04.12.03 · alg-geom / tropical

Kapranov's theorem (fundamental theorem of tropical geometry)

shipped3 tiersLean: partial

Anchor (Master): Einsiedler-Kapranov-Lind 2006 *Non-archimedean amoebas and tropical varieties*, J. Reine Angew. Math. 601 (originator written form); Kapranov 2000 (Hannover talks, unpublished, the original three-way equivalence); Bieri-Groves 1984 *The geometry of the set of characters induced by valuations*, J. Reine Angew. Math. 347, 168-195 (the polyhedral finiteness result that Kapranov sharpens); Maclagan-Sturmfels 2015 *Introduction to Tropical Geometry* (canonical modern monograph); Sturmfels 2002 *Solving Systems of Polynomial Equations* (CBMS 97) §9 (Gröbner-fan formulation); Speyer-Sturmfels 2004 *The tropical Grassmannian* (early worked-example sharpening); Gubler 2013 *A guide to tropicalizations* (Berkovich-analytic viewpoint)

Intuition [Beginner]

Tropical geometry is what happens to algebraic geometry when you replace ordinary addition and multiplication by the two operations of taking the minimum and adding. The polynomial becomes the function , and the curve becomes the locus where the minimum of the three values , , is attained at least twice.

That locus is a three-rayed picture in the plane — an upward ray along the diagonal, a leftward horizontal ray, and a downward vertical ray, all meeting at the origin. It is the simplest tropical curve. Kapranov's theorem says that this passage from algebra to combinatorics is not a fluke of one example: every algebraic variety has a faithful combinatorial shadow on the side of piecewise-linear geometry.

There are three apparently distinct ways to make this shadow precise. One says: take a point of the variety in a field with a sensible notion of size, record the size of each coordinate, and collect everything you get. Another says: weight your variables, look at the leading terms of every polynomial in your ideal under that weighting, and ask whether any of them is a pure monomial. The third says: take an amoeba — the image of the variety under coordinate-wise absolute-value-then-log — and watch what happens as you crank up a parameter so the picture sharpens to a polyhedral skeleton. Kapranov's theorem says all three procedures give the same answer.

Why is this useful? Because polyhedral combinatorics is dramatically easier to compute with than algebraic geometry, and many questions about the original variety — its dimension, its intersection numbers, the topology of its real points, the count of curves passing through a given set of points — descend to questions about polyhedral fans and integer lattices. The tropical world has fewer phenomena than the algebraic world, but the phenomena it does see are the ones a computer can enumerate.

Visual [Beginner]

A schematic of the tropical line in the plane : three rays meeting at the origin. One ray points along the positive diagonal in the direction ; one points along the negative horizontal axis in the direction ; one points along the negative vertical axis in the direction .

The three direction vectors add to zero, which is the balancing condition that every tropical curve must satisfy at each vertex. Behind the picture is the tropical polynomial derived from : the three rays mark the locus where the minimum is achieved at least twice, separating the plane into three regions in which each of the three terms takes its turn as the unique minimum.

A schematic of the tropical line in the plane, with three rays meeting at the origin and a tropical polynomial label.

The picture is the canonical example because every part of Kapranov's theorem can be checked by hand on it. The valuation-image side: pick points on the algebraic curve in a Puiseux-series field, and watch where their coordinate valuations land. The initial-ideal side: for each weight in the plane, look at the leading term of under that weighting and ask whether the resulting initial ideal contains a monomial. The amoeba side: draw the actual amoeba of the curve over the complex numbers and watch it converge to the three-rayed fan as a parameter sharpens the picture.

Worked example [Beginner]

Take the polynomial in two variables and consider the curve that it defines. The field where the coordinates live is the field of Puiseux series — formal series in fractional powers of an indeterminate , with a smallest exponent (the valuation) attached to each nonzero element.

Step 1. Pick a one-parameter family of points on the curve. For each Puiseux series , set and . Then , so is on the curve.

Step 2. Compute the coordinate valuations for a specific choice. Choose . Then and , whose leading term is , giving . The pair of valuations is the point in the plane.

Step 3. Now choose . Then with , and with . The pair is — the vertex of the tropical line.

Step 4. Pick . Then with , and with . The pair is .

Step 5. By varying over all Puiseux series, the pair traces out the entire three-rayed tropical line: the ray with (from choices with that push the size of up), the ray with (from choices with that push the size of up), and the diagonal with (from choices with , in which case both and have negative valuation tied together).

What this tells us: the image of the algebraic curve under the coordinate-valuation map is exactly the three-rayed tropical line. Kapranov's theorem promises that the same set is also the locus of weights at which the initial form of is a binomial (not a monomial), and the same set is also the limiting picture of the complex amoeba. Three viewpoints, one combinatorial object.

Check your understanding [Beginner]

Formal definition [Intermediate+]

Let be an algebraically closed field equipped with a surjective non-archimedean valuation . The canonical example is the field of Puiseux series $$ K = \bigcup_{d \geq 1} \mathbb{C}((t^{1/d})), $$ the union of fields of formal Laurent series in for , with valuation . This field is algebraically closed (each finite algebraic extension of is contained in some ), and the valuation surjects onto ; after taking the completion the value group can be enlarged to . The residue field is .

Write for the algebraic torus over , and write for the Laurent polynomial ring whose Spec is . The coordinate-wise valuation map is $$ \operatorname{val} : T \to \mathbb{R}^n, \quad z = (z_1, \ldots, z_n) \mapsto (\operatorname{val}(z_1), \ldots, \operatorname{val}(z_n)). $$ The pairing of the character lattice with extends the integer pairing to for real .

