This paper was converted on www.awesomepapers.org from LaTeX by an anonymous user.
Want to know more? Visit the Converter page.

The Weyl group of type A1A_{1} root systems extended by an abelian group111 The research for this article was supported by a postdoctoral fellowship at the Department of Mathematics and Statistics at Dalhousie University.

Georg W. Hofmann
Abstract

We investigate the class of root systems RR obtained by extending an A1A_{1}-type irreducible root system by a free abelian group GG. In this context there are two reflection groups with respect to a discrete symmetric space TT associated to RR, namely, the Weyl group 𝒲\mathcal{W} of RR and a group 𝒰\mathcal{U} with a so-called presentation by conjugation. We show that the natural homomorphism 𝒰𝒲\mathcal{U}\to\mathcal{W} is an isomorphism if and only if an associated subset Tab{0}T^{\mathrm{ab}}\setminus\{0\} of G2=G/2GG_{2}=G/2G is 2-independent, i.e. its image under the map G2G2G2,gggG_{2}\to G_{2}\otimes G_{2},~g\mapsto g\otimes g is linearly independent over the Galois field 𝔽2\mathbb{F}_{2}. Mathematics Subject Classification 2000: 20F55, 17B65, 17B67, 22E65 Key Words and Phrases: Weyl group, root system, presentation by conjugation, discrete symmetric space

1 Introduction

We consider a root system RR extended by an abelian group GG, a notion that is introduced in [Yos04]. It generalizes the concepts of extended affine root systems (see [AAB+97], for instance) and affine root systems in the sense of [Sai85], both of which are generalizations of root systems of affine Kac-Moody algebras (see [MP95], for instance). The Weyl group 𝒲\mathcal{W} of RR is not necessarily a Coxeter group, so a more general presentation is needed to capture the algebraic structure of 𝒲\mathcal{W}. The group 𝒰\mathcal{U} is given by the so-called presentation by conjugation with respect to RR:

𝒰(r^α)αR×|\displaystyle\mathcal{U}\cong\big{\langle}(\hat{r}_{\alpha})_{\alpha\in R^{\times}}~|~ r^α=r^β if α and β are linearly dependent,\displaystyle\hat{r}_{\alpha}=\hat{r}_{\beta}\text{~~~if $\alpha$ and $\beta$ are linearly dependent,}
r^α2=1,r^αr^βr^α1=r^rα(β); for α,βR.\displaystyle\hat{r}_{\alpha}^{2}=1,~~\hat{r}_{\alpha}\hat{r}_{\beta}\hat{r}_{\alpha}^{-1}=\hat{r}_{r_{\alpha}(\beta)};\text{ for }\alpha,\beta\in R\big{\rangle}.

There is a natural group homomorphism from 𝒰\mathcal{U} onto 𝒲\mathcal{W}.

Question  Is 𝒰𝒲\mathcal{U}\to\mathcal{W} injective? In other words, does 𝒲\mathcal{W} have the presentation by conjugation with respect to RR?

This question has been studied for various root systems in [Kry00], [Aza99], [Aza00], [AS07], [AS08], [Hof07], [Hof08].

In this note we investigate the case that RR is of type A1A_{1}, i.e. the underlying finite root system consists of two roots. This type of root system RR allows for less rigidity then other types and is therefor of special interest as a prototype. We prove the following result that allows an answer to the question above using an algorithmic approach.

Suppose RR is a type A1A_{1} root system extended by a free abelian group GG. Then a subset TabT^{\mathrm{ab}} of G2=G/2GG_{2}=G/2G can be associated to it in a natural way. This subset is called 2-independent, if its image under G2G2G2,gggG_{2}\to G_{2}\otimes G_{2},g\mapsto g\otimes g is a linearly independent set.

Theorem  The natural homomorphism 𝒰𝒲\mathcal{U}\to\mathcal{W} is an isomorphism if and only if Tab{0}T^{\mathrm{ab}}\setminus\{0\} is 2-independent in G2G_{2}.

This result provides an attractive alternative to a characterization proved in [AS08] using so-called integral collections. Our answer to the question above is more general than that in [AS08] as GG is not required to be finitely generated. We expect that the idea of 2-independence that we have introduced will play an important role in understanding the question for root systems of the types BnB_{n} and CnC_{n}.

