Algebraically Closed Metric Valued Fields

• Algebraically closed metric valued fields are non-Archimedean fields whose ultrametric is defined by a valuation, ensuring model-theoretic tameness and clear geometric structure.
• They support quantifier elimination and C-minimality across multi-sorted languages, allowing definable sets to be characterized precisely as combinations of balls, cosets, and lattices.
• Effective computability aspects, including valuation extension and the Mal’cev phenomena, provide practical insights into the computational model theory of such fields.

An algebraically closed metric valued field is equivalently an algebraically closed non-Archimedean valued field equipped with the ultrametric induced by its valuation. In the standard ACVF setting, one considers a field $K$ with a valuation $v:K\to \Gamma\cup\{\infty\}$, valuation ring $\mathcal{O}=\{x\in K:v(x)\ge 0\}$, maximal ideal $M=\{x:v(x)>0\}$, residue field $k=\mathcal{O}/M$, and basic balls $B(a,\gamma)=\{x:v(x-a)\ge \gamma\}$ and $B^\circ(a,\gamma)=\{x:v(x-a)>\gamma\}$. The metric is definable from the valuation, and ACVF is the model completion of the theory of valued fields, admitting quantifier elimination and the structural analysis of definable sets, definable types, imaginaries, and effective presentations that make the subject central both in model theory and in metric formulations of valued fields (Hrushovski, 2014).

1. Valuation-theoretic and ultrametric foundations

A valued field is a field $K$ together with a valuation $v:K\to \Gamma\cup\{\infty\}$ into an ordered abelian group $\Gamma$ such that $v:K\to \Gamma\cup\{\infty\}$0 iff $v:K\to \Gamma\cup\{\infty\}$1, $v:K\to \Gamma\cup\{\infty\}$2, and $v:K\to \Gamma\cup\{\infty\}$3, with equality if $v:K\to \Gamma\cup\{\infty\}$4. The valuation is assumed surjective, $v:K\to \Gamma\cup\{\infty\}$5 is the value group, $v:K\to \Gamma\cup\{\infty\}$6 is the valuation ring, $v:K\to \Gamma\cup\{\infty\}$7 its maximal ideal, and $v:K\to \Gamma\cup\{\infty\}$8 the residue field. This is the basic non-Archimedean setup underlying both the algebraic and the metric viewpoints (Harrison-Trainor, 2016).

The valuation induces an ultrametric structure. Given a strictly order-preserving embedding $v:K\to \Gamma\cup\{\infty\}$9, one defines $\mathcal{O}=\{x\in K:v(x)\ge 0\}$0 for $\mathcal{O}=\{x\in K:v(x)\ge 0\}$1, $\mathcal{O}=\{x\in K:v(x)\ge 0\}$2, and $\mathcal{O}=\{x\in K:v(x)\ge 0\}$3. Equivalently, one may define the metric directly from valuation balls. Closed and open balls are definable by valuation inequalities, and in discretely valued situations one often writes $\mathcal{O}=\{x\in K:v(x)\ge 0\}$4. The non-Archimedean triangle inequality becomes $\mathcal{O}=\{x\in K:v(x)\ge 0\}$5, so the metric neighborhoods are precisely the valuation balls (Kovacsics et al., 2018).

In ACVF, the valuation is nontrivial, and the algebraic closure is taken in the valued-field sense used by the theory. The metric therefore does not add new primitive structure: it is definable from $\mathcal{O}=\{x\in K:v(x)\ge 0\}$6, and conversely the model-theoretic treatment of the metric sort is controlled by valuation inequalities. A crucial limitation is that metric completeness is not part of ACVF per se: when $\mathcal{O}=\{x\in K:v(x)\ge 0\}$7 is algebraically closed and maximally complete, the induced ultrametric space is complete, but the general theory ACVF does not impose maximal completeness (Yin, 2010).

