arXiv reader

Character of Irreducible Representations Restricted to Finite Order Elements – An Asymptotic Formula

Shrawan Kumar and Dipendra Prasad

arXiv
2412.01742v1 · PDF · 2 December 2024
Categories
math.RT, math.AG, math.GR
Licence
CC BY 4.0 The site shows the text.

Abstract

Let G be a connected reductive group over the complex numbers and let T\subset G be a maximal torus. For any t\in T of finite order and any irreducible representation V(\lambda) of G of highest weight \lambda, we determine the character ch(t, V(\lambda)) by using the Lefschetz Trace Formula due to Atiyah-Singer and explicitly determining the connected components and their normal bundles of the fixed point subvariety (G/P)^t\subset G/P (for any parabolic subgroup P). This together with Wirtinger’s theorem gives an asymptotic formula for ch(t, V(n\lambda)) when n goes to infinity.

Items 25

Theorem

Theorem 1.1

C2OO

For any t\in T of finite order and any integer n\geqslant 0,

\Theta_{n\lambda}(t^{-1})=\sum_{v\in Y_t} D_v^t(n,\lambda),

where

D_v^t(n,\lambda)= \sum_{k\geqslant 0} \int_{X_P^t(v)}\,\,\frac{c_1\left(\mathcal{L}(\lambda)_{|_{X_P^t(v)}}\right)^k\frac{n^k}{k!}e^{-nv\lambda}(t)\cdot \operatorname{\mathrm{td}}(X_P^t(v))}{\prod_{\alpha\in(vR^+_P)\backslash R((G^t)^o)}\, [1-e^\alpha(t)e^{- c_1(\overline{\mathcal{L}}(\alpha)_{|_{X_P^t(v)}})}]},

where Y_t denotes the set of connected components of Z_{G}(t), R^+_P is the set of roots of the unipotent radical of P, \operatorname{\mathrm{td}} is the Todd genus of the tangent bundle, the notation Y_t and X_P^t(v) are explained in Proposition 2.1 and the notation \overline{\mathcal{L}}(\alpha) is as in the identification (?).

Corollary

Corollary 1.2

QQTQ

For any fixed 0\leq p<r, \chi_{\lambda_{|\left\{n\equiv p(mod\,r)\right\}}}^t is a polynomial function in n of degree \leq \max_{v\in Y_t}\left\{d_v\right\}.

Moreover, the coefficient of n^{d_v} in D_v^t(n,\lambda) restricted to \{ n\equiv p (mod\, r) \} is equal to

\frac{1}{d_v!}\int_{X_P^t(v)} c_1 \left(\mathcal{L}(\lambda)_{|_{X_P^t(v)}}\right)^{d_v} \frac{e^{-pv\lambda}(t)}{\prod_{\alpha\in \left(v R^+_P\right)\backslash R\left(\left(G^t\right)^o\right)}\left(1-e^{\alpha}(t)\right)}.

In particular, if there is a unique v_o\in Y_t such that d_{v_o}= \max_{v\in Y_t}\left\{d_v\right\}. Then, for any fixed 0\leq p< r,

(\chi^t_\lambda)_{|_{n\equiv p (mod \ r)}}

is a polynomial function of degree exactly equal to d_{v_o}.

Used by Proof
Theorem

Theorem 1.3

J55Y

Assume that t\in T is of order 2. Then, the function \left(\chi_{\lambda}^t\right)_{|\{ n\in 2\mathbbm{Z}_{\geq0}\}} is a polynomial function of degree exactly equal to d:= max_{v\in Y_t}\left\{d_v\right\}.

Proposition

Proposition 2.1

597T

With the notation as above, the set of connected components of X_P^S is bijectively parameterized by

Y_S:=\left\{v\in W^P:~B^S\cdot vP/P=vP/P \right\}.

For v\in Y_S, let X_P^S(v) be the corresponding connected component of X_P^S. Then, the morphism

\hat{\phi}_v:(G^S)^o\longrightarrow X^S_P, \quad g \mapsto g\cdot vP/P

descends to a (G^S)^o-equivariant variety isomorphism

\phi_v:(G^S)^o/\left((G^S)^o\cap(vPv^{-1})\right)\overset{\sim}{\longrightarrow} X_P^S(v).

Observe that the Borel subgroup B^S\subset (G^S)^o\cap (vPv^{-1}), since v\in Y_S. Moreover, since X_P^S is smooth, its connected components coincide with its irreducible components.