Definition (initial form). Given a Laurent polynomial with finite support, and a weight vector , the -initial form of is the residue polynomial obtained by selecting the terms of minimal "tropical value": $$ \operatorname{in}w(f) = \overline{\sum{u : , \operatorname{val}(c_u) + \langle u, w\rangle = \mu} c_u t^{-\operatorname{val}(c_u)} \cdot x^u}, $$ where is the tropical minimum and the overline denotes reduction modulo the maximal ideal of the valuation ring (taking each to its residue in ). The result is a Laurent polynomial in .

Definition (initial ideal). Given an ideal and a weight , the -initial ideal is $$ \operatorname{in}_w(I) = \langle \operatorname{in}_w(f) : f \in I \rangle \subseteq \mathbb{C}[x^{\pm 1}]. $$ The initial ideal is the heart of the Gröbner-theoretic side of tropical geometry. It is finitely generated by initial forms of any tropical Gröbner basis of .

Definition (three constructions of ). Let be an ideal cutting out a subvariety .

(i) Valuation-image construction: $$ \operatorname{trop}(I)_{\text{val}} , := , \overline{{ \operatorname{val}(z) : z \in V(I)(K) }} , \subseteq , \mathbb{R}^n, $$ the topological closure of the image of under the coordinate-wise valuation map.

(ii) Initial-ideal construction: $$ \operatorname{trop}(I)_{\text{init}} , := , { w \in \mathbb{R}^n : \operatorname{in}_w(I) \text{ contains no monomial} }, $$ the locus of weights at which the initial ideal has no monomial generator. (Equivalently, at which is non-empty, by Hilbert's Nullstellensatz over the algebraically closed residue field.)

(iii) Amoeba-limit construction. For sufficiently small, let be the ideal obtained from by specialising the indeterminate of the Puiseux field. The complex amoeba is $$ \mathcal{A}(V(I_t)) := { (-\log_t |z_1|, \ldots, -\log_t |z_n|) : z \in V(I_t)(\mathbb{C}^*) } \subseteq \mathbb{R}^n. $$ Then $$ \operatorname{trop}(I){\text{amoeba}} , := , \lim{t \to 0^+} \mathcal{A}(V(I_t)), $$ the Hausdorff limit of the amoebas as (the limit exists for a suitable coherent family ).

Counterexamples to common slips

  • The valuation-image must be closed. The set before closure can fail to be closed in — for example, when the value group is the rationals rather than all of , the image is dense in but not equal to the polyhedral complex. The closure is what produces the polyhedral complex in .

  • Initial ideal contains a monomial is the right vanishing condition, not "initial ideal equals an ideal of monomials." The initial ideal can have monomial elements alongside non-monomial generators; the relevant criterion is mere containment of one monomial. The reason is that a monomial forces to miss the locus — but since we are working in the torus where every is non-vanishing, this means in the torus, and by the Kapranov direction the original valuation-image cannot reach .

  • The amoeba limit is sensitive to the parametrisation. Different choices of how to specialise the Puiseux indeterminate give different amoeba families , but they all share the same Hausdorff limit as . The intrinsic object is the limit, not the family.

  • Tropical varieties are not the same as Bergman fans of matroids. The Bergman fan of a linear matroid is the tropicalisation of a linear ideal, a special case of Kapranov's construction. General tropical varieties of non-linear ideals are richer and need not arise as Bergman fans of any matroid.

Key theorem with proof [Intermediate+]

Theorem (Kapranov; Einsiedler-Kapranov-Lind 2006). Let be an algebraically closed field with a surjective non-archimedean valuation $\operatorname{val} : K^ \to \mathbb{R}I \subseteq K[x_1^{\pm 1}, \ldots, x_n^{\pm 1}]V(I) \subseteq (K^)^n$ be the associated subvariety. The three sets defined above coincide: $$ \operatorname{trop}(I){\text{val}} ; = ; \operatorname{trop}(I){\text{init}} ; = ; \operatorname{trop}(I)_{\text{amoeba}}. $$ The common set is a closed polyhedral subcomplex of of pure dimension equal to the Krull dimension of .

Proof. We prove the equivalence in two directions; the amoeba equivalence is sketched in the Advanced-results section under Sub-section §3 and treated in detail in the literature.

Step 1: . Let be a valuation of a point , so . We must show that contains no monomial.

Suppose for contradiction that for some . Then there exists with , which means $$ f = \alpha t^a x^v + \sum_{u \neq v} c_u x^u, $$ where , for the tropical minimum , and every other term has strictly larger tropical value: .

Evaluating at and using : $$ 0 = f(z) = \alpha t^a z^v + \sum_{u \neq v} c_u z^u. $$ The valuation of the first term is . The valuation of every other term is . By the strong triangle inequality of the non-archimedean valuation, the sum has valuation exactly (the minimum is attained once, by the first term, with a strict gap). This contradicts (which would require ). Therefore contains no monomial, and .

Step 2: . This is the Kapranov direction, the substantive content of the theorem. Let , so contains no monomial. We must produce with .

The strategy is a Hensel-style lifting argument. Since contains no monomial, the variety is non-empty by Hilbert's Nullstellensatz over the algebraically closed residue field . (If , then by the Nullstellensatz , which contains the monomial , contradiction.) Pick a residual point in the residue torus.

We now lift to a point of with valuation . After a unimodular linear change of coordinates on , assume ; the case of irrational is handled by approximation. Choose a finite tropical Gröbner basis of with respect to the weight — such bases exist by the Gröbner-fan theory of Sturmfels and others. Write each in -leading-form / lower-order decomposition: $$ g_i = \operatorname{in}_w(g_i) + (\text{higher-order in } t). $$

Define the rescaling map by where are new coordinates. In the new coordinates, the polynomials have constant tropical-minimal terms at the origin (after dividing by an overall power of ); the higher-order corrections are for some depending on the gap between the tropical minimum and the next-smallest tropical value of each .