2. Languages, quantifier elimination, and minimality

Two language choices dominate the literature: a three-sorted language with field, residue-field, and value-group sorts, and the two-sorted $\mathcal{O}=\{x\in K:v(x)\ge 0\}$8 language centered on the quotient $\mathcal{O}=\{x\in K:v(x)\ge 0\}$9. In the three-sorted framework one has sorts $M=\{x:v(x)>0\}$0, $M=\{x:v(x)>0\}$1, and $M=\{x:v(x)>0\}$2, with function symbols $M=\{x:v(x)>0\}$3, $M=\{x:v(x)>0\}$4, and a predicate for the valuation ring. In $M=\{x:v(x)>0\}$5 one has the field sort $M=\{x:v(x)>0\}$6, the $M=\{x:v(x)>0\}$7-sort, and the map $M=\{x:v(x)>0\}$8; the induced map $M=\{x:v(x)>0\}$9 recovers the valuation (Hrushovski, 2014, Yin, 2010).

Framework	Basic data	Principal result
Three-sorted ACVF	$k=\mathcal{O}/M$0, $k=\mathcal{O}/M$1, $k=\mathcal{O}/M$2, $k=\mathcal{O}/M$3, $k=\mathcal{O}/M$4	Quantifier elimination; $k=\mathcal{O}/M$5 and $k=\mathcal{O}/M$6 stably embedded
$k=\mathcal{O}/M$7	$k=\mathcal{O}/M$8, $k=\mathcal{O}/M$9, $B(a,\gamma)=\{x:v(x-a)\ge \gamma\}$0, induced $B(a,\gamma)=\{x:v(x-a)\ge \gamma\}$1	Quantifier elimination; ACVF is C-minimal
$B(a,\gamma)=\{x:v(x-a)\ge \gamma\}$2, $B(a,\gamma)=\{x:v(x-a)\ge \gamma\}$3	Add a section $B(a,\gamma)=\{x:v(x-a)\ge \gamma\}$4 of $B(a,\gamma)=\{x:v(x-a)\ge \gamma\}$5 or a section $B(a,\gamma)=\{x:v(x-a)\ge \gamma\}$6 of $B(a,\gamma)=\{x:v(x-a)\ge \gamma\}$7	Quantifier elimination in the expanded languages

ACVF eliminates quantifiers in these standard languages. In the three-sorted setting this yields the reduction of mixed formulas to valuation and residue conditions; in particular, $B(a,\gamma)=\{x:v(x-a)\ge \gamma\}$8 and $B(a,\gamma)=\{x:v(x-a)\ge \gamma\}$9 are stably embedded, and definable subsets of $B^\circ(a,\gamma)=\{x:v(x-a)>\gamma\}$0 or $B^\circ(a,\gamma)=\{x:v(x-a)>\gamma\}$1 are definable over parameters from those sorts alone (Hrushovski, 2014). In the $B^\circ(a,\gamma)=\{x:v(x-a)>\gamma\}$2 setting, Yimu Yin proved that ACVF admits quantifier elimination, and the same holds for the expansions $B^\circ(a,\gamma)=\{x:v(x-a)>\gamma\}$3 and $B^\circ(a,\gamma)=\{x:v(x-a)>\gamma\}$4 obtained by adding a section of the whole $B^\circ(a,\gamma)=\{x:v(x-a)>\gamma\}$5-sort or of the residue field (Yin, 2010).

Minimality phenomena organize the field-sort geometry. ACVF is C-minimal: every definable subset of the field sort is a boolean combination of balls. In pure characteristic $B^\circ(a,\gamma)=\{x:v(x-a)>\gamma\}$6, ACVF$B^\circ(a,\gamma)=\{x:v(x-a)>\gamma\}$7 is also b-minimal, and so are the section expansions $B^\circ(a,\gamma)=\{x:v(x-a)>\gamma\}$8 and $B^\circ(a,\gamma)=\{x:v(x-a)>\gamma\}$9. The expansions are not C-minimal, because sections allow definable choices of representatives across $K$0-cosets or residue classes. Local C-minimality separates them more finely: $K$1 is locally C-minimal, while $K$2 is not (Yin, 2010).

These results explain why the metric field sort remains geometrically tractable despite the non-Archimedean setting. In unexpanded ACVF, the definable subsets of $K$3 are assembled from balls and $K$4-fibers, whereas expansions by sections increase definable choice and thereby alter local ball geometry.

