---
title: Lusztig's Weight Map Overview
url: https://www.emergentmind.com/topics/lusztig-s-weight-map
type: topic
---

# Lusztig's Weight Map Overview

The expression *Lusztig’s weight map* does not designate a single standard construction across the literature surveyed here. This suggests that the phrase is best treated as an umbrella label for several Lusztig-style assignments in representation theory, algebraic groups, Hecke algebras, and quantum groups: a polynomial-valued map \(\mu\mapsto \mathfrak M_\lambda^\mu(q)\) refining weight multiplicities, a Coxeter-theoretic weight function \(L\), the group-valued map \(\phi_G:G\to \Irr(W)\) whose fibers are Lusztig’s strata, the weight/idempotent part of the quantum Frobenius on modified quantum groups, and Lusztig–Vogan-type maps from orbit data to dominant weights [1406.1453] [1205.3762] [1909.11539] [1607.02117] [1711.00148].

## 1. Terminological scope and principal meanings

In the cited arXiv literature, several nearby constructions are explicitly present, while the phrase *weight map* itself is often absent. In particular, one finds Lusztig’s \(q\)-analogues of weight multiplicities, a *weight function* \(L\) in Coxeter and Hecke theory, the map \(\phi_G\) attached to Lusztig’s strata, Lusztig’s map \(\Psi\) for unipotent classes, the asymptotic Hecke algebra map \(\phi\), and crystal-theoretic weight or energy functions rather than a single universally adopted object [1909.11539] [1205.3762] [2110.07148] [1210.0223] [2412.20757].

| Setting | Representative construction | Source and target |
|---|---|---|
| Weight multiplicities | \(\mu\mapsto \mathfrak M_\lambda^\mu(q)\) | weights \(\mu\) to \(\mathbb Z[q]\) |
| Coxeter/Hecke theory | \(L:W\to \mathbb Z\) | Coxeter group to parameter values |
| Reductive groups | \(\phi_G(g)=\mathbf j_{W_s}^W\rho_u^{W_s}\) | \(G\) to \(\Irr(W)\) |
| Quantum groups | \(1_\lambda\), \(H_\alpha\), Frobenius on \(\dot U\) | idempotent/Cartan weight data |
| Nilpotent-orbit theory | \(\gamma:\Omega\to \Lambda^+\), \(LV_p\) | orbit data to dominant weights |

The common feature is not a fixed codomain but a recurrent role for *weight data*: highest weights, root gradings, toral characters, Weyl-group parameters, or nilpotent-orbit labels. A plausible implication is that any encyclopedia treatment must be comparative rather than singular.

## 2. Lusztig \(q\)-weight multiplicities as a polynomial-valued weight map

The most literal weight-theoretic construction in the cited material is the assignment, for fixed dominant \(\lambda\), sending a weight \(\mu\) to Lusztig’s \(q\)-analogue of the \(\mu\)-weight multiplicity in \(V_\lambda\). With
\[
\frac{1}{\prod_{\alpha\in \Delta_+}(1-qe^\alpha )}=\sum_{\mu\in Q_+} {\mathcal P}_q(\mu)e^\mu,
\]
Panyushev defines
\[
\mathfrak M_\lambda^\mu(q)=\sum_{w\in W}\varepsilon(w)\,{\mathcal P}_q\bigl(w(\lambda+\rho)-(\mu+\rho)\bigr).
\]
This polynomial satisfies \(\mathfrak M_\lambda^\mu(q)\equiv 0\) unless \(\lambda-\mu\in Q_+\); if \(\mu\preccurlyeq\lambda\), then \(\mathfrak M_\lambda^\mu(q)\) is monic of degree \(\operatorname{ht}(\lambda-\mu)\); \(\mathfrak M_\lambda^\lambda(q)\equiv 1\); and \(\mathfrak M_\lambda^\mu(1)=m_\lambda^\mu\), so evaluation at \(q=1\) recovers the ordinary weight multiplicity [1406.1453].

