On the uniqueness of maximal immediate extensions of valued differential fields (Q1628495)

Recall the following well-known theorems about an equicharacteristic zero valued field \(K\): Theorem A'. Any two maximal immediate extensions of \(K\) are isomorphic over \(K\), and any two algebraically maximal immediate extensions of \(K\) are isomorphic over \(K\). Theorem B'. If \(K\) is henselian, then \(K\) is algebraically maximal. The paper under review establishes analogues of these theorems in the category of valued differential fields with small derivation, with an additional assumption on the value group \(\Gamma\) of \(K\). For relevant definitions, see [\textit{M. Aschenbrenner} et al., Asymptotic differential algebra and model theory of transseries. Princeton, NJ: Princeton University Press (2017; Zbl 1430.12002)]: Theorem A. If the differential residue field \(\mathbf{k}\) is linearly surjective and \(\Gamma\) is the union of its finite rank convex subgroups, then any two maximal immediate extensions of \(K\) are isomorphic over \(K\), and any two differential-algebraically maximal differentially algebraic immediate extensions of \(K\) are isomorphic over \(K\). Theorem B. If \(K\) is differential-henselian, has few constants, and \(\Gamma\) is the union of its finite rank convex subgroups, then \(K\) is differential-algebraically maximal. The key idea for these theorems involves isolating and establishing the so-called differential-henselian configuration property. This is analogous to the key idea for Theorem A' which essentially states that when you view a pseudocauchy sequence along a minimal polynomial, eventually the degree \(1\) term dominates. They also establish the first known result for differential-henselizations: Theorem C. Suppose \(K\) is asymptotic, \(\mathbf{k}\) is linearly surjective, and \(\Gamma\) has the differential-henselian configuration property. Then \(K\) has a differential-henselization, and any two differential-henselizations of \(K\) are isomorphic over \(K\). Reviewer comment: The paper actually proves more than what is stated in Theorem A. and Theorem B. By Proposition 3.1 and Lemma 3.4, it follows that every \(\Gamma\) whose archimedean classes are well-ordered has the differential-henselian configuration property. Thus ``is the union of its finite rank convex subgroups'' can be replaced with ``has well-ordered archimedean classes'' in those theorems. Furthermore, the condition on \(\Gamma\) in Theorem C. can be completely removed. This is shown in the second author's thesis.