3. Definable types, stable domination, and imaginaries

A definable type in ACVF is a family of Boolean retractions $K$5 compatible with auxiliary variables, and it behaves functorially under definable maps by pushforward and under products by the tensor construction $K$6. Orthogonality to $K$7 means that for every definable map into the value group, the pushforward of the type is constant. Domination formalizes the principle that a type on $K$8 can be controlled by data in $K$9 or $v:K\to \Gamma\cup\{\infty\}$0 through a definable map (Hrushovski, 2014).

The central equivalence is that, for definable types in ACVF, stable domination is equivalent to orthogonality to $v:K\to \Gamma\cup\{\infty\}$1. It is also equivalent to symmetry of the type and to commuting under tensor product with every definable type. Thus the stably dominated part of ACVF is precisely the part that does not move the value group and is controlled by residue-field data (Hrushovski, 2014).

Basic examples already exhibit the metric content. The generic type of $v:K\to \Gamma\cup\{\infty\}$2, obtained from the residue map $v:K\to \Gamma\cup\{\infty\}$3 and the generic type of $v:K\to \Gamma\cup\{\infty\}$4, is a complete definable type on $v:K\to \Gamma\cup\{\infty\}$5, orthogonal to $v:K\to \Gamma\cup\{\infty\}$6, with the valuation additivity profile

$v:K\to \Gamma\cup\{\infty\}$7

for $v:K\to \Gamma\cup\{\infty\}$8 realizing the type. Closed and open balls are definable, and their generic types are stably dominated when the radius parameter is fixed modulo $v:K\to \Gamma\cup\{\infty\}$9. This makes the basic metric neighborhoods themselves carriers of model-theoretically canonical generic behavior (Hrushovski, 2014).

Elimination of imaginaries is achieved via geometric sorts. The sorts

$\Gamma$0

code lattices and lattice-plus-residue-coset data. Lemma 6.1 shows that definable $\Gamma$1-submodules of $\Gamma$2, their cosets, finite subsets of $\Gamma$3, and certain valuation-defined unipotent cosets are coded in the geometric sorts $\Gamma$4. The main theorem then states that ACVF eliminates imaginaries in $\Gamma$5 (Hrushovski, 2014).

The decomposition theorem for definable types sharpens this picture. A definable type $\Gamma$6 on a variety $\Gamma$7 decomposes into a definable type $\Gamma$8 on $\Gamma$9 and an $v:K\to \Gamma\cup\{\infty\}$00-germ $v:K\to \Gamma\cup\{\infty\}$01 of pro-definable maps into the stable completion $v:K\to \Gamma\cup\{\infty\}$02, with $v:K\to \Gamma\cup\{\infty\}$03. Up to generic reparametrization, the pair $v:K\to \Gamma\cup\{\infty\}$04 is canonical. In this sense, stably dominated definable types are the model theoretic incarnation of a Berkovich point (Hrushovski, 2014).

4. Definable subsets and functions in the metric field sort

The C-minimality of ACVF implies that definable subsets of the field sort are boolean combinations of balls. In $v:K\to \Gamma\cup\{\infty\}$05, one also has $v:K\to \Gamma\cup\{\infty\}$06-balls $v:K\to \Gamma\cup\{\infty\}$07, and the analysis of definable sets in products proceeds through polydiscs, $v:K\to \Gamma\cup\{\infty\}$08-polydiscs, and special bijections that convert definable sets into deformed $v:K\to \Gamma\cup\{\infty\}$09-pullbacks (Yin, 2010).

For definable functions in C-minimal expansions of ACVF, Cubides Kovacsics and Delon established two factorization principles over the value group. Type (I) asserts that if $v:K\to \Gamma\cup\{\infty\}$10 is definable near infinity, then there is a definable $v:K\to \Gamma\cup\{\infty\}$11 such that, outside a sufficiently large ball, $v:K\to \Gamma\cup\{\infty\}$12 depends only on $v:K\to \Gamma\cup\{\infty\}$13. Type (II) asserts that if $v:K\to \Gamma\cup\{\infty\}$14 is a definable local C-isomorphism, then outside a finite subset it locally factorizes over $v:K\to \Gamma\cup\{\infty\}$15: on sufficiently small balls, $v:K\to \Gamma\cup\{\infty\}$16 depends only on $v:K\to \Gamma\cup\{\infty\}$17. Both statements admit uniform versions for definable families (Kovacsics et al., 2018).

