← Back to arXiv
arXivLogicarXiv:2608.14033

Definability of Hausdorff Limits for Lipschitz Cells in O-minimal Structures

The paper tackles a question about geometric limits of families of sets defined by mathematical formulas. In standard real analysis, if you have a continuously varying collection of sets, each described by the same type of formula, and you look at what shape those sets converge to (in the sense of Hausdorff distance, which measures how far apart two sets are in space), the limiting set is also describable by a formula of the same logical type. This was established by Lou van den Dries using a framework called o-minimality, which is a powerful setting for "tame" geometry where definable sets cannot be too wild. The authors want to extend this result beyond the ordinary real numbers to more exotic mathematical universes called non-Archimedean real closed fields, which contain infinitely large and infinitely small numbers.

The challenge is that the classical proof relies on the fact that closed and bounded regions of ordinary Euclidean space are compact, meaning every sequence has a convergent subsequence. This compactness simply fails in non-Archimedean settings. The authors work around this by restricting attention to a well-behaved class of sets called Lipschitz cells, which are sets whose boundaries cannot oscillate too wildly, with a uniform bound on how steep they can be. They replace compactness arguments with a construction called dense completions and track convergence through very long sequences indexed by elements of the field itself. They also reinterpret Hausdorff distance as taking values in an ordered completion rather than just real numbers.

The main result is that, under these conditions, the Hausdorff limit of such a family still exists and is definable within the dense completion of the base model. A key ingredient is a model-theoretic property called stable embeddedness of tame pairs, which ensures that logical definability is preserved when passing between different levels of the construction. The authors also prove a uniform version: not only does each individual limit set have a formula describing it, but the entire collection of limit sets across all parameter values is itself a definable family. This generalizes the classical theorem to a much broader and more algebraically exotic context.

Read original →