2 Discrete symmetric spaces and their reflection groups

In this section we provide the basic terminology for the following sections. The notion of a discrete symmetric space is a special case of the symmetric spaces introduced in [Loo69]. The associated category of reflection groups is introduced in [Hof08] and more details can be found there.

Definition 2.1

(Discrete symmetric space) Let TT be a set with a (not necessarily associative) multiplication

μ:T×TT,(s,t)st.\displaystyle\mu:T\times T\to T,~(s,t)\mapsto s{\cdot}t.

Then the pair (T,μ)(T,\mu) is called a discrete symmetric space if the following conditions are satisfied for all ss, tt and rTr\in T:

  1. (S1)

    ss=s,s{\cdot}s=s,

  2. (S2)

    s(st)=t,s{\cdot}(s{\cdot}t)=t,

  3. (S3)

    r(st)=(rs)(rt).r{\cdot}(s{\cdot}t)=(r{\cdot}s){\cdot}(r{\cdot}t).

By abuse of language, we will sometimes say that TT is a discrete symmetric space instead of saying that (T,μ)(T,\mu) is a discrete symmetric space. If st=ts{\cdot}t=t for all ss and tTt\in T then we call μ\mu the trivial multiplication. \diamond

For the remainder of this section, let TT be a discrete symmetric space.

Definition 2.2

(Reflection group) Let 𝒳\mathcal{X} be a group acting on TT. We will denote the element in TT obtained by xx acting on tt by x.tx.t. Let

𝒳:T𝒳,tt𝒳\hskip 1.42262pt\begin{picture}(1.0,5.0)(-0.5,-3.0)\circle*{2.0}\end{picture}\hskip 1.42262pt^{\mathcal{X}}:~T\to\mathcal{X},~t\mapsto t^{\mathcal{X}}

be a function. Then (𝒳,𝒳)(\mathcal{X},\hskip 1.42262pt\begin{picture}(1.0,5.0)(-0.5,-3.0)\circle*{2.0}\end{picture}\hskip 1.42262pt^{\mathcal{X}}) is called a TT-reflection group, if the following conditions are satisfied:

  1. (G1)

    The group 𝒳\mathcal{X} is generated by the set T𝒳:={t𝒳|tT}T^{\mathcal{X}}:=\{t^{\mathcal{X}}~|~t\in T\}.

  2. (G2)

    For all ss and tTt\in T we have t𝒳.s=ts.t^{\mathcal{X}}.s=t{\cdot}s.

  3. (G3)

    For all ss and tTt\in T we have t𝒳s𝒳(t𝒳)1=(ts)𝒳.t^{\mathcal{X}}s^{\mathcal{X}}(t^{\mathcal{X}})^{-1}~=~(t{\cdot}s)^{\mathcal{X}}.

  4. (G4)

    For every tTt\in T we have (t𝒳)2=1.(t^{\mathcal{X}})^{2}=1.

If we do not need to specify the map 𝒳\hskip 1.42262pt\begin{picture}(1.0,5.0)(-0.5,-3.0)\circle*{2.0}\end{picture}\hskip 1.42262pt^{\mathcal{X}} we will also say that 𝒳\mathcal{X} is a reflection group instead of saying that (𝒳,𝒳)(\mathcal{X},\hskip 1.42262pt\begin{picture}(1.0,5.0)(-0.5,-3.0)\circle*{2.0}\end{picture}\hskip 1.42262pt^{\mathcal{X}}) is a reflection group. \diamond

Definition 2.3

(Reflection morphism) Let 𝒳\mathcal{X} and 𝒴\mathcal{Y} be TT-reflection groups. Then a group homomorphism φ:𝒳𝒴\varphi:\mathcal{X}\to\mathcal{Y} is called a TT-morphism, if φ(t𝒳)=t𝒴\varphi(t^{\mathcal{X}})=t^{\mathcal{Y}} for every tTt\in T. \diamond

Let the group 𝒰\mathcal{U} be given by the presentation

𝒰:=(t𝒰)tT|(t𝒰)2=1 and t𝒰s𝒰(t𝒰)1=(ts)𝒰 for s and tT.\displaystyle\mathcal{U}:=\big{\langle}(t^{\mathcal{U}})_{t\in T}~\big{|}~(t^{\mathcal{U}})^{2}=1\text{~~and~~}t^{\mathcal{U}}s^{\mathcal{U}}(t^{\mathcal{U}})^{-1}=(t{\cdot}s)^{\mathcal{U}}\text{~~~for $s$ and $t\in T$}\big{\rangle}.