Under an additional asymptotic hypothesis on unary $v:K\to \Gamma\cup\{\infty\}$18-definable functions—eventual $v:K\to \Gamma\cup\{\infty\}$19-linearity—these factorizations lift to the $v:K\to \Gamma\cup\{\infty\}$20-sort. The paper formulates an RV-factorization at infinity and a local RV-factorization for families of local C-isomorphisms, with finite definable partitions, integral exponents, and definable $v:K\to \Gamma\cup\{\infty\}$21-valued parameters. Under definable completeness and eventual linearity, the corresponding limiting values exist and recover the $v:K\to \Gamma\cup\{\infty\}$22-parameters (Kovacsics et al., 2018).

In value-group $v:K\to \Gamma\cup\{\infty\}$23, this leads to uniform polynomial boundedness: for every definable family $v:K\to \Gamma\cup\{\infty\}$24, there exist an integer $v:K\to \Gamma\cup\{\infty\}$25 and a definable $v:K\to \Gamma\cup\{\infty\}$26 such that $v:K\to \Gamma\cup\{\infty\}$27 whenever $v:K\to \Gamma\cup\{\infty\}$28. The paper states this in particular for C-minimal expansions of $v:K\to \Gamma\cup\{\infty\}$29 and $v:K\to \Gamma\cup\{\infty\}$30, both of which have value group $v:K\to \Gamma\cup\{\infty\}$31 (Kovacsics et al., 2018).

In characteristic $v:K\to \Gamma\cup\{\infty\}$32, definable completeness and eventual $v:K\to \Gamma\cup\{\infty\}$33-linearity imply a local Jacobian theorem. Outside finitely many points, a definable local C-isomorphism has locally the Jacobian property: on suitable balls it is injective, differentiable with nonvanishing derivative, $v:K\to \Gamma\cup\{\infty\}$34 is constant, and

$v:K\to \Gamma\cup\{\infty\}$35

More generally, any definable family $v:K\to \Gamma\cup\{\infty\}$36 admits a definable partition $v:K\to \Gamma\cup\{\infty\}$37 such that each fiber $v:K\to \Gamma\cup\{\infty\}$38 is finite, $v:K\to \Gamma\cup\{\infty\}$39 is locally constant, and $v:K\to \Gamma\cup\{\infty\}$40 has locally the Jacobian property (Kovacsics et al., 2018).

5. Effective extension of valuations and computable presentations

The computability theory of valued fields introduces a distinct but closely related perspective on algebraically closed metric valued fields. For a computable algebraic valued field $v:K\to \Gamma\cup\{\infty\}$41, the Hensel irreducibility set is

$v:K\to \Gamma\cup\{\infty\}$42

The main theorem states that the following are equivalent: first, for every computable embedding $v:K\to \Gamma\cup\{\infty\}$43 into an algebraic extension, the valuation extends computably to a computable valuation on $v:K\to \Gamma\cup\{\infty\}$44; second, the set $v:K\to \Gamma\cup\{\infty\}$45 is computable (Harrison-Trainor, 2016).

This is an effectiveness condition related to Hensel’s lemma. The proof uses an increasing chain of normal finite extensions, effective enumeration of valuation extensions to each stage, and a theorem distinguishing extensions of valuations by elements whose values separate candidate extensions. Conversely, one diagonalizes against purported computable extensions to recover computability of $v:K\to \Gamma\cup\{\infty\}$46. If $v:K\to \Gamma\cup\{\infty\}$47 is Henselian and computable, then $v:K\to \Gamma\cup\{\infty\}$48 is computable “trivially,” and the extension property holds for all algebraic embeddings (Harrison-Trainor, 2016).

The metric interpretation is direct. Extending $v:K\to \Gamma\cup\{\infty\}$49 extends the ultrametric topology, hence the balls $v:K\to \Gamma\cup\{\infty\}$50 and all valuation-definable open sets. Proposition “embeds-acvf” ensures that every computable nontrivially valued field effectively embeds into a computable algebraically closed valued field with a computable extension of the valuation. The stronger main theorem identifies exactly when such an extension can be carried out along any fixed algebraic embedding (Harrison-Trainor, 2016).