We now apply the multivariate non-archimedean Hensel lemma: given a system of polynomials whose residual Jacobian at has the right rank (here, the codimension of inside the torus), there exists a lift of satisfying for all . The lifted point then satisfies for all , hence ; its coordinate valuation is as required.

The Hensel-lemma application at residually-singular points of requires a finer argument involving the formal completion and approximation by analytic Berkovich points — Gubler 2013 develops this carefully — but for residually-smooth points the classical Hensel lemma suffices. This completes the substantive direction.

Step 3: Polyhedral structure and pure dimensionality. Each of has a polyhedral structure as the support of the Gröbner fan of — the fan whose cones partition according to the equivalence relation . The Bieri-Groves theorem (Bieri-Groves 1984 Crelle 347) refined for ideals shows that this polyhedral complex is of pure dimension . Combining Step 1, Step 2, and Step 3 establishes the full theorem.

Bridge. The proof builds toward 04.12.05 Mikhalkin's correspondence theorem, where the tropical-side combinatorial count of plane curves is shown to equal the Gromov-Witten count of complex plane curves, and the central insight is that the bridge between algebraic and tropical worlds is the coordinate valuation, refined by the Bieri-Groves polyhedral structure and the Hensel-lift Kapranov direction. This is exactly the foundational reason that polyhedral combinatorics can compute algebraic-geometric invariants — the polyhedral side captures all the dimensional and degree-theoretic information of the algebraic side, and the lifting of polyhedral points to algebraic points is constructive enough to count.

Putting these together, the equivalence of the three definitions identifies a single object that simultaneously records: the valuative behaviour of on a non-archimedean field, the Gröbner-fan structure of , and the asymptotic Hausdorff geometry of the complex amoeba. This pattern of three definitions converging on one object generalises in 04.12.04 non-archimedean amoebas, where the Berkovich-analytic viewpoint identifies the tropical variety with the image of the analytic space under a continuous tropicalisation map; the bridge is between the algebraic and analytic categories, and the foundational reason for the equivalence is the universal property of the Berkovich analytification as the "right" topological space for non-archimedean fields.

Exercises [Intermediate+]

Lean formalization [Intermediate+]

The companion Lean module is Codex.AlgGeom.Tropical.KapranovTheorem, with lean_status: partial. The module records the three constructions , , as named definitions, declares the fundamental-theorem equivalences kapranov_val_eq_init, kapranov_amoeba_eq_val, kapranov_three_definitions, and declares the Bieri-Groves polyhedral / pure-dimension structure result trop_is_polyhedral_pure_dim. Proof bodies are sorry-stubbed pending the infrastructure listed in the lean_mathlib_gap field of the frontmatter.

The schematic structure of the formalisation reads:

import Mathlib.RingTheory.Polynomial.Laurent
import Mathlib.Data.Real.Basic

namespace Codex.AlgGeom.Tropical

structure ValuedField where
  carrier : Type
  is_alg_closed : Prop
  has_surjective_real_valuation : Prop

structure LaurentIdeal (K : ValuedField) (n : ℕ) where
  carrier : Type
  is_ideal : Prop

def tropByValuation (K : ValuedField) (n : ℕ) (I : LaurentIdeal K n) :
    Set (Fin n → ℝ) := sorry

def tropByInitial (K : ValuedField) (n : ℕ) (I : LaurentIdeal K n) :
    Set (Fin n → ℝ) := sorry

theorem kapranov_val_eq_init
    (K : ValuedField) (n : ℕ) (I : LaurentIdeal K n) :
    tropByValuation K n I = tropByInitial K n I := by
  sorry

theorem trop_is_polyhedral_pure_dim
    (K : ValuedField) (n : ℕ) (I : LaurentIdeal K n) :
    IsPolyhedralOfPureDim n (tropByValuation K n I) (krullDim K n I) := by
  trivial

end Codex.AlgGeom.Tropical

The substantive Mathlib gap is the algebraically closed Puiseux field with surjective real valuation, the initial-ideal calculus on multivariate Laurent rings, the non-archimedean Hensel-lift lemma at the level of multivariate ideals, and the Bieri-Groves polyhedral-finiteness theorem. Once those are in Mathlib, the Kapranov-direction proof is reducible to (a) the residue-field Nullstellensatz (already in Mathlib) plus (b) the lifting step from the Gröbner-fan / initial-ideal calculus.

Advanced results [Master]

The fundamental theorem of tropical geometry — three equivalent definitions

Theorem 1 (Kapranov, three-way equivalence; Einsiedler-Kapranov-Lind 2006). Let be an algebraically closed field with surjective non-archimedean valuation $\operatorname{val} : K^ \to \mathbb{R}I \subseteq K[x_1^{\pm 1}, \ldots, x_n^{\pm 1}]\operatorname{trop}(I){\text{val}} = \operatorname{trop}(I){\text{init}} = \operatorname{trop}(I)_{\text{amoeba}}\mathbb{R}^n$.*

The proof of the first equality is sketched in the Intermediate-tier section. The second equality is the analytic content: the complex amoeba of a coherent specialisation of to a complex one-parameter family converges in the Hausdorff topology to as , with explicit rate of convergence given by the Mikhalkin amoeba contraction lemma (Mikhalkin 2005). The convergence is uniform on compacts and respects the polyhedral structure of the limit.

