Surreal numbers, derivations and transseries (Q1708554)

This article concerns Conway's real closed field of surreal numbers \textbf{No} [\textit{J. H. Conway}, On numbers and games. London: Academic Press (1976; Zbl 0334.00004)]. Recall that these numbers, which encompass all at once real numbers and ordinal numbers, can also be viewed as (bounded) generalized power series in the sense of [\textit{H. Hahn}, Wien. Ber. 116, 601--655 (1907; JFM 38.0501.01)]. Moreover, after [\textit{H. Gonshor}, An introduction to the theory of surreal numbers. Cambridge: Cambridge University Press (1986; Zbl 0595.12017)], \textbf{No} comes equipped with Gonshor's extension of the real exponentiation. Hence, there remained an important pending question to complete this real analytic approach: could \textbf{No} be endowed with a compatible derivation? This is what the authors achieve in the present article: \textbf{Theorem} (Theorem A, Theorem 6.30). The field \textbf{No} admits a derivation \(D :\textbf{No}\rightarrow\textbf{No}\) satisfying the following properties: (1) Leibniz' rule: \(D(xy) = xD(y) + yD(x)\); (2) strong additivity: \(D\left(\sum_{i\in I} x_i\right)=\sum_{i\in I}D(x_i)\) if \((x_i : i \in I)\) is summable; (3) compatibility with exponentiation: \(D(\text{exp}(x)) = \text{exp}(x)D(x)\); (4) constant field \(\mathbb R\): \(\ker(D) = \mathbb R\); (5) \(H\)-field: if \(x > N\), then \(D(x) > 0\). The notion of \(H\)-field is an abstract version of that of Hardy fields, due to \textit{M. Aschenbrenner} and \textit{L. van den Dries} [Math. Z. 242, No. 3, 543--588 (2002; Zbl 1066.12002)]. In fact, they obtain the following stronger result: \textbf{Theorem} (Theorem B, Theorem 7.7). The field \textbf{No} of surreal numbers equipped with \(D\) is a Liouville closed H-field with small derivation in the sense of [loc. cit.], that is, \(D\) is surjective and sends infinitesimals to infinitesimals. This was subsequently used in [\textit{M. Aschenbrenner} et al., J. Eur. Math. Soc. (JEMS) 21, No. 4, 1179--1199 (2019; Zbl 1470.12004)] to show that \textbf{No} is a universal domain for \(H\)-fields, in particular for Hardy fields. To obtain their main result, the authors resume and extend results of the PhD thesis of Schmeling (see also [\textit{J. van der Hoeven}, Transseries and real differential algebra. Berlin: Springer (2006; Zbl 1128.12008)]): \textbf{Theorem} (Theorem C, Theorem 8.10). \textbf{No} is a field of transseries in the sense of Schmeling. More precisely, they proceed by: \begin{itemize} \item writing surreal numbers in a so-called \textit{Ressayre normal form} to reduce to a smaller subclass of \textbf{No}; \item identifying the subclass of \textit{log-atomic numbers} as \textit{\(\lambda\)-numbers} (a generalization of the \(\kappa\)-numbers of [\textit{S. Kuhlmann} and \textit{M. Matusinski}, Order 32, No. 1, 53--68 (2015; Zbl 1388.12013)]) and as \textit{levels} in the sense of [\textit{D. Marker} and \textit{C. Miller}, Rev. Mat. Univ. Complutense Madr. 10, 241--249 (1997; Zbl 0879.03012)]; \item providing a formula for the derivatives of the latter; \item proving the so-called axiom (T4) of Schmeling and exploiting it to extend the derivation formula to the whole of \textbf{No}. \end{itemize} One of their key ingredients is what they call the \textit{nested truncation rank} which allows them to prove by induction the important and difficult summability questions. We advise the reader to go through the very informative detailed introduction of the paper.