There is map 𝒰:T𝒰,tt𝒰\hskip 1.42262pt\begin{picture}(1.0,5.0)(-0.5,-3.0)\circle*{2.0}\end{picture}\hskip 1.42262pt^{\mathcal{U}}:~T\to\mathcal{U},~t\mapsto t^{\mathcal{U}} associated to the presentation. An action of 𝒰\mathcal{U} on TT can be defined satisfying t𝒰.s=tst^{\mathcal{U}}.s=t{\cdot}s for all ss and tTt\in T. With respect to this action the pair (𝒰,𝒰)(\mathcal{U},\hskip 1.42262pt\begin{picture}(1.0,5.0)(-0.5,-3.0)\circle*{2.0}\end{picture}\hskip 1.42262pt^{\mathcal{U}}) is a TT-reflection group. There is a unique TT-morphism from 𝒰\mathcal{U} into any other TT-reflection group.

Definition 2.4

The pair (𝒰,𝒰)(\mathcal{U},\hskip 1.42262pt\begin{picture}(1.0,5.0)(-0.5,-3.0)\circle*{2.0}\end{picture}\hskip 1.42262pt^{\mathcal{U}}) is called the initial TT-reflection group. \diamond

3 Type A1A_{1} root systems extended by an abelian group

In this section we introduce the concept of a type A1A_{1} root system extended by an abelian group GG in an ad hoc manner. Thus we avoid presenting the details of the definition for more general types.

Let (G,+)(G,+) be an abelian group. Define the multiplication

G×GG,(g,h)gh=2gh.\displaystyle G\times G\to G,~(g,h)\mapsto g{\cdot}h=2g-h. (1)

Now let TT be a generating subset of GG such that 0T0\in T and GTTG{\cdot}T\subseteq T. It is straightforward to verify that TT with the restriction of the multiplication above is a discrete symmetric space. The set R:=T×{1,1}R:=T\times\{1,-1\} is a type A1A_{1} root system extended by the abelian group GG in the sense of [Yos04] or [Hof08].

Consider the two-element group 𝒱:={1,1}\mathcal{V}:=\{1,-1\} with its action on GG characterized by 1g=g-1g=-g for all gGg\in G. Set 𝒜:=G𝒱\mathcal{A}:=G\rtimes\mathcal{V}. Then 𝒜\mathcal{A} acts on TT via

(g,v).t=2g+vt.\displaystyle(g,v).t=2g+vt.

The map

𝒜:T𝒜,tt𝒜=(t,1)\displaystyle\hskip 1.42262pt\begin{picture}(1.0,5.0)(-0.5,-3.0)\circle*{2.0}\end{picture}\hskip 1.42262pt^{\mathcal{A}}:~T\to\mathcal{A},~t\mapsto t^{\mathcal{A}}=(t,-1)

turns 𝒜\mathcal{A} into a TT-reflection group.

In general, if BB is a group, AA is an abelian group, and f:B×BAf:B\times B\to A is a cocycle, then the set A×BA\times B with the multiplication given by

(a,b)(a,b)=(a+a+f(b,b),bb)\displaystyle(a,b)(a^{\prime},b^{\prime})=\big{(}a+a^{\prime}+f(b,b^{\prime}),bb^{\prime}\big{)}

defines a group denoted by A×fBA\times_{f}B which is a central extension of BB.