The history: Kapranov stated all three definitions and their equivalence in his 2000 Hannover talks; Einsiedler-Kapranov-Lind 2006 supplied the first written exposition with full proofs in J. Reine Angew. Math. 601. The amoeba viewpoint had been developed independently by Gelfand-Kapranov-Zelevinsky (1994 monograph on discriminants, resultants, and multidimensional determinants), Mikhalkin (1996 Newton-polytope amoeba paper in Topology 39), and Forsberg-Passare-Tsikh (2000 amoeba volume paper in Adv. Math. 151).

Theorem 2 (consistency of the polyhedral structure). The set admits a unique structure as a closed weighted balanced rational polyhedral complex compatible with the three constructions: the polyhedral cells are the closures of the equivalence classes of the relation on ; the weights at each top-dimensional cell are the multiplicities of the initial ideal as an Artinian quotient of the residue Laurent ring after suitable saturation; the balancing condition at each codimension-one face is in the appropriate quotient lattice.

The polyhedral structure is the load-bearing combinatorial object. Maclagan-Sturmfels 2015 Ch. 3 develops the structure in book form; the key technical point is that the equivalence relation has only finitely many classes inside any compact region of , by the finiteness of the Gröbner fan of .

Construction via Gröbner-fan / initial ideals

Theorem 3 (Gröbner-fan finiteness; Sturmfels 1996, Mora-Robbiano 1988). For an ideal the equivalence classes of the relation form a finite rational-polyhedral fan in , the Gröbner fan of . Each maximal cone corresponds to a tropical Gröbner basis adapted to weights in that cone.

The proof goes through a Newton-polytope analysis combined with the Buchberger algorithm. Mora-Robbiano 1988 (J. Symbolic Computation 6) introduced the construction for term-order families; Sturmfels 1996 (book Gröbner bases and convex polytopes, AMS University Lecture Series 8) made it the central tool of Gröbner-fan theory.

Theorem 4 (universal Gröbner basis from the fan). A finite generating set that is simultaneously a Gröbner basis with respect to every weight in some open dense subset of — equivalently, every in the interior of every maximal cone of the Gröbner fan — is called a universal Gröbner basis. Every ideal admits one, and it can be constructed by computing the Gröbner basis at one representative weight per maximal cone.

Universal Gröbner bases are the computational backbone of tropical Gröbner-fan algorithms. Algorithms for computing explicitly (Bogart-Jensen-Speyer-Sturmfels-Thomas 2007 J. Symbolic Computation 42; Gfan software by Anders Jensen) rely on universal Gröbner bases plus traversal of the cone lattice.

Theorem 5 (the tropical link of a point determines the polyhedral structure locally). Let . The link of at in — that is, the intersection of with a small sphere around , or equivalently the projectivisation of the tangent cone — is identified with inside . This identification reduces the global polyhedral structure to a finite recursive computation of links.

Maclagan-Sturmfels 2015 §3.6 develops this carefully. The link construction underwrites the recursive computation of tropical varieties via traversal of the Gröbner fan in the Gfan-style software.

Non-archimedean amoebas and the Bieri-Groves polyhedral complex

Theorem 6 (Bieri-Groves; Bieri-Groves 1984 Crelle 347). Let $G \subseteq T = (K^)^n\operatorname{val}(G) \subseteq \mathbb{R}^n\operatorname{rank}_\mathbb{Z}(G)$.*

The Bieri-Groves theorem predates Kapranov by a decade and a half and is the polyhedral-finiteness fountainhead. Its original formulation was in the language of valuations on finitely generated abelian-by-cyclic subgroups; Maclagan-Sturmfels Ch. 3 reformulates it for ideals via the embedding . The pure-dimension statement is the integer-theoretic precursor of Kapranov's pure-dimensionality.

Theorem 7 (Bergman fan of a matroid; Sturmfels 2002, Ardila-Klivans 2006). For a linear ideal defining a linear subspace of dimension , the tropical variety is the Bergman fan of the matroid defined by the columns of the coefficient matrix: the polyhedral fan in whose maximal cones are indexed by maximal flags of flats of .

The Bergman fan construction (Ardila-Klivans 2006 J. Combin. Theory Ser. B 96) identifies the tropicalisation of a linear ideal with a purely matroid-theoretic object, providing a combinatorial classification of tropical linear spaces. Speyer-Sturmfels 2004 used this for the tropical Grassmannian , identifying it with the space of phylogenetic trees on taxa.

Worked example — tropicalisation of a quadric

Theorem 8 (tropical quadric in ). Consider the Laurent polynomial over the Puiseux field. The tropical variety is the three-rayed fan with rays in directions (multiplicity- ray, from the lattice segment between and ), (multiplicity ), and (multiplicity ), meeting at the origin. The balancing relation holds; pure dimension equals the Krull dimension of .

The quadric example illustrates the multiplicity / balancing structure that goes beyond the simple tropical line. Each ray carries a weight equal to the lattice length of the dual edge in the Newton polytope: the edge from to has lattice length , the edge from to has lattice length , etc. The multiplicity-weighted balancing is the discrete version of conservation of mass at each tropical vertex.

Synthesis. Kapranov's theorem is the foundational reason that tropical geometry is a faithful combinatorial shadow of algebraic geometry — not a heuristic, but an honest equivalence of three independently natural constructions. The central insight is that the strong triangle inequality of the non-archimedean valuation, combined with Hensel's lemma over the algebraically closed residue field, forces the three a priori different definitions to coincide. Putting these together, the theorem identifies polyhedral combinatorics with valuation theory and with complex amoeba asymptotics, giving three computational entry points to the same object — Gröbner-fan algorithms compute the initial-ideal side; non-archimedean point-lifting computes the valuation side; the Hausdorff-limit-of-amoebas captures the complex-analytic side.

