathfrak{g})F_{i}^{\langle i,\lambda\rangle+1}).\end{split} Take v\rightarrow 1 , we can obtain an irreducible highest weight \mathfrak{g} -module L_{1}(\lambda) . 1.1. Decomposition rule and Restriction Rule Given a Kac-Moody algebra \mathfrak{g} or its quantized enveloping algebra \mathbf{U}(\mathfrak{g}) , one can consider the category of integrable highest weight \mathfrak{g} (or \mathbf{U}(\mathfrak{g}) )-modules. The irreducible integrable highest weight \mathfrak{g} -modules L_{1}(\lambda) (or \mathbf{U}(\mathfrak{g}) -modules L(\lambda) ) are parameterized by dominant integral weights \lambda of \mathfrak{g} . For any two dominant integral weights \lambda^{1} and \lambda^{2} , the tensor product L_{1}(\lambda^{1})\otimes L_{1}(\lambda^{2}) is also an integrable highest weight \mathfrak{g} -module, and decomposed into a direct sum of irreducible integrable highest weight \mathfrak{g} -modules (2) L_{1}(\lambda^{1})\otimes L_{1}(\lambda^{2})\cong\bigoplus_{\mu,\alpha\in S(\mu)}L_{1}(\mu)^{\alpha}, here \mu runs over dominant integral weights of \mathfrak{g} and S(\mu) is the set of direct summands of L_{1}(\lambda^{1})\otimes L_{1}(\lambda^{2}) , which is isomorphic to L_{1}(\mu) . This is called the decomposition rule (or Littlewood-Richardson rule), and the order m_{\mu}^{\lambda^{1},\lambda^{2}} of S(\mu) is called the decomposition coefficient. Similarly, we can consider a Levi subalgebra \mathfrak{l} of \mathfrak{g} , and consider the restricted module \mathbf{res}^{\mathfrak{g}}_{\mathfrak{l}}L_{1}(\lambda) . As an integrable highest weight \mathfrak{l} -module, \mathbf{res}^{\mathfrak{g}}_{\mathfrak{l}}L_{1}(\lambda) is also ecomposed into a direct sum of irreducible integrable highest weight \mathfrak{l} -modules (3) \mathbf{res}^{\mathfrak{g}}_{\mathfrak{l}}L_{1}(\lambda)=\bigoplus_{\mu_{\mathfrak{l}},\alpha\in S(\mu_{\mathfrak{l}})}L_{1}(\mu_{\mathfrak{l}})^{\alpha}, here \mu_{\mathfrak{l}} runs over dominant integral weights of \mathfrak{l} and S(\mu_{\mathfrak{l}}) is the set of direct summands of \mathbf{res}^{\mathfrak{g}}_{\mathfrak{l}}L(\lambda) , which is isomorphic to L_{1}(\mu_{\mathfrak{l}}) . This is called the restriction rule, and the order m_{\mu_{\mathfrak{l}}}^{\lambda} of S(\mu_{\mathfrak{l}}) is called the restriction coefficient. It is an important question to determine the decomposition rule and restriction rule. The author in [ 20 ] construct Demazure modules from paths and root operators, and prove that the decomposition rule and restriction rule is determined by certain paths. Later, Kashiwara proves similar Demazure character formula for his crystals in [ 15 ] . Their results give a combinatorial solution to this question. In this article, we expect to give a categorical and geometric answer to this question. 1.2. Perverse sheaves and Canonical bases for quantum groups and their representations For a given symmetric generalized Cartan matrix C , there is a finite quiver Q=(I,H,\Omega) associated to C . We assume that Q doesn’t have loops but may have multiple arrows. In [ 22 ] , [ 24 ] , [ 25 ] and [ 26 ] , G.Lusztig has considered a certain class of perverse sheaves on the moduli spaceof representations of Q . These semisimple complexes are called Lusztig’s sheaves. The Grothendieck group of Lusztig’s sheaves has a bialgebra structure, which is canonically isomorphic to the integral form {{}_{\mathcal{A}}\mathbf{U}^{+}}(\mathfrak{g}) (or {{}_{\mathcal{A}}\mathbf{U}^{-}}(\mathfrak{g}) ) of the positive (or negative) part of the quantum group. Moreover, the simple Lusztig sheaves forms the canonical basis. Under the canonical map ( 1 ), the nonzero images of elements in the canonical basis provide a basis of L(\lambda) , which is called the canonical basis of L(\lambda) by Lusztig. By a algebraic construction, Lusztig also defines the canonical basis for certain tensor products in [ 23 ] , and his construction is generalized by [ 3 ] . One can also see details in Section 2.1. Inspired by [ 44 ] and [ 19 ] , the authors of [ 12 ] and [ 10 ] consider certain localizations of Lusztig’s category \mathcal{Q} on multi-framed quivers, and give a sheaf theoretic realization of the irreducible integrable highest weight module L(\lambda) and their tensor products. They also prove that the nonzero simple perverse sheaves in localization coincide with the canonical basis defined by [ 3 ] and [ 23 ] . See details in Section 3. To determine the decomposition rule, it is a natural idea to decompose the canonical basis of the tensor product into canonical bases of direct summands. However, the canonical basis is not compatible with direct summands in general, but only compatible with certain subquotients M[\geqslant\mu] , M[>\mu] and M[\geqslant\mu]/M[>\mu] . See details in Section 2.2. Fortunately, the informations of these subquotients are enough to determine the decomposition rule. Our first main result is to give a sheaf theoretic realization of those sub/quotient based modules mentioned above. More precisely, we base on the results in [ 10 ] , and define thick subcategories \mathcal{D}^{\geqslant\mu} and \mathcal{D}^{>\mu} for a dominant integral wight \mu of \mathfrak{g} by some micro-local conditions. The Grothendieck groups of localized Lusztig’s sheaves \mathcal{L}^{\geqslant\mu},\mathcal{L}^{>\mu} and \mathcal{L}^{\mu} in these thick subcategories are isomorphic to certain subquotients defined in [ 24 , 27.1 and 27.2] , and nonzero simple perverse sheaves provide the canonical bases of these subquotients. See details in Theorem 4.26 . Moreover, if we focus on a subset J\subseteq I and regard the vertices in I\backslash J as framing, similar micro-local conditions define thick subcategories \mathcal{D}^{\geqslant\mu_{J}} and \mathcal{D}^{>\mu_{J}} for a dominant integral wight \mu_{J} of the Levi subalgebra \mathfrak{l}=\mathfrak{l}_{J} , such that the Grothendieck groups of localized Lusztig’s sheaves \mathcal{L}^{\geqslant\mu_{J}},\mathcal{L}^{>\mu_{J}} and \mathcal{L}^{\mu_{J}} in these thick subcategories are isomorphic to certain subquotients of restricted modules. Simple perverse sheaves also provide the canonical bases, which can determine the restriction rule. See details in Theorem 4.33 . 1.3. Nakajima’s quiver varieties and their relations with canonical bases Using GIT theory, Nakajima defines the quiver varieties and consider their top Borel-Moore homology groups in [ 31 ] . The Hecke correspondence induces operators on Borel-Moore homology groups, which give \mathfrak{g} -module structures on Borel-Moore homology groups of quiver varieties. In particular, the top Borel-Moore homology group of the Lagrangian fiber \mathfrak{L}(\omega) is isomorphic to the irreducible integrable highest weight \mathfrak{g} -module. Later, in [ 33 ] , Nakajima also defines the tensor product varieties, whose top Borel-Moore homology groups realize

Read the complete text