The same paper makes clear that positivity is delicate. If \(\mu\preccurlyeq\lambda\) but \(\mu\) is not a weight of \(V_\lambda\), then \(m_\lambda^\mu=0\), hence \(\mathfrak M_\lambda^\mu(1)=0\), and therefore \(\mathfrak M_\lambda^\mu(q)\) must have some negative coefficients. Broer’s criterion gives an exact characterization of those \(\mu\) for which \(\mathfrak M_\lambda^\mu(q)\in \mathbb N[q]\) for all dominant \(\lambda\): this is equivalent to \((\mu,\nu^\vee)\ge -1\) for all \(\nu\in\Delta^+\). The same work also records the induction lemma
\[
\mathfrak M_\lambda^{\gamma}(q)+\mathfrak M_\lambda^{s_\alpha(\gamma)-\alpha}(q)
=q\bigl(\mathfrak M_\lambda^{\gamma+\alpha}(q)+\mathfrak M_\lambda^{s_\alpha(\gamma)}(q)\bigr),
\]
and the generating-series identity
\[
\sum_{\mu:\,\mu \preccurlyeq \lambda} \mathfrak M_{\lambda}^\mu(q) e^\mu
=\chi_\lambda \,\xi_q,
\qquad
\xi_q=\prod_{\gamma\in\Delta^+}\frac{1-e^{-\gamma}}{1-qe^{-\gamma}}.
\]
These formulas place the map simultaneously in character theory, line-bundle cohomology on \(T^*(G/B)\), and the Gupta–Brylinski bilinear form [1406.1453].

Later work recasts this polynomial-valued map combinatorially. In type \(A\), Kostka–Foulkes polynomials are presented as Lusztig’s \(t\)-analogues of weight multiplicities, and a sign-reversing involution on Kostant partitions yields a positive formula in which the relevant statistic is simply the number of parts:
\[
K_{\lambda,\mu}(t)=\sum_{\boldsymbol\alpha\in D} t^{|\boldsymbol\alpha|}.
\]
Here the decisive grading map is \(\boldsymbol\alpha\mapsto |\boldsymbol\alpha|\), while the nearest literal weight map is either \(\boldsymbol\alpha\mapsto \overline{\boldsymbol\alpha}\) or \(wt(\boldsymbol\alpha)=-\overline{\boldsymbol\alpha}\) [2110.15394]. For type \(C\) and type \(B\) spin weights, KR-crystal models give positive formulas in terms of energy functions, for example
\[
\KL^{C_n}_{\lambda,\mu}(q) = \sum_{T} q^{\overline{D}(\phi_c(T))},
\]
and the same circle of ideas leads to a generalized root-weighting map \(L:\Delta^+\to\mathbb Q\), including the special choice
\[
L_A(\alpha)=
\begin{cases}
1 & \text{if }\alpha=\varepsilon_i-\varepsilon_j,\\
0 & \text{otherwise,}
\end{cases}
\]
which defines a \(\mathfrak{gl}_n\)-version of the type \(C\) \(q\)-weight multiplicity [2412.20757] [2409.02341].

## 3. Weight functions and Hecke-theoretic orderings

In Coxeter and Hecke theory, the closest object to a weight map is Lusztig’s *weight function*
\[
L:W\to \mathbb Z,
\qquad
L(ww')=L(w)+L(w')
\quad\text{whenever}\quad
\ell(ww')=\ell(w)+\ell(w').
\]
For finite Coxeter groups one assumes \(L(s)\ge 0\) for \(s\in S\), and in the equal-parameter case there exists \(a>0\) such that \(L(s)=a\) for all \(s\in S\). This datum determines the unequal-parameter Hecke algebra and controls Lusztig’s \(\mathbf a\)-invariants, families, and the preorder \(\preceq_L\) on \(\Irr(W)\). In type \(B_n\), with generators \(t,s_1,\dots,s_{n-1}\), the function is determined by
\[
b=L(t)\ge 0,\qquad a=L(s_i)>0,
\]
and the resulting preorder is described combinatorially by symbol data \(Z_{a,b}^N(\lambda,\mu)\). The paper proves
\[
E^{(\lambda,\mu)} \preceq_L E^{(\lambda',\mu')}
\Longrightarrow
(\lambda,\mu)\preceq_{a,b}(\lambda',\mu'),
\]
and globally
\[
E\preceq_L E' \implies a_{E'}\le a_E,
\]
with equality only if \(E,E'\) belong to the same family [1205.3762].

For the affine Weyl group of type \(\tilde C_2\), the corresponding notion is a *positive weight function*
\[
L:W\to \mathbb N
\]
with \(L(w)>0\) for \(w\neq e\) and the same additivity on reduced products. It is determined by
\[
L(s_1)=a,\qquad L(s_2)=b,\qquad L(s_0)=c,
\]
or equivalently by the ratios
\[
r_1=\frac ba,\qquad r_2=\frac ca.
\]
The Hecke algebra multiplication is then
\[
T_w T_s=
\begin{cases}
T_{ws}, & \ell(ws)=\ell(w)+1,\\[4pt]
T_{ws}+\bigl(q^{L(s)}-q^{-L(s)}\bigr)T_w, & \ell(ws)=\ell(w)-1.
\end{cases}
\]
The paper uses these parameters to compute Lusztig’s \(\mathbf a\)-function via balanced systems of cell representations and to prove Lusztig’s conjectures \(P1\)–\(P15\) for all choices of positive weight function in type \(\tilde C_2\) [1803.11067].