Proof of Proposition 2.1

Take a connected component \mathcal{Z}\subset X_P^S. It is clearly B^S-stable and hence, by Borel Fixed Point Theorem, contains a B^S-fixed point vP/P (for some v\in W^P) since a B^S-fixed point, in particular, is a T-fixed point. We claim that the map

\hat{\phi}_v:~(G^S)^o\rightarrow X_P^S, \quad g\mapsto g\cdot vP/P

induces an isomorphism

\phi_v:(G^S)^o/\left((G^S)^o\cap(vPv^{-1})\right)\overset{\sim}{\longrightarrow} \mathcal{Z}:

By the Bruhat decomposition, we get

X_P=\sqcup_{w\in W^P}BwP/P,

and hence

X_P^S =\sqcup_{w\in W^P}B^SwP/P.

Since vP/P\in \mathcal{Z}, we get

(G^S)^o\cdot vP/P\subset \mathcal{Z}.

Moreover, the isotropy subgroup of v in (G^S)^o is clearly (G^S)^o\cap (vPv^{-1})\supset B^S, thus we get a (G^S)^o-equivariant embedding

\phi_v:(G^S)^o/\left((G^S)^o\cap(vPv^{-1})\right) \hookrightarrow \mathcal{Z}.

We now prove that any (G^S)^o-orbit in X_P^S is closed: By identity (?), any (G^S)^o-orbit A in X_P^S is a certain union \sqcup_{w\in\theta(A)}B^S wP/P, for some subset \theta(A)\subset W^P. Take v_o\in \theta(A) of smallest length. Then, we prove that

B^S\cdot v_oP/P =v_o P/P.

For if (?) were false, take \mathcal{U}_{\alpha}\subset B^S \subset (G^S)^o such that

\mathcal{U}_\alpha\cdot v_o P/P \supsetneqq v_o P/P,

where \mathcal{U}_\alpha is the one-parameter additive group corresponding to a positive root \alpha. In particular

v_o^{-1}\mathcal{U}_\alpha v_o\nsubset P \mbox{ and hence } v_o^{-1}\mathcal{U}_{-\alpha} v_o\subset \mathcal{U}^P,

where \mathcal{U}^P is the unipotent radical of P, which gives

\mathcal{U}_{-\alpha}v_oP/P=v_oP/P.

By (?), v_o^{-1}\alpha is a negative root and hence

s_\alpha v_o< v_o\quad \mbox{ by [K, Lemma 1.3.13]},

where s_\alpha \in W is the reflection through \alpha. Since \mathcal{U}_\alpha\subset B^S\subset (G^{S})^o, then so is \mathcal{U}_{-\alpha}\subset(G^S)^o and hence the subgroup \langle \mathcal{U}_\alpha, \mathcal{U}_{-\alpha}\rangle\subset G generated by \mathcal{U}_\alpha and \mathcal{U}_{-\alpha} is contained in (G^S)^o. Thus, s_\alpha v_oP/P \in (G^S)^o\cdot v_oP/P. This is a contradiction to the minimal choice of v_o in the (G^S)^o-orbit. Thus (?) is true, proving that any (G^S)^o-orbit in X_P^S is closed. Further, by (?), there are only finitely many (G^S)^o-orbits in X_P^S.

We return to the embedding \phi_v as in (?). We now show that \phi_v is surjective: Since \mathcal{Z} is stable under the action of (G^S)^o, it is a (finite) union of (G^S)^o-orbits. But, each (G^S)^o-orbit is closed in X_P^S and hence \mathcal{Z} is a disjoint union of finitely many closed (G^S)^o-orbits. But, \mathcal{Z} being connected, it is a single (G^S)^o-orbit, proving that \phi_v is a bijective isomorphism.

Consider the map

\psi:Y_S\rightarrow \left\{\mbox{set of the connected components of } X_P^S\right\},

which takes

v\in Y_S \mapsto (G^S) \cdot vP/P.

By the above proof (G^S)^o\cdot vP/P is indeed a connected component of X_P^S and hence \psi is well defined. Moreover, \psi is injective since any compact homogeneous variety of a reductive connected algebraic group over \mathbbm{C} has a unique fixed point under any Borel subgroup, which follows from the Bruhat decomposition.

Further, as proved above, \psi is surjective. This proves the proposition.

Corollary

Corollary 2.2

ZUP5

With the notation as in Proposition 2.1, assume that P=B. Then, each connected component of X^S_B is isomorphic, as a (G^S)^o-variety, with (G^S)^o/B^S. In particular, each connected component of X_B^S is of the same dimension.

