No label defined (Q4254845)

In this thesis a number of subjects are treated which are related to association schemes. Almost all also have some relation to Chevalley groups. The similarity between these problems is that all are (just) beyond the possibility of being solved by hand computations or theoretical considerations. The solutions to the problems show that with the use of computer algebra we are able to solve such problems and, moreover, that these solutions obtained by computer computations give a deeper insight in the problems considered. In this chapter, Chapter 1, we give a summary of each of the other chapters in this thesis. We also show the connections among the chapters and the origins of the problems that are treated in them. In Chapter 2 the basic results on association schemes in connection with eigenvalues are summarized. A theorem that shows that the spectrum of a matrix in the Bose-Mesner algebra of an association scheme can be computed solely from the parameters of the association scheme, is extended to a proposition (Proposition 2.3.1) that treats all cases, where the proof of the former theorem only was valid when the matrix \(L\) of the theorem had distinct eigenvalues. We use this proposition in Chapter 5. The chapter then continues with some methods to compute spectra of matrices closely related to matrices in the Bose-Mesner algebra of an association scheme. This result is used in Chapter 6. In the last section of Chapter 2 we then summarize the theory of Terwilliger algebras. We especially direct our attention to the relation between eigenvalues of the first subconstituent of \(P\)- and \(Q\)-polynomial association schemes and the structure of irreducible modules of endpoint one, which is the main reason for doing the computations in Chapter 6. In Chapter 3 we summarize the theory of Weyl groups and Chevalley groups, as used by us for an implementation of multiplication in Chevalley groups where elements are represented as products of root group elements. Our goal was to be able to compute with Chevalley groups and the graphs of Lie type which originate from them. As it turns out the programs are presently useful only in small cases. One problem we were able to solve is discussed in Chapter 4. In Chapter 4 we introduce the Coxeter graphs and graphs of Lie type, again with an eye on computational aspects. In the last section of this chapter we are then able (with the help of a few days CPU-time on a PC) to find the structure of the graph \((F_{4,1}(\mathbb{F}_2))_d\), that is, the induced subgraph of the graph of Lie type \(F_{4,1}(\mathbb{F}_2)\) on the vertices at maximal distance from a fixed vertex. Chapter 5 then solves a question raised by F. Buekenhout, namely the spectra of the Coxeter graphs \(F_{4,1}\), \(H_{4,1}\) and \(H_{4,4}\). In addition, with the help of the theory of Chapter 2 and with the use of the mathematical computer package Maple, we compute the spectra of all Coxeter graphs and graphs of Lie type with respect to an endpoint in the diagram. These results appeared in the paper [\textit{A. E. Brouwer} and \textit{R. J. Riebeek}, The spectra of Coxeter graphs, J. Algebr. Comb. 9, No. 1, 15-28 (1998)]. Chapter 6 solves a problem raised by Hobart and Ito: we compute the spectrum of the second subconstituent of the bilinear forms graph \(H_q(d, e)\), with the help of the methods developed in Chapter 2. The interest in this graph lies in the fact that it is, in some way, the first non-thin \(P\)- and \(Q\)-polynomial association scheme with classical parameters. As we saw in Chapter 2, the spectrum of the first subconstituent of such a graph is strongly related to the structure of modules of endpoint one. The (still unanswered) question is whether the spectrum of the second subconstituent is related to the structure of irreducible modules of endpoint two. The results of this chapter do not incite to the search for obvious properties of the spectrum to the structure of the modules of endpoint two. A paper which contains the results of this chapter has appeared [see \textit{A. E. Brouwer} and \textit{Rembo Riebeek}, Eur. J. Comb. 29, No. 3, 299-305, Art. No. ej970195 (1998; Zbl 0901.05091)]. In Chapter 7 a new bipartite distance-regular graph with intersection array \(\{45, 44, 36, 5;1, 9, 40, 45\}\) is constructed. Its halved graph is the complement of the strongly regular Berlekamp-van Lint-Seidel graph. This solves a `leftover' of my master's thesis. The structure of this new graph and relations to other well-known combinatorial structures are discussed. The results of this chapter have been published in the paper [\textit{A. E. Brouwer}, \textit{J. H. Koolen} and \textit{R. J. Riebeek}, J. Algebr. Comb. 8, No. 2, 153-156 (1998)].