Several negative phenomena delimit the scope of uniform effectivity. There exists a computable algebraic valued field with a splitting algorithm for which one cannot uniformly compute the number of extensions of $v:K\to \Gamma\cup\{\infty\}$51 to $v:K\to \Gamma\cup\{\infty\}$52 as $v:K\to \Gamma\cup\{\infty\}$53 varies. There also exists a computable algebraic valued field whose Henselization is not computable as a subset of a fixed algebraic closure. These results show that valuation extension and Henselization membership cannot, in general, be read off uniformly from minimal polynomials and valuation data alone (Harrison-Trainor, 2016).

A related obstruction appears for formally $v:K\to \Gamma\cup\{\infty\}$54-adic fields. There exists a computable formally $v:K\to \Gamma\cup\{\infty\}$55-adic field that does not embed into any computable $v:K\to \Gamma\cup\{\infty\}$56-adic closure, and the obstruction is the noncomputability of the divisibility relation in the value group. Conversely, if the dividing set of the value group is computable, then one can build a computable embedding into a computable $v:K\to \Gamma\cup\{\infty\}$57-adic closure (Harrison-Trainor, 2016). Although this is a $v:K\to \Gamma\cup\{\infty\}$58-adic rather than algebraically closed result, it isolates the effective role of the value group in controlling valued-field completions and closures.

6. Mal’cev phenomena, computable dimension, and metric significance

For classes with a suitable pregeometry, the Mal’cev property describes the coexistence of computable presentations with sharply different basis behavior. A class has the Mal’cev property if each infinite-dimensional member has one computable presentation with a computable basis and another with no computable basis, both $v:K\to \Gamma\cup\{\infty\}$59-isomorphic to the original structure. By Goncharov’s theorem, structures with the Mal’cev property have computable dimension $v:K\to \Gamma\cup\{\infty\}$60 (Harrison-Trainor, 2016).

Algebraically closed valued fields of infinite transcendence degree satisfy this property. The proof uses the metatheorem of HT–Melnikov–Montalbán with algebraic independence as the r.i.c.e. pregeometry. For ACVF, quantifier elimination and the valuation-topological description of definable sets imply local indistinguishability of independent tuples: definable sets containing one independent tuple contain valuation-open balls, and independent tuples can be moved into such balls by valuation-preserving algebraic transformations. Decidability of ACVF then permits enumeration of independence diagrams. Hence Conditions G and B hold, and ACVF of infinite transcendence degree has the Mal’cev property (Harrison-Trainor, 2016).

The metric reading of this result is that ACVF admits $v:K\to \Gamma\cup\{\infty\}$61-equivalent computable presentations both with a computable transcendence basis and with no computable basis, while the valuation continues to determine the metric structure. The paper formulates this as computable dimension $v:K\to \Gamma\cup\{\infty\}$62 for ACVF in the metric viewpoint: there are infinitely many non-isomorphic computable metric presentations up to computable isomorphism, with algebraic transcendence controlling the metric structure via valuation (Harrison-Trainor, 2016).

This computational perspective complements the geometric one developed through definable types. The stable completion $v:K\to \Gamma\cup\{\infty\}$63, the coding of lattices and residue cosets in $v:K\to \Gamma\cup\{\infty\}$64 and $v:K\to \Gamma\cup\{\infty\}$65, and the equivalence between stable domination and orthogonality to $v:K\to \Gamma\cup\{\infty\}$66 show that the model theory of algebraically closed metric valued fields decomposes systematically into value-group scales, residue-field data, and valued-field loci. A plausible implication is that the same decomposition explains why ACVF supports both a refined imaginary classification and a robust computability theory: in each case, the geometry of balls, lattices, and residue classes is the organizing substrate (Hrushovski, 2014).

Across these lines of work, algebraically closed metric valued fields emerge as ultrametric structures in which valuation-theoretic data, model-theoretic tameness, and computability-theoretic constraints are tightly coupled. The valuation determines the metric; quantifier elimination and C-minimality govern definable sets; stably dominated types encode the residue-field-controlled part of the geometry; elimination of imaginaries proceeds through geometric sorts coding lattices and cosets; and effective extension theorems identify when the ultrametric structure can be transported computably to algebraic closures.