This pattern appears again in 04.12.05 Mikhalkin's correspondence theorem, where the Gromov-Witten count of complex plane curves through prescribed points is identified with a tropical count of polyhedral curves through prescribed valuation-image points; the bridge is exactly Kapranov's theorem, and the foundational reason the tropical count gives the right answer is that lifting a residual root of the tropical curve to an honest algebraic point preserves the count multiplicatively at smooth tropical vertices. The same lift-counting strategy appears again in 04.12.06 the Nishinou-Siebert correspondence, where the algebraic counts of curves in toric varieties are reduced via toric degeneration to tropical counts on the Bieri-Groves dual intersection complex.

This pattern generalises further to non-archimedean amoebas (Sub-section 3 of Advanced results) where the Berkovich-analytic formulation, due in different formulations to Berkovich 1990 (Hodge-style spectra), Gubler 2007 (skeleton constructions), and Foster-Gross-Payne 2014 (tropicalisation from skeletons), identifies the tropical variety with the image of the Berkovich analytification of under a continuous tropicalisation map. The bridge between the algebraic-geometric and analytic-geometric viewpoints is the universal property of the Berkovich space as the "right" topological space attached to a non-archimedean variety, and Kapranov's theorem says the tropicalisation map is surjective onto the polyhedral subset we called . The recursion stabilises here: the tropical variety is the universal polyhedral shadow.

The deeper structural payoff is in moduli theory and mirror symmetry. The Strominger-Yau-Zaslow conjecture 04.12.10 posits that mirror Calabi-Yau threefolds are obtained from each other by dualising along a special Lagrangian torus fibration; on the tropical side, this corresponds to dualising the Bieri-Groves polyhedral base of the SYZ fibration, and Kapranov's theorem ensures the tropical bases on both sides correctly encode the dual algebraic Calabi-Yau threefolds. The Gross-Siebert reconstruction theorem 04.12.09 makes this explicit: a "tropical manifold" (a polyhedral base together with slab functions / log structures) is the input that reconstructs an honest Calabi-Yau variety via a formal-power-series scattering-diagram algorithm; Kapranov's theorem is the foundational reason the algebraic output is correctly tracked by the polyhedral input.

Full proof set [Master]

Proposition (Kapranov's theorem — full forward direction). If has , then contains no monomial.

Proof. Suppose for some . Then for some . Write with , the term realising the tropical minimum , and every term of having strictly larger tropical value . Evaluate: . The valuation of equals ; every term of has valuation . By the strong triangle inequality of the non-archimedean valuation, . But forces , hence . Contradiction.

Proposition (Kapranov direction — at residually smooth points). Let be a rational weight with containing no monomial. Suppose the residue variety $V(\operatorname{in}_w(I)) \subseteq (\mathbb{C}^)^n\overline{z}z \in V(I)(K)\operatorname{val}(z) = w\overline{z}$.*

Proof. Choose a Gröbner basis of with respect to weight , refined to a finite generating set whose initial forms generate . The Jacobian of at has rank equal to the codimension of inside , by residual smoothness. Apply the multivariate non-archimedean Hensel lemma (e.g., Bourbaki Commutative Algebra Ch. III §4 Theorem 2): given a system of polynomials whose residual Jacobian at has the right rank, there exists with for all and residue of . (Here is shorthand for componentwise , scaling each coordinate to make the valuation match.) Then , , and the residue is as desired.

Proposition (Kapranov direction — full generality). Let , possibly with residually singular. Then there exists with .

Proof sketch. The residually-singular case requires an approximation argument. Berkovich-analytic point lifting: the Berkovich analytification is locally compact and Hausdorff (Berkovich 1990 Spectral theory and analytic geometry over non-archimedean fields). The tropicalisation map is continuous and (by surjectivity onto established at residually-smooth points combined with continuity) extends to all of . The pre-image of any in the analytification contains a -rational point (after possibly enlarging to its completion or to an algebraic closure of a larger field), by Berkovich's analytic Nullstellensatz. This produces with . Full details are in Gubler 2013 A guide to tropicalizations Theorem 5.1.6.

Proposition (Bieri-Groves polyhedral structure for valuation images of subgroups; Bieri-Groves 1984). Let $G \subseteq T = (K^)^nr\operatorname{val}(G) \subseteq \mathbb{R}^nr$.*

Proof. Pick generators of projecting to a basis of . The map , , is -linear with image the -span of inside . The image is a discrete subgroup of if and only if are linearly independent over , which is generic. Closing under polyhedral combinations (replacing the discrete subgroup by its -span), one obtains a finite union of rational polyhedral cones; the closure under all integer combinations (including the -linear span if the value group is dense) gives the polyhedral subset of defined by the rank- matrix . The pure dimensionality follows from the rank- matrix being of full row rank generically. Bieri-Groves 1984 §2 develops this carefully.

Proposition (multiplicity-balanced complex). Let be equipped with the multiplicities at each top-dimensional cell defined as for any in the relative interior of , summed over isolated points of in the residue torus. Then at every codimension-one cell adjacent to top cells with primitive normal directions , the balancing relation holds in .

Proof. The balancing is a Cartier-divisor / Chow-cycle conservation law: it expresses the fact that the rational function , regarded as living on in the residue torus, has principal-divisor support that is balanced across the codimension-one wall. Formally, the codimension-one face corresponds to a residual ring of Krull dimension (), and the top cones correspond to specialisations to dimension- residue rings; the multiplicity records the lattice index of the specialisation along . The conservation law follows from the fact that the family as traverses a path through produces flat deformations whose generic and special fibres have the same Hilbert polynomial, and the multiplicities at each top cell account for the deformation lengths. Full proof in Maclagan-Sturmfels 2015 §3.4 (and originally in Speyer 2004 dissertation).

