A reduction functor, tameness, and Tits form for a class of orders (Q1895566)

The author discusses the representation type -- finite and tame -- of a special class of orders. Let \(D\) be a complete discrete valuation domain with maximal ideal \(\mathfrak p\), the field of fractions \(K\) of \(D\) is assumed to be algebraically closed. A \((0,1)\)-tiled \(D\)-order is of the form \(\Lambda(n) := ({\mathfrak p}^{m_{i,j}})_n\), where \(m_{i,i}=0\), \(m_{i,j}=1\) for \(i>j\) and \(m_{i,j} \in \{0,1\}\) for \(i<j\). If \(n_1 \leq n_2\) the notation \(\Lambda_{n_1} \leq \Lambda_{n_2}\) is used provided \(\Lambda_{n_1}=\varepsilon_{n_1} \cdot \Lambda_{n_2} \cdot \varepsilon_{n_1}\) for \(\varepsilon_{n_1}=\sum^{n_1}_{i=1} e_{i,i}\), where \(e_{i,j}\) is the usual matrix unit. Given now three \((0,1)\)-tiled orders \(\Lambda_{n_i}\), \(0\leq i\leq 2\) with \(0<n_0 \leq n_1 \leq n_2\) with \(\Lambda_0 \leq \Lambda_i\), \(i=1,2\), then the \(\Lambda_0\)-glueing of \(\Lambda_1\) with \(\Lambda_2\) is defined as \[ \Lambda_1 \triangleleft_{\Lambda_0} \Lambda_2 := \bigl \{\lambda=\left( \begin{smallmatrix} \lambda_1 & x \\ y & \lambda_2 \end{smallmatrix} \right) \in \left( \begin{smallmatrix} \Lambda_1 & (D)_{n_1 \times n_2} \\ ({\mathfrak p})_{n_2 \times n_1} & \Lambda_2 \end{smallmatrix} \right):\varepsilon_{n_0} \cdot \lambda_1 \cdot \varepsilon_{n_0}=\varepsilon_{n_0} \cdot \lambda_2 \cdot \varepsilon_{n_0}\bigr\}. \] With this order \(\Lambda := \Lambda_1 \triangleleft_{\Lambda_0} \Lambda_2\) the author associates the reduced Tits quadratic form, a rational quadratic form: \(q_\Lambda:\mathbb{Q}^{n_1+n_2 - n_0+2} \to \mathbb{Q}\). The main results concerning the above order \(\Lambda\) are: Theorem 1. \(\bullet\) \(\Lambda\) is of finite lattice type if and only if its Tits quadratic form is weakly positive. \(\bullet\) If \(\lambda\) is of tame representation type, then the form \(q_\Lambda\) is weakly non-negative. The converse holds if \(\Lambda_0\) is hereditary and \(\Lambda_i\), \(i=1,2\) are of finite lattice type. Theorem 2. Assume that in the definition of \(\Lambda\) we have \(n_0=n_1=n_2\). Then \(\bullet\) \(\Lambda\) is of finite lattice type, if and only if \(\Lambda_0\) is hereditary. \(\bullet\) The following conditions are equivalent: -- \(\Lambda\) is of tame lattice type. -- The Tits quadratic form is weakly non-negative. -- For each triple \(\{i, j, k\}\) of different elements in \(\{1, \dots, n_0\}\) the order \(\eta \cdot \Lambda_0 \cdot \eta\) with \(\eta=e_{i,i}+e_{j,j}+e_{k,k}\) is not isomorphic to one of the orders \[ \left( \begin{smallmatrix} D & {\mathfrak p} & {\mathfrak p}\\ {\mathfrak p} & D & {\mathfrak p}\\ {\mathfrak p} & {\mathfrak p} & D \end{smallmatrix} \right), \;\left(\begin{smallmatrix} D & {\mathfrak p} & D \\ {\mathfrak p} & D & {\mathfrak p} \\ {\mathfrak p} & {\mathfrak p} & D \end{smallmatrix} \right),\;\left( \begin{smallmatrix} D & D & {\mathfrak p} \\ {\mathfrak p} & D & {\mathfrak p}\\ {\mathfrak p} & {\mathfrak p} & D \end{smallmatrix} \right),\;\left(\begin{smallmatrix} D & {\mathfrak p} & {\mathfrak p} \\ {\mathfrak p} & D & D \\ {\mathfrak p} & {\mathfrak p} & D \end{smallmatrix} \right), \] \(\bullet\) The category of \(\Lambda\)-lattices is of polynomial growth if and only if there exists at most one pair \((i, j)\) of indices with \(i < j\) and \(m_{i,j} = 1\). The proof is based on a reduction theory to socle projective representations -- introduced by \textit{C. M. Ringel} and the reviewer [J. Algebra 60, 11-42 (1979; Zbl 0438.16021)] -- of posets.