[Paper Review] Non-zero integral friezes

Bruce Fontaine|arXiv (Cornell University)|Sep 21, 2014

Algebraic structures and combinatorial models3 references3 citations

TL;DR

This paper extends Coxeter-Conway friezes by allowing non-zero integers (positive and negative) instead of just positive integers, establishing that for Dynkin types $A_n$, $B_n$, $C_n$, $D_n$, and $G_2$, the number of non-zero integral friezes is 1, 2, or 4 times the number of positive friezes. The key contribution is a complete classification of these friezes via group actions and symmetry, revealing structural connections to cluster algebras and Galois invariance over rings of integers.

ABSTRACT

We study non-zero integral friezes for Dynkin types $A_n$, $B_n$, $C_n$, $D_n$ and $G_2$. These differ from standard Coxeter-Conway (positive) friezes by allowing any non-zero integer to appear. In each case we show that there are either $1$, $2$ or $4$ times as many non-zero friezes as positive friezes. This is a first step for considering friezes over general rings of integers.

Motivation & Objective

• To generalize positive friezes (Coxeter-Conway) by allowing non-zero integers, enabling study of friezes over general rings of integers.
• To determine the number of non-zero integral friezes for each Dynkin type $A_n$, $B_n$, $C_n$, $D_n$, and $G_2$.
• To understand the role of symmetries and Galois actions in classifying these friezes, particularly via involutions and $G$-invariance.
• To lay foundational groundwork for studying friezes over rings like $\mathbb{Z}[\zeta]$ for roots of unity $\zeta$.

Proposed method

• Use of involutions (e.g., $\sigma$, $\sigma_1$, $\sigma_2$) to relate non-positive to positive friezes, showing that sign flips preserve admissibility under certain conditions.
• Lift non-zero integral friezes of type $C_n$, $B_n$, $G_2$ to $G$-invariant friezes on $A_{2n-1}$, $D_{n+1}$, and $D_4$, respectively, using group actions.
• Apply Galois descent: $G$-invariant $\Delta$-friezes descend to $\Delta/G$-friezes, and non-zero integral $\Delta/G$-friezes lift to $G$-invariant $\Delta$-friezes.
• Use of triangulation labelings with $\pm1$ boundary states and Ptolemy relations to characterize admissible labelings and their extensions.
• Leverage the Laurent phenomenon and positivity of cluster algebra exchange relations to ensure integer outputs under non-positive assignments.
• Analyze the action of the group $G$ (generated by $\sigma_1$, $\sigma_2$) on cluster variables, particularly distinguishing between $G$-invariant and non-invariant friezes.

Experimental results

Research questions

• RQ1 How many non-zero integral friezes exist for each Dynkin type $A_n$, $B_n$, $C_n$, $D_n$, and $G_2$?
• RQ2 What is the relationship between the number of non-zero integral friezes and the number of positive friezes in each Dynkin type?
• RQ3 Which symmetries (e.g., sign-flip involutions) preserve admissibility of frieze labelings, and how do they affect classification?
• RQ4 Can non-zero integral friezes over $\mathbb{Z}[\zeta]$ for roots of unity $\zeta$ be classified, and are there finitely many?

Key findings

• For $A_n$, the number of non-zero integral friezes is twice the number of positive friezes when $n$ is odd, and equal when $n$ is even.
• For $B_n$ and $C_n$, there are exactly twice as many non-zero integral friezes as positive friezes.
• For $D_n$, there are four times as many non-zero integral friezes as positive ones when $n$ is even, and twice as many when $n$ is odd.
• For $G_2$, there are exactly 9 non-zero integral friezes, all of which are positive.
• The classification relies on lifting to $G$-invariant friezes on $A_{2n-1}$, $D_{n+1}$, or $D_4$, and using group actions to count distinct non-positive friezes.
• The results suggest that non-zero integral friezes over rings like $\mathbb{Z}[\zeta]$ may be finite, supporting the conjecture that such friezes are finite for any root of unity $\zeta$.