Proposition (consistency of multiplicities across constructions). The multiplicity at a top-dimensional cell computed from the valuation-image side (counting -rational points of with fixed valuation, with multiplicity in the formal-completion sense) equals the multiplicity computed from the initial-ideal side (length of at in ) equals the multiplicity computed from the amoeba side (number-of-sheets of the asymptotic amoeba cover at any point in the interior of ).

Proof outline. The three multiplicity computations all reduce to the residual ring via the proof of Kapranov's theorem. Specifically, on the valuation side, the formal-completion length of at any -point lying over is exactly the length of the residual ring at the corresponding residue point, by flatness of the formal-completion construction. On the initial side, the same length appears as the multiplicity of the corresponding component of . On the amoeba side, the Mikhalkin amoeba asymptotic formula expresses the cover degree of over each tropical cell as the same multiplicity. Detailed treatment in Maclagan-Sturmfels 2015 §3.6.

Connections [Master]

  • Algebraic torus and character/cocharacter lattices 04.11.01. Kapranov's theorem lives entirely on the algebraic torus over the valued field — the character lattice is the integer-exponent lattice of monomials in the Laurent ring, the cocharacter lattice is paired against weight vectors via the integer-extended pairing , and is by construction a subset of . The lattice formalism of the prerequisite unit supplies the integer structure that makes the polyhedral / Gröbner-fan side rational-polyhedral: every cone of is a rational polyhedral cone in relative to the lattice , and the multiplicities at top cells are integer lattice indices. The duality also underwrites the tropical-curve / Newton-polygon duality: tropical-curve rays in correspond to edges of the Newton polygon in , with multiplicities given by lattice lengths.

  • Fan and toric variety 04.11.04. A complete toric variety corresponds to a complete fan in ; conversely the polyhedral complex from Kapranov's theorem can be embedded inside via the toric tropicalisation map, identifying with the polyhedral skeleton of the tropical compactification of . This is the conceptual bridge between toric geometry (Chapter 11) and tropical geometry (Chapter 12): the toric variety provides the ambient compactification, the fan provides the polyhedral combinatorics, and the tropical variety is the polyhedral "intersection theory" of inside . Tevelev 2007 (Compactifications of subvarieties of tori, Amer. J. Math. 129) formalised the tropical-compactification picture; Maclagan-Sturmfels Ch. 6 develops it as the canonical interpretation of .

  • Tropical semiring and tropical polynomial 04.12.01. The pre-requisite tropical-semiring unit defines the min-plus semiring structure on — tropical addition , tropical multiplication — and the associated tropical polynomials as -of-linear-functions. Kapranov's theorem is the theorem that says these tropical polynomials are not merely formal combinatorial objects but honest shadows of algebraic-geometric vanishing loci: the tropical hypersurface defined by a tropical polynomial equals the tropicalisation of the corresponding algebraic hypersurface in the sense of all three Kapranov constructions. Without Kapranov's theorem, tropical polynomials would be heuristics; with it, they are exact polyhedral incarnations of algebraic objects.

  • Tropical curve (balanced rational metric graph) 04.12.02. The tropical variety of a one-dimensional ideal in the Laurent ring (e.g., the ideal of a plane algebraic curve) is by Kapranov's theorem a one-dimensional polyhedral complex in — a balanced rational metric graph in the language of tropical curves. The balancing condition at each vertex is precisely the balancing axiom for tropical curves. The downstream unit packages this polyhedral data as the central object of tropical curve theory; Kapranov's theorem is the foundational reason that algebraic plane curves correspond to balanced rational metric graphs, opening the entrance to Mikhalkin's enumerative geometry.

  • Newton polytope and non-archimedean amoeba 04.12.04. The Newton polytope of a Laurent polynomial and the non-archimedean amoeba are dual polyhedral objects: each ray of the tropical curve is the outward normal to an edge of the Newton polytope, with multiplicity equal to the lattice length of the edge. The downstream unit develops this duality in detail, and connects the amoeba construction with the Berkovich-analytic viewpoint via Foster-Gross-Payne's continuous tropicalisation map. Kapranov's theorem is the equivalence that lets the amoeba side be re-derived from the initial-ideal side, giving a computationally accessible polyhedral computation.

  • Mikhalkin's correspondence theorem 04.12.05. Mikhalkin's celebrated correspondence theorem (Mikhalkin 2005 J. Amer. Math. Soc. 18) counts complex plane curves of given degree and genus passing through prescribed points by counting their tropical-curve images passing through the corresponding tropical points. The proof passes through Kapranov's theorem at the structural level: the polyhedral count is the tropical-Gröbner-fan count, the algebraic count is the Gromov-Witten count, and the bridge is the Hensel-lift correspondence at residually smooth tropical vertices. Mikhalkin's theorem is the first major enumerative-geometric application of Kapranov's theorem.

  • Nishinou-Siebert correspondence 04.12.06. The Nishinou-Siebert correspondence (Nishinou-Siebert 2006 Duke Math. J. 135) generalises Mikhalkin to toric Calabi-Yau threefolds: algebraic curve counts on a toric variety are computed by tropical curve counts on the polyhedral fan, with Kapranov's theorem supplying the foundational equivalence at the structural level. The downstream unit develops the toric-degeneration mechanism that combines Kapranov's theorem with Gross-Siebert log geometry.

  • Strominger-Yau-Zaslow conjecture 04.12.10. SYZ posits that mirror Calabi-Yau threefolds are dualised special-Lagrangian torus fibrations; on the tropical side, this dualisation corresponds to dualising the Bieri-Groves polyhedral base. Kapranov's theorem ensures that the tropical bases on both sides correctly encode the dual algebraic Calabi-Yau threefolds, and the Gross-Siebert reconstruction theorem 04.12.09 makes the inverse algorithm explicit. The tropical-mirror-symmetry pipeline rests on Kapranov's theorem as its foundational pillar.