The set (GG)×G×𝒱(G\wedge G)\times G\times\mathcal{V} with the multiplication

(l,g,v)(l,g,v):=(l+l+g(vg),g+vg,vv)\displaystyle(l,g,v)(l^{\prime},g^{\prime},v^{\prime}):=(l+l^{\prime}+g\wedge(vg^{\prime}),g+vg^{\prime},vv^{\prime})

is a group. We denote it by (GG)×G𝒱(G\wedge G)\times_{\wedge}G\rtimes\mathcal{V}. It can equally be interpreted as the semidirect product of the Heisenberg group (GG)×G(G\wedge G)\times_{\wedge}G with 𝒱\mathcal{V} or a central extension of 𝒜\mathcal{A} by GGG\wedge G with cocycle f:𝒜2GG,((g,v),(g,v))g(vg).f:~\mathcal{A}^{2}\to G\wedge G,~\big{(}(g,v),(g^{\prime},v^{\prime})\big{)}\mapsto g\wedge(vg^{\prime}).

Set

𝒲:T(GG)×G𝒱,tt𝒲=(0,t,1).\displaystyle\hskip 1.42262pt\begin{picture}(1.0,5.0)(-0.5,-3.0)\circle*{2.0}\end{picture}\hskip 1.42262pt^{\mathcal{W}}:~T\to(G\wedge G)\times_{\wedge}G\rtimes\mathcal{V},t\mapsto t^{\mathcal{W}}=(0,t,-1).

Let 𝒲\mathcal{W} be the subgroup of (GG)×G𝒱(G\wedge G)\times_{\wedge}G\rtimes\mathcal{V} generated by T𝒲T^{\mathcal{W}}. Then (𝒲,𝒲)(\mathcal{W},\hskip 1.42262pt\begin{picture}(1.0,5.0)(-0.5,-3.0)\circle*{2.0}\end{picture}\hskip 1.42262pt^{\mathcal{W}}) is a TT-reflection group with the action of 𝒲\mathcal{W} on TT induced by the action of 𝒜\mathcal{A} on TT.

Definition 3.1

The group 𝒲\mathcal{W} is called the Weyl group of RR. \diamond

This definition of the Weyl group coincides with the definition of Weyl groups given in [Hof08] if GG is free abelian and the one given in [Aza99] if GG is finitely generated free abelian.

4 The abelian 2-group case

In this section we investigate the case where GG is an elementary abelian 2-group. So we may think of GG as a vector space over the Galois field 𝔽2\mathbb{F}_{2} with two elements. From (1) it immediately follows that TT has the trivial multiplication.

Denote by GsymGG\otimes_{\mathrm{sym}}G the subgroup of GGG\otimes G generated by the elements of the set {vv|vG}.\{v\otimes v~|~v\in G\}. The group homomorphism

GGGsymG characterized by ghghhg\displaystyle G\otimes G\to G\otimes_{\mathrm{sym}}G\text{~~characterized by~~}g\otimes h\mapsto g\otimes h-h\otimes g

factors through GGG\wedge G giving a group homomorphism

π:GGGsymG characterized by uvuvvu.\displaystyle\pi:~G\wedge G\to G\otimes_{\mathrm{sym}}G\text{~~characterized by~~}u\wedge v\mapsto u\otimes v-v\otimes u.

If BB is an ordered basis of GG then {b1b2|b1<b2B}\{b_{1}\wedge b_{2}~|~b_{1}<b_{2}\in B\} is a basis of GGG\wedge G. Its image under π\pi is linearly independent, so π\pi is injective.

Define the map

φ:(GG)×GGsymG,(t,g)(π(t)+gg).\displaystyle\varphi:~(G\wedge G)\times_{\wedge}G\to G\otimes_{\mathrm{sym}}G,~(t,g)\mapsto(\pi(t)+g\otimes g).
Theorem 4.1

The map φ\varphi is a group isomorphism such that φ(0,g)=gg\varphi(0,g)=g\otimes g for all gGg\in G.

Proof.

To see that φ\varphi is a group homomorphism let s,tGGs,t\in G\wedge G and g,hGg,h\in G. Then

φ((s,g)(t,h))\displaystyle\varphi\big{(}(s,g)(t,h)\big{)} =φ(s+t+gh,g+h)\displaystyle=\varphi(s+t+g\wedge h,g+h)
=φ(s)+π(t)+gh+hg+(g+h)(g+h)\displaystyle=\varphi(s)+\pi(t)+g\otimes h+h\otimes g+(g+h)\otimes(g+h)
=φ(s)+π(t)+gg+hh=φ(s,g)+φ(t,h).\displaystyle=\varphi(s)+\pi(t)+g\otimes g+h\otimes h=\varphi(s,g)+\varphi(t,h).