Used by Proof
Proof of Corollary 2.2

By the above proposition applied to the case of P=B, we get that any connected component \mathcal{Z} of X_B^S is isomorphic with (G^S)^o/\left((G^S)^o\cap(vBv^{-1})\right), for some v\in Y_S \subset W. Moreover, B^S\subset (G^S)^o\cap(vBv^{-1}). But, vBv^{-1} being a solvable group and B^S being a maximal solvable subgroup of (G^S)^o, we get that B^S=(G^S)^o\cap(vBv^{-1}). This proves the Corollary.

Example

Example 2.3

MOTL

For a general parabolic P, X_P^S does not necessarily have components of the same dimension. For example, take G=\operatorname{\mathrm{GL}}_n(\mathbbm{C}) and P a maximal parabolic subgroup such that X_P=\mathbbm{P}^{n-1}. Take S\subset T be the subgroup of order 2 generated by the diagonal element \left((+1)^m,(-1)^{n-m}\right), which is +1 along the first m entries of the diagonal and -1 in the following (n-m) entries. In this case,

X^S_P=\mathbbm{P}^{m-1}\sqcup \mathbbm{P}^{n-m-1}.

This provides a counter example if n\neq 2m.

Remark

Remark 2.4

0QMF

(1) Following the notation and assumptions of Proposition 2.1,

Y_S=\left\{ v\in W^P:~ \mathcal{U}_v^S=e \right\},

where \mathcal{U}_v\subset G is the unipotent group with Lie algebra

\bigoplus_{\alpha\in R^+ \cap v R_P^-} \mathfrak{g}_\alpha = \bigoplus_{\alpha\in R^+ \cap v R^-} \mathfrak{g}_\alpha,

where R^+ is the set of positive roots of G and R^-_P is the set of roots of the opposite unipotent radical of P. This follows since

B^S v~P/P= \mathcal{U}_v^S\cdot vP/P.

1ex

(2) If S is generated by a finite order t\in T which is regular, i.e., e^\alpha(t)\neq 1 for any root \alpha of G, then B^S=T (and hence (G^S)^o=T), Y_S=W^P and the connected components of X_P^S are all points (use Proposition 2.1).

Lemma

Lemma 2.5

21Y6

Following Proposition 2.1, we get the following:

\sum_{v\in Y_S} \# \left( W((G^S)^o)/W\left((G^S)^o\cap (vPv^{-1})\right)\right)=\# W^P,

where W(H) is the Weyl group of H.

In particular,

\# Y_S\geq \frac{\# W^P}{\#W((G^S)^o)}.

For P=B, in fact we have the equality

\#Y_S =\frac{\# W}{\#W((G^S)^o)}.
Used by Proof
Proof of Lemma 2.5

The identity (2.5) follows from Proposition 2.1 since wP/P\in X_P^S for any w\in W^P and \phi_v: (G^S)^o/\left((G^S)^o\cap vPv^{-1}\right)\overset{\sim}{\rightarrow}X_P^S(v) is a (G^S)^o-equivariant isomorphism for any v\in Y_S and, moreover, \sqcup_{v\in Y_S}X_P^S(v)=X_P^S.

Inequality (?), of course, follows immediately from identity (2.5).

To prove identity (?), observe that for P=B, (G^S)^o\cap v Pv^{-1}=B^S for any v\in Y_S (cf. proof of Corollary 2.2).

Theorem

Theorem 3.1

YGY3

With the notation as above, for any integer n\geqslant 0,

\operatorname{\mathrm{ch}}\left(t^{-1},V(n\lambda)\right)=\sum_{v\in Y_S} D^t_v(n,\lambda),

where

D^t_v(n,\lambda):= \sum_{k\geqslant 0} \int_{X_P^t(v)}\,\,\frac{c_1\left(\mathcal{L}(\lambda)_{|_{X_P^t(v)}}\right)^k\frac{n^k}{k!}e^{-nv\lambda}(t)\cdot \operatorname{\mathrm{td}}(X_P^t(v))}{\prod_{\alpha\in(vR^+_P)\backslash R((G^t)^o)}\, [1-e^\alpha(t)e^{- c_1(\overline{\mathcal{L}}(\alpha)_{|_{X_P^t(v)}})}]},

as earlier R^+_P is the set of roots of the unipotent radical of P and \operatorname{\mathrm{td}} is the Todd genus of the tangent bundle,

Proof of Theorem 3.1

Use [AS, Theorem 4.6] together with the identities (2.5) and (?) for the line bundle \mathcal{L}(n\lambda) over X_P. By the Borel-Weil-Bott Theorem,

\begin{aligned} &&H^i\left(X_P,\mathcal{L}(n\lambda)\right)=0,\quad \mbox{for all } i>0\\ \mbox{and} && H^0\left(X_P,\mathcal{L}(n\lambda)\right)= V(n\lambda)^*. \end{aligned}

