Domination, fibrations and splitting

This article is concerned with finite rank stability theory, and more precisely two classical ways to decompose a type using minimal types. The first is its domination equivalence to a Morley power of minimal types, and the second its semi-minimal analysis, both of which are useful in applications. Our main interest is to explore how these two decompositions are connected. We prove that neither determine the other in general, and give more precise connections using various notions from the model theory literature such as uniform internality, proper fibrations and disintegratedness.

💡 Research Summary

This paper investigates the relationship between two classical decompositions of a finite‑rank stationary type in a superstable theory: domination‑decomposition and semi‑minimal analysis. Domination‑decomposition expresses a type p as domination‑equivalent to a Morley product of minimal types, while semi‑minimal analysis builds a chain of definable maps f with fibers that are almost internal to minimal types. The authors ask whether one decomposition determines the other and find that, in general, it does not.

The central results are threefold. First, Theorem 3.12 (Theorem A) shows that if a fiber of a definable map f from p to f(p) is almost internal to a minimal type r, then p is domination‑equivalent to f(p) ⊗ r^{(m)} for some m ≤ U‑rank of that fiber. Moreover, m equals the full U‑rank precisely when f is uniformly almost r‑internal, i.e., all fibers share the same internality parameters. This links uniform internality to the size of the Morley power appearing in the domination‑decomposition.

Second, the authors establish a dichotomy for the value of m when the fibers have no proper fibrations (Definition 3.14). In this situation either m = 0 (the fiber contributes nothing to the domination product) or m equals the full U‑rank of the fiber. Lemma 3.15 provides a construction that, given a domination of a minimal type, yields a definable map whose image is an R‑internal type, where R is the family of A‑conjugates of the minimal type. Using this, Theorem 3.19 identifies the U‑rank of each minimal component in the domination‑decomposition with the U‑rank of the corresponding R‑reduction of p. This refines Buechler’s level theory by giving a type‑by‑type description independent of the original level arguments.

Third, Theorem 4.4 (Theorem B) treats the case where each fiber is almost internal to a disintegrated minimal type (a minimal type whose induced pregeometry is trivial). It proves a sharp dichotomy: either any two distinct fibers are orthogonal (hence completely independent), or there exists a minimal disintegrated type r such that p is inter‑algebraic with f(p) ⊗ r^{(n)}, where n is the U‑rank of any fiber. This phenomenon underlies recent work of Freitag and Nagloo on relations among solutions of Painlevé equations.

The paper also discusses when the minimal types appearing in a domination‑decomposition can be taken over the same parameter set A. While this fails in general (e.g., Hrushovski’s ℵ₁‑categorical example without minimal types over acl^{eq}(∅)), it holds in important theories such as DCF₀ (differentially closed fields of characteristic 0) and CCM (compact complex manifolds).

Overall, the authors provide a systematic comparison of domination‑decomposition and semi‑minimal analysis, showing that each captures different aspects of a type’s structure. Uniform almost internality, the absence of proper fibrations, and disintegratedness serve as precise bridges between the two decompositions, yielding concrete criteria for when one can read off the minimal components of a domination‑decomposition from a semi‑minimal analysis and vice versa. The results have immediate applications to the model‑theoretic study of differential equations, especially in computing r‑reductions for generic solutions of rational differential systems. Future work may extend these ideas to simple theories and develop effective algorithms for r‑reduction in concrete settings.