It is clear that φ\varphi is surjective, since it has a generating set in its image.

Since we are working with characteristic 2, the map

GGG,vvvG\to G\vee G,~v\mapsto v\vee v

is an injective group homomorphism. We denote by GsymGG\vee_{\mathrm{sym}}G the additive subgroup of GGG\vee G generated by {gg|gG}\{g\vee g~|~g\in G\}. So we have a group isomorphism

GsymGG.\displaystyle G\vee_{\mathrm{sym}}G\to G.

Its composition with the quotient homomorphism GsymGGsymGG\otimes_{\mathrm{sym}}G\to G\vee_{\mathrm{sym}}G yields a homomorphism

:GsymGGsatisfyinggg=g.\sqrt{\hskip 1.42262pt\begin{picture}(1.0,5.0)(-0.5,-3.0)\circle*{2.0}\end{picture}\hskip 1.42262pt}:~G\otimes_{\mathrm{sym}}G\to G~~~~~\text{satisfying}~~~~~\sqrt{g\otimes g}=g.

It vanishes on the image of π\pi, since

π(gh)\displaystyle\sqrt{\pi(g\wedge h)} =ghhg=gh+hg\displaystyle=\sqrt{g\otimes h-h\otimes g}=\sqrt{g\otimes h+h\otimes g}
=(g+h)(g+h)gghh\displaystyle=\sqrt{(g+h)\otimes(g+h)}-\sqrt{g\otimes g}-\sqrt{h\otimes h}
=g+hgh=0\displaystyle=g+h-g-h=0

for all gg and hGh\in G.

To show that φ\varphi is injective, let (t,v)ker(φ)(t,v)\in\ker(\varphi), so π(t)=vv\pi(t)=v\otimes v. Taking the square root on both sides yields 0=v0=v. We conclude π(t)=0\pi(t)=0. Since π\pi is injective we obtain t=0t=0. ∎

In this section the action of 𝒱\mathcal{V} on GG is trivial, so the reflection group 𝒜\mathcal{A} is given by the direct product 𝒜=G×𝒱\mathcal{A}=G\times\mathcal{V}. The Weyl group 𝒲\mathcal{W} is given as the subgroup of ((GG)×G)×𝒱\big{(}(G\wedge G)\times_{\wedge}G\big{)}\times\mathcal{V} generated by the image of

𝒲:T((GG)×G)×𝒱,tt𝒲=(0,t,1).\displaystyle\hskip 1.42262pt\begin{picture}(1.0,5.0)(-0.5,-3.0)\circle*{2.0}\end{picture}\hskip 1.42262pt^{\mathcal{W}}:~T\to\big{(}(G\wedge G)\times_{\wedge}G\big{)}\times\mathcal{V},~t\mapsto t^{\mathcal{W}}=(0,t,-1).

Due to the preceding theorem, the Weyl group can also be given as the subgroup of (GsymG)×𝒱\big{(}G\otimes_{\mathrm{sym}}G)\times\mathcal{V} generated by the image of

𝒲:T(GsymG)×𝒱,tt𝒲=(tt,1).\displaystyle\hskip 1.42262pt\begin{picture}(1.0,5.0)(-0.5,-3.0)\circle*{2.0}\end{picture}\hskip 1.42262pt^{\mathcal{W}}:~T\to(G\otimes_{\mathrm{sym}}G)\times\mathcal{V},~t\mapsto t^{\mathcal{W}}=(t\otimes t,-1). (2)

Let F:=F(T{0})F:=F(T\setminus\{0\}) be the free vector space on the set T{0}T\setminus\{0\} with the embedding ι:T{0}F\iota:T\setminus\{0\}\to F. The initial reflection group is given by 𝒰=F×𝒱\mathcal{U}=F\times\mathcal{V} with the map

T𝒰,t{(ι(t),1)if t0(0,1)if t=0.\displaystyle T\to\mathcal{U},~t\mapsto\begin{cases}(\iota(t),-1)&\text{if $t\neq 0$}\\ (0,-1)&\text{if $t=0$.}\end{cases}
Definition 4.2

A subset MM of GG is called 2-dependent, if the elements of the set {gg|gM}\{g\otimes g~|~g\in M\} are linearly dependent in GGG\otimes G. The set MM is called 2-independent if it is not 2-dependent. \diamond

Example 4.3
  1. a)

    A linearly independent subset MM of GG is 2-independent, due to the homomorphism \sqrt{\hskip 1.42262pt\begin{picture}(1.0,5.0)(-0.5,-3.0)\circle*{2.0}\end{picture}\hskip 1.42262pt} used in the proof of Theorem 4.1.