Thus,

\begin{aligned} \sum_{i\geq0}(-1)^i \mbox{Trace}\left(t,~H^i\left(X_P,\mathcal{L}(n\lambda)\right)\right)&= \mbox{ch}\left(t, V(n\lambda)^*\right)\\ &= \mbox{ch}\left(t^{-1}, V(n\lambda)\right). \end{aligned}
Remark

Remark 3.2

928L

The above theorem remains true (by the same proof) for any standard parabolic subgroup P replacing P_\lambda as long as \lambda extends to a character of P; in particular, for P=B. Applying the above theorem for P=B and n=1, we get that \operatorname{\mathrm{ch}}\left(t^{-1},V(\lambda)\right) is a piecewise polynomial function in \lambda. In fact, it is a polynomial function restricted to the dominant elements of any coset X(T)/d\cdot X(T) (d being the order of t), where X(T) is the character group of T.

Corollary

Corollary 3.3

PMTA

For any fixed 0\leq p<r, \chi_{\lambda_{|\left\{n\equiv p(mod\,r)\right\}}}^t is a polynomial function in n of degree \leq \max_{v\in Y_S}\left\{d_v\right\}.

Moreover, the coefficient of n^{d_v} in D_v^t(n,\lambda) restricted to \{ n\equiv p (mod\, r) \} is equal to

\frac{1}{d_v!}\int_{X_P^t(v)} c_1 \left(\mathcal{L}(\lambda)_{|_{X_P^t(v)}}\right)^{d_v} \frac{e^{-pv\lambda}(t)}{\prod_{\alpha\in \left(v R^+_P\right)\backslash R\left(\left(G^t\right)^o\right)}\left(1-e^{\alpha}(t)\right)}.
Used by Proof
Proof of Corollary 3.3

The polynomial behavior of \chi_{\lambda|\left\{n\equiv p (mod r)\right\}} follows immediately from Theorem 3.1. To prove (1.2), again use Theorem 3.1 together with the fact that the constant term of the Todd genus of any manifold is 1 (cf. [F, Example 3.2.4]).

Remark

Remark 3.4

GKFU

Since \mathcal{L}(\lambda) is an ample line bundle over X_P=G/P, by Wirtinger theorem (cf. [GH, Chapter 0, \S2]), for any v\in Y_S,

\int_{X_P^t(v)}c_1\left(\mathcal{L}(\lambda)|_{X_P^t(v)}\right)^{d_v}>0.
Corollary

Corollary 3.5

9KUG

Following the notation and assumptions as in Theorem 3.1, assume further that there is a unique v_o\in Y_S such that d_{v_o}= \max_{v\in Y_S}\left\{d_v\right\}. Then, for any fixed 0\leq p< r,

\operatorname{\mathrm{ch}}\left(t^{-1}, V(n\lambda)_{|_{n\equiv p (mod \ r)}}\right)

is a polynomial function of degree exactly equal to d_{v_o}.

Corollary

Corollary 3.6

8CWO

With the notation and assumptions as in Theorem 3.1 and with the additional assumption that P=B,

\operatorname{\mathrm{ch}}\left(t^{-1}, V(n\lambda)\right)=\sum_{v\in Y_S}\sum_{k\geqslant 0}\, \int_{\overset{o}{X}^t} \frac{c_1\left(\mathcal{L}(v\lambda)_{|_{\overset{o}{X}^t}}\right)^k \frac{n^k}{k!} e^{-nv\lambda} (t)\operatorname{\mathrm{td}} \overset{o}{X}^t} {\prod_{\alpha\in(vR^+)\backslash \overset{o}{R}^t}\left[ 1-e ^\alpha (t) \cdot e^{-c_1 \left( \mathcal{L}(\alpha)_{ |_{\overset{o}{X}^t}}\right)}\right]},

where \overset{o}{X}^t:=(G^t)^o/B^t and \overset{o}{R}^t denotes the set of roots of (G^t)^o.