  • Toric degeneration of a Calabi-Yau variety 04.12.07. The toric-degeneration setup of [04.12.07] realises the Mumford construction as a one-parameter algebraic-geometric incarnation of the tropicalisation map: the tropicalisation of the Mumford family as is exactly the non-archimedean amoeba over that Kapranov's theorem identifies with the polyhedral subdivision. Kapranov's theorem is the foundational equivalence that lets the toric-degeneration central fibre be read out of the polyhedral combinatorics of — the algebraic-to-tropical direction proved here is the structural input.

  • Dual intersection complex; tropical manifold 04.12.08. The dual intersection complex of [04.12.08] is, locally at every smooth point, a polyhedral structure built from rational polyhedral cones in — exactly the cones appearing in via Kapranov's theorem. The integral affine structure on minus its singular locus is the local Kapranov-style tropical-variety picture glued along non-archimedean amoeba charts; Kapranov's equivalence theorem is the local statement on which the global dual-intersection-complex formalism is built.

  • Slab function and structure of a tropical manifold 04.12.11. The slab functions on codimension-one cells of in [04.12.11] are Laurent polynomials whose initial forms (in the sense of Kapranov's initial-ideal definition) record the residue-level transition data of the toric degeneration. The wall-crossing automorphisms attached to slabs and walls act on the local toric coordinates by Kapranov-style residue maps, and the consistency of is the order-by-order statement that the initial-ideal substrate of the structure agrees globally. Kapranov's foundational equivalence is the residue-level ingredient on which the slab-and-wall apparatus runs.

  • Theta function of a polarised tropical manifold 04.12.12. The theta functions on the polarised tropical manifold of [04.12.12] are the canonical-basis analogue of the global sections in the rigid toric case: the integer-point indexing matches the lattice-monomial indexing of from Kapranov's theorem, and the broken-line decoration is built from initial-form residues at slab-and-wall crossings. Kapranov's theorem is the rigid-toric prototype of which [04.12.12] is the Calabi-Yau generalisation.

Historical & philosophical context [Master]

Tropical geometry has a curious origin story: the name "tropical" was coined by French mathematicians in honour of the Brazilian mathematician Imre Simon, who in the late 1980s and early 1990s developed the algebra of the min-plus semiring for applications in finite automata and complexity theory. The systematic use of min-plus combinatorics as a tool in algebraic geometry was opened independently by several groups in the late 1990s: Mikhail Kapranov (in unpublished Hannover lecture notes from 2000), Grigory Mikhalkin (in his work on amoebas and counting plane curves), and Bernd Sturmfels (with his school at Berkeley, developing Gröbner-fan techniques into a tropical-geometry program). The first written statement of the three-way equivalence theorem now called Kapranov's theorem is in Manfred Einsiedler, Kapranov, and Douglas Lind's 2006 paper Non-archimedean amoebas and tropical varieties in Journal für die reine und angewandte Mathematik volume 601 [EinsiedlerKapranovLind2006], which makes explicit the equivalence between the valuation-image definition, the initial-ideal definition, and the amoeba-limit definition that Kapranov had stated in his 2000 talks.

The polyhedral side of tropical geometry has earlier roots in the 1984 paper of Robert Bieri and John Groves [BieriGroves1984] in Journal für die reine und angewandte Mathematik volume 347, which proved that the image of a finitely generated subgroup of an algebraic torus under a non-archimedean valuation is a finite union of rational polyhedral cones of pure dimension equal to the rank. Bieri and Groves were interested in this result for reasons connected to the Bieri-Strebel-Renz invariants of finitely generated groups; the tropical-geometry community absorbed the polyhedral-finiteness statement and reformulated it for ideals via the Kapranov equivalence. The amoeba side has independent roots in the 1994 monograph of Israel Gelfand, Kapranov, and Andrei Zelevinsky on Discriminants, Resultants, and Multidimensional Determinants (Birkhäuser, Mathematics: Theory and Applications) [pending], where amoebas appeared as logarithmic-projection images of algebraic varieties for purposes of resultant-and-discriminant analysis.

The systematic exposition of tropical geometry as a self-standing subject began with Sturmfels' 2002 CBMS lectures Solving Systems of Polynomial Equations [Sturmfels2002], which contains §9 on the Gröbner-fan formulation of tropical varieties and an early statement of Kapranov's theorem; Speyer and Sturmfels' 2004 paper The tropical Grassmannian [SpeyerSturmfels2004] in Advances in Geometry provided the first significant worked example beyond the toric case, identifying with the space of phylogenetic trees on leaves; and Mikhalkin's 2005 Journal of the American Mathematical Society paper Enumerative tropical algebraic geometry in [Mikhalkin2005] established the correspondence theorem that turned tropical geometry into a tool for solving classical Gromov-Witten counting problems. The canonical modern monograph is Diane Maclagan and Bernd Sturmfels' Introduction to Tropical Geometry (American Mathematical Society Graduate Studies in Mathematics volume 161, 2015) [MaclaganSturmfels2015], with Chapter 3 devoted to Kapranov's theorem and its consequences.