  2. b)

    Set G=(𝔽2)2G=(\mathbb{F}_{2})^{2}. Then the set MM of all nonzero vectors in GG is 2-independent, since the matrices

    (1000),(0001),(1111)\begin{pmatrix}1&0\\ 0&0\end{pmatrix},\begin{pmatrix}0&0\\ 0&1\end{pmatrix},\begin{pmatrix}1&1\\ 1&1\end{pmatrix}

    are linearly independent.

  3. c)

    Set G=(𝔽2)nG=(\mathbb{F}_{2})^{n}. Any subset MM of GG with cardinality |M|>n(n+1)2|M|>\frac{n(n+1)}{2} is 2-dependent, since dim𝔽2(GsymG)=n(n+1)2\dim_{\mathbb{F}_{2}}\big{(}G\otimes_{\mathrm{sym}}G\big{)}=\frac{n(n+1)}{2}.

Theorem 4.4

The reflection morphism 𝒰𝒲\mathcal{U}\to\mathcal{W} is injective if and only if the set T{0}T\setminus\{0\} is 2-independent in GG.

Proof.

We will use the form of the Weyl group 𝒲\mathcal{W} given in (2). Suppose 𝒰𝒲\mathcal{U}\to\mathcal{W} is not injective. Then there is a non-trivial element in its kernel. This element can be written as (i=1nι(ti),σ)GsymG×𝒱\big{(}\sum_{i=1}^{n}\iota(t_{i}),\sigma\big{)}\in G\otimes_{\mathrm{sym}}G\times\mathcal{V} for distinct elements t1,t2,,tnT{0}t_{1},~t_{2},\dots,t_{n}\in T\setminus\{0\} and σ𝒱\sigma\in\mathcal{V}. It follows that σ=1\sigma=1 and i=1ntiti=0.\sum_{i=1}^{n}t_{i}\otimes t_{i}=0. So t1t_{1}, t2,,tnt_{2},\dots,t_{n} are 2-dependent. This implies that T{0}T\setminus\{0\} is 2-dependent.

Conversely, suppose T{0}T\setminus\{0\} is 2-dependent, say i=1ntiti=0\sum_{i=1}^{n}t_{i}\otimes t_{i}=0 for distinct elements t1,t2,,tnT{0}t_{1},~t_{2},\dots,t_{n}\in T\setminus\{0\} and n1n\geq 1. Then (i=1nti,0)\big{(}\sum_{i=1}^{n}t_{i},0\big{)} is a nontrivial element in the kernel of 𝒰𝒲\mathcal{U}\to\mathcal{W}. ∎

Denote the reflection morphism 𝒰𝒲\mathcal{U}\to\mathcal{W} above by φ\varphi. Then Example 4.3 yields

Corollary 4.5
  1. (i)

    The map φ\varphi is injective if T{0}T\setminus\{0\} is a basis of GG.

  2. (ii)

    The map φ\varphi is not injective if |T{0}|>n(n+1)2|T\setminus\{0\}|>\frac{n(n+1)}{2}, where n=dim(G)n=\dim(G).

  3. (iii)

    If T=GT=G, then φ\varphi is an isomorphism if and only if dim(G)2\dim(G)\leq 2.

5 The free abelian case

In this section let GG be a free abelian group. We will reduce the situation to that of the former section. More details can be found in [Hof08] Section 2, in particular in Construction 2.10.

