Remarks and problems about algorithmic descriptions of groups
Emmanuel Rauzy
TL;DR
The work develops a unified, lattice- and numbering-based framework for global decision problems in finitely generated groups, bridging local descriptions (word problem) and global descriptions (finite presentations) via multiple computability lenses. It proves Rice–Shapiro-type results for recursive presentations and a Rice theorem for co-recursive presentations, and establishes an algorithmic characterization of finitely presented groups through the equivalence $\nu_{FP} \equiv \nu_{RP} \wedge \nu_{MQA}$, with the compactness of FP groups linked to Scott topology on marked groups. The paper further analyzes marked quotient algorithms, both in the absolute setting and relative to classes of groups, and demonstrates how Marked Quotient content interacts with finite presentations and group varieties, including the lamplighter example. It also discusses Adian–Rabin results as incomplete answers within this framework, outlining how the topology and hierarchy perspectives illuminate undecidability and classification limits in group theory. Overall, the work offers a rigorous, extensible program for algorithmic descriptions of groups beyond traditional finite presentations, with implications for meta-decision problems and the structure of algorithmic group theory.
Abstract
Motivated by a theorem of Groves and Wilton, we propose the study of the lattice of numberings of isomorphism classes of marked groups as a rigorous and comprehensive framework to study global decision problems for finitely generated groups. We establish the Rice and Rice-Shapiro Theorems for recursive presentations, and establish similar results for co-recursive presentations. We give an algorithmic characterization of finitely presentable groups in terms of semi-decidability of two decision problems: the word problem and the marked quotient problem, which we introduce. We explain how this result can be used to define algorithmic generalizations of finite presentations. Finally, we discuss how the Adian-Rabin Theorem provides incomplete answers in several respects.
