The spectrum of a Coxeter transformation, affine Coxeter transformations, and the defect map (Q1822600)

From MaRDI portal
scientific article
Language Label Description Also known as
English
The spectrum of a Coxeter transformation, affine Coxeter transformations, and the defect map
scientific article

    Statements

    The spectrum of a Coxeter transformation, affine Coxeter transformations, and the defect map (English)
    0 references
    0 references
    1989
    0 references
    Let A be an \(\ell \times \ell\) matrix over \(K={\mathbb{R}}\) or \({\mathbb{C}}\) satisfying (i) \(A_{ii}=2\), (ii) \(A_{ij}=0\) iff \(A_{ji}=0\). Assume that W is a reflection group acting on a vector space V over K with a basis \(\{\alpha_ 1,...,\alpha_{\ell}\}\) and that W is generated by the reflections \(\{r_ 1,...,r_{\ell}\}\) defined by \(r_ i\alpha_ j=\alpha_ j-A_{ij}\alpha_ i\). Let \(\Gamma_ 0\) be the primitive graph of A, i.e. \(\Gamma_ 0\) is the graph with \(\ell\)-nodes in which the ith and jth nodes are joined by an edge iff \(A_{ij}\neq 0\). Assume that the graph has no odd cycles. Then \(\Gamma_ 0\) can be partitioned into a disjoint union of sets \(\Sigma_ 1\), \(\Sigma_ 2\) such that \(r_ ir_ j=r_ jr_ i\) if i,j are both in \(\Sigma_ 1\) or \(\Sigma_ 2\). Let \(R_ k\), \(k=1,2\) be the product of the reflections coming from \(\Sigma_ 1\), \(\Sigma_ 2\) and set \(R=R_ 1R_ 2\). R is called the Coxeter transformation of W. If A is the Cartan matrix associated to a finite-dimensional simple Lie algebra, then in fact all products of the distinct simple reflections lie in the same conjugacy class; this is false in general. In this paper, the authors prove that if the set of eigenvalues of R are \(\{e^{\chi_ 1},...,e^{\chi_{\ell}}\}\), then the eigenvalues of A are \(\{\) 4 \(cosh^ 2(\chi_ 1/4),...,4 \cosh^ 2(\chi_{\ell}/4)\}\). As a simple corollary one deduces that R is of finite order iff A is of finite type. The observation that \(spec(2-A)=spec(A-2)\) proves that the eigenvalues of A occur in pairs \(\{\) \(\lambda\),4-\(\lambda\) \(\}\). Thus the eigenvectors occur in pairs as well. If A is an affine Cartan matrix, then 0 is an eigenvalue of A. The eigenvector corresponding to the \(\lambda =4\) eigenvalue arises naturally in the study of the defect map which is defined as follows. Let \(\delta\) be the element of V which is fixed by W. Then W induces a group of transformations \(\bar W\) on V/R\(\delta\). \(\bar W\) is the corresponding finite Weyl group and the image \(\bar R\) of R has finite order say m. Let \({\tilde \delta}\) be the map \(1-R^ m: V\to K\delta\). \({\tilde \delta}\) is called the defect map and the authors give a description of the matrix of this map.
    0 references
    0 references
    Coxeter transformation
    0 references
    Cartan matrix
    0 references
    eigenvalues
    0 references
    affine Cartan matrix
    0 references
    defect map
    0 references
    Weyl group
    0 references
    0 references