Let 𝒰\mathcal{U} be the initial TT-reflection group and let 𝒲\mathcal{W} be the Weyl group. The abelianizations 𝒰ab\mathcal{U}^{\mathrm{ab}} and 𝒲ab\mathcal{W}^{\mathrm{ab}} are TabT^{\mathrm{ab}}-reflection groups, where TabT^{\mathrm{ab}} is the image of TT under the quotient homomorphism GG2=G/2GG\to G_{2}=G/2G. This is a discrete symmetric space with the trivial multiplication. More precisely 𝒰ab\mathcal{U}^{\mathrm{ab}} is the initial TabT^{\mathrm{ab}}-reflection group and 𝒲ab=(G2G2)×G2×𝒱\mathcal{W}^{\mathrm{ab}}=\big{(}G_{2}\wedge G_{2}\big{)}\times_{\wedge}G_{2}\times\mathcal{V} is the Weyl group for the discrete symmetric space TabT^{\mathrm{ab}}.

The TT-reflection morphism 𝒰𝒲\mathcal{U}\to\mathcal{W} yields a TabT^{\mathrm{ab}}-morphism 𝒰ab𝒲ab\mathcal{U}^{\mathrm{ab}}\to\mathcal{W}^{\mathrm{ab}} and there is a group homomorphism ψ\psi making the following diagram commute:

According to [Hof08] Theorem 4.16 the map ψ\psi is an isomorphism. With Theorem 4.4 we have obtained the main result of this article:

Theorem 5.1

The TT-reflection homomorphism 𝒰𝒲\mathcal{U}\to\mathcal{W} is an isomorphism if and only if Tab{0}T^{\mathrm{ab}}\setminus\{0\} is 2-independent in G/2GG/2G.

Corollary 4.5 gives more information in some specific cases. In particular, it confirms the observation made in [Hof07] and [AS08] that 𝒰𝒲\mathcal{U}\to\mathcal{W} is not always injective. If nn is the rank of GG then testing for 2-dependence involves testing for linear dependence of |T{0}||T\setminus\{0\}| vectors in an n(n+1)2\frac{n(n+1)}{2}-dimensional vector space over the Galois field 𝔽2\mathbb{F}_{2}. This is more practical than testing for the existence of a so-called non-trivial integral collection according to [AS08] Theorem 5.16. This theorem also requires GG to be finitely generated, a hypothesis that we don’t require for our Theorem 5.1.

The hypotheses “free” for GG is only used to apply Theorem 4.16 of [Hof08]. We would be interested in understanding if it could be weakened to “torsion free”, “involution free” or even omitted completely.

References

  • [AAB+97] B. N. Allison, S. Azam, S. Berman, Y. Gao, and A. Pianzola. Extended affine Lie algebras and their root systems. Memoirs of the American Mathematical Society, 126(603), March 1997.
  • [AS07] S. Azam and V. Shahsanaei. Simply laced extended affine Weyl groups (a finite presentation). Publ. Res. Inst. Math. Sci., 43(2):403–424, 2007.
  • [AS08] S. Azam and V. Shahsanaei. Presentation by conjugation for A1{A}_{1}-type extended affine Weyl groups. J. Algebra, 319(4):1428–1449, 2008.
  • [Aza99] S. Azam. Extended affine Weyl groups. J. Algebra, 214(2):571–624, 1999.
  • [Aza00] S. Azam. A presentation for reduced extended affine Weyl groups. Comm. Algebra, 28(1):465–488, 2000.
  • [Hof07] G. W. Hofmann. Weyl groups with Coxeter presentation and presentation by conjugation. J. Lie Theory, 17:337–355, 2007.
  • [Hof08] G. Hofmann. The abelianizations of Weyl groups of root systems extended by abelian groups. J. Algebra, 2008. To appear.
  • [Kry00] Y. Krylyuk. On automorphisms and isomorphisms of quasi-simple Lie algebras. J. Math. Sci. (New York), 100(1):1944–2002, 2000.
  • [Loo69] O. Loos. Symmetric spaces. I: General theory. W. A. Benjamin, Inc., New York-Amsterdam, 1969.
  • [MP95] R. V. Moody and A. Pianzola. Lie Algebras With Triangular Decompositions. John Wiley and Sons, 1. edition, 1995.
  • [Sai85] K. Saito. Extended affine root systems. I. Coxeter transformations. Publ. Res. Inst. Math. Sci., 21(1):75–179, 1985.
  • [Yos04] Y. Yoshii. Root systems extended by an abelian group and their Lie algebras. J. Lie Theory, 14(2):371–394, 2004.