In this Hecke-theoretic sense, *weight* means parameter data rather than a map from weights of a representation. The distinction from the polynomial map \(\mu\mapsto \mathfrak M_\lambda^\mu(q)\) is substantial.

## 4. Maps to Weyl-group data: strata, unipotent classes, and affine cells

A different Lusztig-style construction appears for a connected reductive algebraic group \(G\) over an algebraically closed field. If \(g=su\) is the Jordan decomposition, with \(W_s\) the Weyl group of \(G_s^\circ\), then
\[
\phi_G(g)=\mathbf j_{W_s}^W\,\rho_u^{W_s},
\]
where \(\rho_u^{W_s}\) is the Springer representation attached to \(u\) with trivial local system and \(\mathbf j_{W_s}^W\) is Lusztig–Spaltenstein truncated induction. Its fibers
\[
X=\phi_G^{-1}(\chi),\qquad \chi\in \Irr(W),
\]
are Lusztig’s strata. The paper proves that each stratum is locally closed, is a union of sheets, and has irreducible components exactly the sheets it contains; these results hold in arbitrary characteristic [1909.11539]. The paper does not use the phrase *weight map* for \(\phi_G\), but it is one of the most important maps in the Lusztig-strata literature.

For unipotent classes, Lusztig’s map
\[
\Psi:\underline G\to \underline W
\]
is studied on spherical unipotent conjugacy classes. If \(w_\gamma\in W\) is the unique element such that \(Bw_\gamma B\cap \gamma\) is dense in \(\gamma\), then
\[
\Psi(\gamma)=W\cdot w_\gamma
\]
for spherical unipotent \(\gamma\). The same paper proves that a conjugacy class in a finite Weyl group has a unique maximal length element if and only if it has a maximum in Bruhat order [1210.0223]. Here again, the object is a map to Weyl-group conjugacy classes, but not a weight map in the multiplicity-theoretic sense.

An affine analogue appears in the map
\[
\widetilde{\operatorname{AV}}:\operatorname{Cells}(W_{\mathrm{aff}})\to \underline W,
\]
constructed from the singular supports of IC sheaves on the affine flag variety. The main theorem identifies it with the Kazhdan–Lusztig map on the Langlands dual side:
\[
\widetilde{\operatorname{AV}}
=
\operatorname{KL}_{G^\vee}\circ \underline c^{-1},
\]
where \(\underline c\) is Lusztig’s bijection from nilpotent orbits in \(G^\vee\) to two-sided cells of \(W_{\mathrm{aff}}\). The proof uses affine Springer fibers and Yun’s global Springer theory, and the result is presented as evidence for a conjecture of Lusztig on strata in a connected reductive group [2403.07080].

## 5. Modified quantum groups, idempotents, and additive weight operators

In the modified quantum-group setting, weights are encoded by orthogonal idempotents rather than by a single map. Lusztig’s modified algebra \(\dot U\) has generators
\[
1_\lambda,\qquad E_{\lambda+\alpha_i,\lambda},\qquad F_{\lambda,\lambda+\alpha_i},
\]
with
\[
1_\lambda 1_{\lambda'}=\delta_{\lambda,\lambda'}\,1_\lambda.
\]
For a unital \(\dot U\)-module \(M\), the \(\lambda\)-weight space is
\[
M_\lambda=1_\lambda M.
\]
The same paper identifies Lusztig’s modified algebra with modified forms of multi-parameter twisted Hopf algebras by an isomorphism preserving the idempotents:
\[
\psi(1'_\lambda)=1_\lambda,
\qquad
\psi(E'_{\lambda+\alpha_i,\lambda})=e_{i,\lambda}E_{\lambda+\alpha_i,\lambda},
\qquad
\psi(F'_{\lambda,\lambda+\alpha_i})=f_{i,\lambda}F_{\lambda,\lambda+\alpha_i}.
\]
This means that the weight labels are preserved exactly, while the root generators are rescaled by explicit weight-dependent scalars [1312.0838].

For the quantum Frobenius on \(\dot U_{\mathbb O_p}\), the weight behavior becomes an actual partial map on idempotents. The paper defines
\[
\mathrm{Fr}(E_n^{(a)})=
\begin{cases}
\mathsf E_{n/p}^{(a/p)} & \text{if } p\mid a,n,\\
0 & \text{otherwise},
\end{cases}
\qquad
\mathrm{Fr}(F_n^{(a)})=
\begin{cases}
\mathsf F_{n/p}^{(a/p)} & \text{if } p\mid a,n,\\
0 & \text{otherwise}.
\end{cases}
\]
A plausible implication is that on idempotents one has the corresponding rule \(1_n\mapsto 1_{n/p}\) when \(p\mid n\), and \(0\) otherwise. The paper then categorifies this using a thickening functor that sends weight \(n\) to weight \(np\) [1607.02117].

