Crystal growth and Witt rings (Q2639888)

Let (W,G) be a finitely generated abstract Witt ring and let \(\hat a_ n\) be the number of isometry classes of n-dimensional quadratic forms over G. The power series \(\hat z(y)=\sum \hat a_ ny^ n\) is said to be the generating function of \((W,G)\). \textit{W. Scharlau} [Acta Arith. 54, 51-59 (1989; Zbl 0693.10018)] conjectured that \(\hat z(y)\) is always a rational function with the unique pole at \(y=1\), and proved this when \((W,G)\) is torsion free or satisfies SAP. Here the conjecture is proved in the general case. The series \(\hat z(y)\) is replaced by \[ \hat Z(X,y)=\sum \hat A_ n(X)y^ n, \] where \(\hat A_ n(X)=\sum \hat f_ n(a_ 1,...,a_ r)x_ 1^{a_ 1}...x_ r^{a_ r}\) and \(\hat f_ n(a_ 1,...,a_ r)\) is the number of isometry classes of n-dimensional forms with total signature \((a_ 1,...,a_ r)\in {\mathbb{Z}}^ r\), r being the number of signatures of (W,G). Using a generalized version of \textit{D. A. Klarner}'s results on generating functions of growing mathematical crystals [Discrete Appl. Math. 3, 47-52, 113-117 (1981; Zbl 0466.05007)], the author shows that \(\hat Z(X,y)\) is a rational function with the denominator given explicitly as the product of linear factors in y. From this result Scharlau's conjecture follows.