Proof of Corollary 3.6

To deduce the corollary from Theorem 3.1, we need to observe that under the isomorphism of Proposition 2.1 (for any v\in Y_S):

\phi_v:~\overset{o}{X}^t\rightarrow X^t_B(v),\quad gB^t\mapsto gvB/B,

the line bundle \mathcal{L}(\mu)_{|_{X_B^t(v)}} pulls back to

\phi_v^* \left(\mathcal{L}(\mu)_{|_{X_B^t(v)}}\right)\simeq \mathcal{L}(v\mu)_{|_{\overset{o}{X}^t}}

under the isomorphism

\left[g,\mathbbm{1}_{-v\mu}\right] \mapsto\left[gv,\mathbbm{1}_{-\mu}\right],\quad \mbox{for } g\in(G^t)^o,

where \mathbbm{1}_{-v\mu} is a basis of \mathbbm{C}_{-v\mu}. Moreover,

\phi_v^* \left(\overline{\mathcal{L}}(\mu)_{|_{X_B^t(v)}}\right)=\mathcal{L}(\mu)_{|_{\overset{o}{X}^t}}.
Remark

Remark 3.7

WGW0

The Todd genus of any flag variety G/P is determined by Brion [B, \S3].

Example

Example 3.8

M2AV

Let the assumption be as in Theorem 3.1. Assume further that t\in T is a regular element (i.e., e^\alpha(t)\neq 1 for any root \alpha of G) of finite order r. Then, the function n\mapsto \mbox{ch}\left(t^{-1}, V(n\lambda)\right) is the constant function 1, when n is restricted to r\mathbbm{Z}_{\geq0}.

Theorem

Theorem 4.1

AHQE

Follow the notation and assumptions as in Corollary 3.3 and assume further that t\in T is of order 2. Then, the function \chi_{\lambda|\left\{ n\in 2\mathbbm{Z}_{\geq0}\right\}}^t is a polynomial function of degree exactly equal to d:= max_{v\in Y_S}\left\{d_v\right\}.

Proof of Theorem 4.1

By Corollary 3.3, the coefficient of n^{d_v} in D_v^t(n,\lambda) restricted to 2\mathbbm{Z}_{\geq0} is equal to

\frac{1}{d_v!}\int_{X^t_P(v)} c_1\left( \mathcal{L}(\lambda)_{|_{X^t_P(v)}}\right)^{d_v}\cdot \frac{1}{\prod_{\alpha\in\left(vR^+_P\right)\backslash R\left(\left(G^t\right)^o\right)} \left(1-e^\alpha(t)\right)}.

But since t is of order 2, e^\alpha(t)=\pm1. Moreover, for \alpha\notin R\left(\left(G^t\right)^o\right), e^\alpha(t)=-1. Thus, the above sum reduces to

\frac{1}{d_v!}\int_{X_P^t(t)} \,c_1\left(\mathcal{L}(\lambda)_{|_{X^t_P(v)}}\right)^{d_v}\cdot 2^{ -\# \left(vR_P^+\backslash R\left(\left(G^t\right)^o\right)\right)}.

By Wirtinger’s Theorem (cf. Remark 2.3), the integral \int_{X_P^t(v)} c_1 \left(\mathcal{L}(\lambda)_{|_{X^t_P(v)}}\right)^{d_v}>0.

Since, by Theorem 3.1,

\chi_\lambda^t (n) =\sum_{v\in Y_S} D^t_v\left(n, \lambda\right),

the theorem follows.

Example

Example 4.2

D3KT

Consider G=\operatorname{\mathrm{GL}}_n(\mathbbm{C}) and t=\left((+1)^m,(-1)^{n-m}\right) as in Example 2.3. In the case n=2m and m odd, using Theorem 3.1, it can be seen that for

\lambda:= \left(\lambda_1=\lambda_2=\dots =\lambda_m\geq \lambda_{m+1}=\lambda_{m+2}=\dots =\lambda_n\right)

with \lambda_1 and \lambda_{m+1} of opposite parity,

\operatorname{\mathrm{ch}}(t, V(\lambda))= 0.

To prove this observe that, by Example 2.3, X_P^t= \mathbbm{P}^{m-1}\sqcup \mathbbm{P}^{m-1}. Moreover, in this case, Y_S= \{1, v_o\}, where v_o is the cycle (1, n, n-1, n-2, \dots, 2). Further, e^{-\lambda}(t) = - e^{-v_o\lambda}(t).

A similar result can be obtained for \operatorname{\mathrm{Sp}}(2n) and \operatorname{\mathrm{SO}}(n).

Source and licence

The text on this page is from arXiv:2412.01742v1 by Shrawan Kumar and Dipendra Prasad, under the licence CC BY 4.0. The corpus made these changes: it read the items out of the LaTeX source, it expanded the macros of the paper, it put each proof after its statement, and it converted the text to HTML. The formulas are the TeX of the paper, and KaTeX renders them. A formula or a reference can look different from the PDF. The page is published under CC BY 4.0.

The record of each item is in analysis/objects/2412/2412.01742.jsonl.