A further refinement appears in Lusztig’s quantum group of divided powers, where one constructs primitive elements
\[
H_\alpha=\frac{K_\alpha^{2\ell_\alpha}-1}{\Phi_{\ell_\alpha}(v_\alpha^2)}.
\]
After specialization, they satisfy
\[
\Delta(H_\alpha)=1\otimes H_\alpha+H_\alpha\otimes 1,
\qquad
[H_\alpha,E_\beta]=\frac{2(\alpha,\beta)}{(\alpha,\alpha)}E_\beta.
\]
These \(H_\alpha\) realize the additive Cartan part of the unrolled small quantum group inside \(U_q^{\mathcal L}(\mathfrak g)\) and recover full additive weight data from the multiplicative \(K\)-operators [1702.05164].

On the geometric side of Hall-type realizations, Lusztig’s algebra \(\mathbf f\) carries the grading
\[
|\theta_i|=i\in \mathbb NI,\qquad |xy|=|x|+|y|,
\]
while an \(I\)-graded vector space \(\mathbf V\) has dimension vector
\[
\underline{\dim}\,\mathbf V=\sum_{i\in I}(\dim_{\mathbb K}V_i)\,i\in \mathbb NI,
\]
which the paper states can also be viewed as an element in the weight lattice \(P\). This is the basic weight bookkeeping behind the geometric realization of Lusztig’s symmetries on the whole quantum group [1704.00166].

## 6. Lusztig–Vogan type maps and distinguished weights

A map whose codomain is literally a dominant weight lattice appears in the Lusztig–Vogan setting. For \(G=GL_n(\mathbb C)\), the bounded derived category \(D^b(\mathrm{Coh}^G(\mathcal N))\) has one basis indexed by \(\Omega\), the set of pairs \((\mathcal O,\mathcal E)\) of nilpotent orbits and irreducible \(G\)-equivariant vector bundles, and another indexed by \(\Lambda^+\), the dominant weights. Bezrukavnikov’s theorem gives a unique bijection
\[
\gamma:\Omega\to \Lambda^+.
\]
In type \(A\), the paper provides a direct combinatorial computation: if \((\alpha,v)\) encodes the orbit and bundle data, then
\[
\gamma(\alpha,v)=\mathrm{dom}\bigl(2_1(\alpha,v)+2\rho_\alpha\bigr),
\]
where \(2_1(\alpha,v)\in \Lambda_\alpha^+\) is produced by a recursive algorithm. This is one of the clearest instances of a Lusztig-style map to actual weights [1711.00148].

A later development studies the iterated Lusztig–Vogan bijection in type \(A\). Writing
\[
LV=\kappa\circ E^{-1}\circ \phi,
\qquad
LV_p=\frac1p\circ LV,
\]
the paper defines distinguished weights as those dominant weights whose repeated images under \(LV_p\) eventually become zero. It proves the anti-symmetry
\[
\lambda_i=-\lambda_{n+1-i}
\]
for every distinguished weight \(\lambda\), and gives the asymptotic formula
\[
|\Lambda_{n,k}^+|
\sim
\frac{a_{\left\lfloor \frac{n+1}2\right\rfloor}}{\left\lfloor \frac n2\right\rfloor!}\,
k^{\left\lfloor \frac n2\right\rfloor},
\]
where the \(a_i\) are the telephone numbers defined by
\[
a_0=a_1=1,\qquad a_i=a_{i-1}+(i-1)a_{i-2}.
\]
This places the weight map in a recursive dynamical setting on dominant weights [2505.19460].

Taken together, these constructions suggest that *Lusztig’s weight map* is not a single invariant but a family of context-dependent assignments. In weight-multiplicity theory it is the \(q\)-deformed map \(\mu\mapsto \mathfrak M_\lambda^\mu(q)\); in Hecke theory it is the parameter function \(L\); in geometric representation theory it may be \(\phi_G\), \(\Psi\), or \(\widetilde{\operatorname{AV}}\); in quantum groups it is encoded by idempotents, Frobenius divisibility, and primitive Cartan operators; and in Lusztig–Vogan theory it becomes an explicit map from orbit data to dominant weights. The unifying theme is the controlled transfer of weight information across algebraic, geometric, and combinatorial models.

Source: https://www.emergentmind.com/topics/lusztig-s-weight-map