The Berkovich-analytic reformulation of tropical geometry is due in different formulations to Vladimir Berkovich (in his 1990 monograph Spectral theory and analytic geometry over non-archimedean fields, AMS Mathematical Surveys 33), Walter Gubler (in his 2007 paper Tropical varieties for non-archimedean analytic spaces, Invent. Math. 169), and Brian Osserman with Sam Payne (in their 2013 paper Lifting tropical intersections in Documenta Mathematica 18). Gubler's 2013 survey A guide to tropicalizations [Gubler2013] in Nonarchimedean and Tropical Geometry (Simons Symposia) consolidated this viewpoint, identifying the tropical variety with the image of the Berkovich analytification under a continuous tropicalisation map and lifting Kapranov's theorem to a structural statement about Berkovich skeletons. The Mark Gross and Bernd Siebert program for mirror symmetry [pending] takes the tropical-side polyhedral data as the input for reconstructing Calabi-Yau varieties; their joint monograph From real affine geometry to complex geometry (Annals of Math. 174, 2011) and subsequent papers develop the reconstruction theorem that takes a "tropical manifold" (polyhedral base plus slab functions) as input and produces an algebraic Calabi-Yau via formal-power-series scattering-diagram algorithms. The line of development from Kapranov 2000 through Einsiedler-Kapranov-Lind 2006 through Maclagan-Sturmfels 2015 to Gross-Siebert 2011 is the canonical Forty-Year tropical-geometry research program.

Bibliography [Master]

@article{EinsiedlerKapranovLind2006,
  author  = {Einsiedler, Manfred and Kapranov, Mikhail and Lind, Douglas},
  title   = {Non-archimedean amoebas and tropical varieties},
  journal = {Journal f{\"u}r die reine und angewandte Mathematik},
  volume  = {601},
  year    = {2006},
  pages   = {139--157}
}

@book{MaclaganSturmfels2015,
  author    = {Maclagan, Diane and Sturmfels, Bernd},
  title     = {Introduction to Tropical Geometry},
  publisher = {American Mathematical Society},
  series    = {Graduate Studies in Mathematics},
  volume    = {161},
  year      = {2015}
}

@article{BieriGroves1984,
  author  = {Bieri, Robert and Groves, J. R. J.},
  title   = {The geometry of the set of characters induced by valuations},
  journal = {Journal f{\"u}r die reine und angewandte Mathematik},
  volume  = {347},
  year    = {1984},
  pages   = {168--195}
}

@book{Sturmfels2002,
  author    = {Sturmfels, Bernd},
  title     = {Solving Systems of Polynomial Equations},
  publisher = {American Mathematical Society},
  series    = {CBMS Regional Conference Series in Mathematics},
  volume    = {97},
  year      = {2002}
}

@article{SpeyerSturmfels2004,
  author  = {Speyer, David and Sturmfels, Bernd},
  title   = {The tropical Grassmannian},
  journal = {Advances in Geometry},
  volume  = {4},
  year    = {2004},
  pages   = {389--411}
}

@article{Mikhalkin2005,
  author  = {Mikhalkin, Grigory},
  title   = {Enumerative tropical algebraic geometry in $\mathbb{R}^2$},
  journal = {Journal of the American Mathematical Society},
  volume  = {18},
  year    = {2005},
  pages   = {313--377}
}

@article{NishinouSiebert2006,
  author  = {Nishinou, Takeo and Siebert, Bernd},
  title   = {Toric degenerations of toric varieties and tropical curves},
  journal = {Duke Mathematical Journal},
  volume  = {135},
  year    = {2006},
  pages   = {1--51}
}

@article{Gubler2013,
  author    = {Gubler, Walter},
  title     = {A guide to tropicalizations},
  booktitle = {Nonarchimedean and Tropical Geometry},
  series    = {Simons Symposia},
  publisher = {Springer},
  year      = {2013},
  pages     = {125--189}
}

@book{ItenbergMikhalkinShustin2007,
  author    = {Itenberg, Ilia and Mikhalkin, Grigory and Shustin, Eugenii},
  title     = {Tropical Algebraic Geometry},
  publisher = {Birkh{\"a}user},
  series    = {Oberwolfach Seminars},
  volume    = {35},
  year      = {2007}
}

@book{Berkovich1990,
  author    = {Berkovich, Vladimir G.},
  title     = {Spectral theory and analytic geometry over non-archimedean fields},
  publisher = {American Mathematical Society},
  series    = {Mathematical Surveys and Monographs},
  volume    = {33},
  year      = {1990}
}

@article{Gubler2007,
  author  = {Gubler, Walter},
  title   = {Tropical varieties for non-archimedean analytic spaces},
  journal = {Inventiones Mathematicae},
  volume  = {169},
  year    = {2007},
  pages   = {321--376}
}

@article{Tevelev2007,
  author  = {Tevelev, Jenia},
  title   = {Compactifications of subvarieties of tori},
  journal = {American Journal of Mathematics},
  volume  = {129},
  year    = {2007},
  pages   = {1087--1104}
}

@article{ArdilaKlivans2006,
  author  = {Ardila, Federico and Klivans, Caroline J.},
  title   = {The Bergman complex of a matroid and phylogenetic trees},
  journal = {Journal of Combinatorial Theory, Series B},
  volume  = {96},
  year    = {2006},
  pages   = {38--49}
}

@article{BogartJensenSpeyerSturmfelsThomas2007,
  author  = {Bogart, Tristram and Jensen, Anders N. and Speyer, David and Sturmfels, Bernd and Thomas, Rekha R.},
  title   = {Computing tropical varieties},
  journal = {Journal of Symbolic Computation},
  volume  = {42},
  year    = {2007},
  pages   = {54--73}
}

@article{GrossSiebert2011,
  author  = {Gross, Mark and Siebert, Bernd},
  title   = {From real affine geometry to complex geometry},
  journal = {Annals of Mathematics},
  volume  = {174},
  year    = {2011},
  pages   = {1301--1428}
}