Real radiciality and monoreal extensions

Detailed Technical Summary: Real Radiciality and Monoreal Extensions

Problem Statement and Motivation
The paper addresses the development of an analogue to purely inseparable (radicial) extensions within the context of real algebraic geometry. In classical algebra, a field extension $K \to L$ is purely inseparable if every element's minimal polynomial has a single root in any algebraically closed field extension, which corresponds to the extension inducing a bijection on the spectrum level. However, in real algebra, formally real fields have characteristic zero, meaning all algebraic extensions are separable. Consequently, the classical notion of inseparability does not apply directly.

The authors are motivated by a geometric question: given a real algebraic variety $V$, does there exist a "biggest" real algebraic variety $W$ between $V$ and its normalization that induces a bijection on real points? While seminormalization solves this for algebraically closed fields, the paper notes (citing [6]) that such a construction fails in real algebraic geometry because real points can disappear as complex points, preventing surjectivity. Therefore, the authors shift focus from bijectivity to injectivity at the level of the real spectrum. The central problem is to define and characterize "real radicial" extensions—those where morphisms into real closed fields admit at most one lifting—and to study "monoreal" extensions, where such liftings exist and are unique.

Methodology and Definitions
The authors employ tools from real algebraic geometry, specifically the theory of real closed fields and the real spectrum ($\text{Specr}$).

1. Monoreal Extensions:

   • Definition: An irreducible polynomial $P \in K[X]$ is monoreal if it admits exactly one root in any real closed field extension of $K$. An algebraic extension $K \to L$ is monoreal if the minimal polynomial of every element in $L$ is monoreal.
   • Characterization: For finite extensions, $K \to L$ is monoreal if and only if any morphism $K \to R$ (where $R$ is a real closed field) admits exactly one lifting to $L$.
   • Properties: The authors establish that monoreal extensions are stable under composition and restriction. They prove the existence and uniqueness of a monoreal closure for any formally real field, analogous to the purely inseparable closure in classical algebra.

2. Real Radiciality:

   • Definition: An extension $K \to L$ is real radicial if any morphism $K \to R$ admits at most one lifting to $L$. This relaxes the existence condition of monoreality to focus purely on injectivity.
   • Ring Extensions: The concept is generalized to ring extensions $A \to B$. An extension is real radicial if distinct morphisms $B \to R$ remain distinct when composed with $A \to B$. This is characterized by the injectivity of the induced map on the real spectrum, $\text{Specr}(B) \to \text{Specr}(A)$.

3. Polyreal Extensions:

   • To explore factorization, the authors define polyreal extensions as those where no non-trivial element is monoreal (minimal polynomials have zero or $\ge 2$ real roots). They investigate whether algebraic extensions can be factorized into monoreal and polyreal parts.

Key Results and Contributions

• Stability and Closure (Section 1):

  • Monoreal extensions are stable under composition, restriction, and base change (Propositions 1.8, 1.9, 1.10).
  • The monoreal closure of a field $K$ exists, is unique up to isomorphism, and is the maximal monoreal extension contained in any algebraic formally real extension (Theorem 1.13).
  • A geometric description is provided for the monoreal closure of $\mathbb{R}(t)$, identifying it with a field of continuous maps $f: \mathbb{R} \to \mathbb{R} \cup \{\pm \infty\}$ whose graphs coincide with their Zariski closures up to finitely many points (Proposition 1.16).
  • The paper demonstrates that a factorization of algebraic extensions into a monoreal part followed by a polyreal part is possible and unique (Theorem 1.25), whereas the reverse order is not generally possible.

• Relation to the Real Spectrum (Section 2):

  • Bijectivity vs. Injectivity: A finite extension is monoreal if and only if the induced map on the real spectrum is a homeomorphism (bijective).
  • Real Radiciality: An algebraic extension is real radicial if and only if the induced map on the real spectrum is injective (Proposition 2.3).
  • Degree Constraints: For finite real radicial extensions, a dichotomy exists: either the extension has odd degree and is monoreal, or it has even degree and the target field is not formally real (Proposition 2.7).
  • Counterexamples: The paper provides examples showing that real radiciality is not stable under restriction (e.g., $\mathbb{Q} \to \mathbb{Q}(\sqrt{2}) \to \mathbb{Q}(i, \sqrt{2})$) and that injectivity on the real spectrum does not imply surjectivity.

• Ring Extensions and Geometric Applications (Section 3):

  • The authors characterize real radicial ring extensions via the injectivity of the map $\text{Specr}(B) \to \text{Specr}(A)$ (Theorem 3.4).
  • They investigate the relationship between injectivity on the real spectrum and injectivity on the set of real prime ideals ($R\text{-Spec}$). While injectivity on the real spectrum implies injectivity on real residue fields, it does not generally imply injectivity on real prime ideals (Remark 3.6).
  • Geometric Theorem: In the context of affine algebraic varieties over a real closed field, if a dominant finite type morphism is injective on real closed points, then the induced map on real prime ideals is injective (Proposition 3.7).
  • Main Geometric Result: For finitely generated algebras, the map on the real spectrum is injective if and only if the map on real prime ideals is injective and the residue field extensions are real radicial (Theorem 3.8).
  • Bijection under Centrality: If $A \to B$ is a finite extension of real domains, $A$ is central, and the map on the real spectrum is injective, then the map is a bijection (Proposition 3.9).

Significance and Claims
The paper claims to establish a robust framework for "real radiciality," serving as the real algebraic counterpart to classical radiciality. The primary significance lies in shifting the focus from bijectivity (which is often impossible in real geometry due to the disappearance of real points) to injectivity.

The authors demonstrate that:

1. Monoreal extensions provide a well-behaved class of extensions with a unique closure, analogous to purely inseparable extensions.
2. Real radiciality is the correct condition to ensure injectivity at the real spectrum level.
3. In the geometric context (affine varieties), injectivity at the real spectrum level is closely linked to injectivity on real prime ideals, provided the morphism behaves well on real closed points.

The paper concludes that while a "seminormalization" (a variety inducing a bijection on real points) does not generally exist in real algebraic geometry, the study of injective morphisms via real radicial extensions offers a viable and structured alternative for understanding the topology of real algebraic varieties. The work relies on standard tools from real algebra (real spectrum, real closed fields) and cites foundational works by Grothendieck [8] and others in real algebraic geometry [3, 5, 10].