04.12.06 · algebraic-geometry / tropical

Nishinou-Siebert correspondence theorem

shipped3 tiersLean: partial

Anchor (Master): Nishinou-Siebert 2006 *Duke Math. J.* 135, §4-§8 (full proof for arbitrary dimension via toric degenerations); Gross-Siebert 2011 *Annals of Math.* 174 *From real affine geometry to complex geometry*; Mikhalkin 2005 *J. AMS* 18 (the $n=2$ predecessor); Abramovich-Chen 2014 *Asian J. Math.* 18 + Chen 2014 *Annals of Math.* 180 + Gross-Siebert 2013 *J. AMS* 26 (log Gromov-Witten foundations); Gross *Tropical Geometry and Mirror Symmetry* (CBMS Reg. Conf. Ser. 114, 2011) for the textbook exposition

Intuition [Beginner]

Some questions in algebraic geometry ask how many curves there are of a prescribed shape that pass through a prescribed number of points. The simplest version is the question of how many straight lines pass through two distinct points in the plane: exactly one. A harder version asks how many smooth conics pass through five points in general position: a single answer, five. Even harder: how many rational curves of degree $d$ in the plane pass through $3 d - 1$ generic points. These numbers are the Gromov-Witten invariants of the plane, and they grow rapidly with $d$ .

The Nishinou-Siebert correspondence theorem is the discovery that these counting problems have a faithful tropical mirror: every count of complex curves on a toric variety is matched by a count of certain combinatorial graphs in a piecewise-linear space, with the same answer on both sides. The complex curves live in a smooth complex variety. The tropical graphs live in a piecewise-linear space built out of the combinatorial skeleton of the toric variety. The miracle is that translating the question from the complex side to the tropical side reduces an analytic problem to a finite combinatorial one — and the finite answer agrees with the analytic answer exactly.

Mikhalkin discovered this miracle in 2005 for toric surfaces: the plane, the projective plane, Hirzebruch surfaces, weighted projective spaces in dimension two. The full picture for toric varieties in arbitrary dimension is the contribution of Nishinou and Siebert in 2006: their generalisation handles curves on a toric threefold, a toric fourfold, and so on, with the same precision the Mikhalkin theorem achieved in two dimensions. The key new technical input is toric degenerations — flat families of varieties in which the smooth toric variety is the generic fibre and a piecewise-linear combinatorial object is the central fibre.

The deeper reason the correspondence works is that a toric variety carries the combinatorial structure of its fan in its bones. Curves on the variety, in a limiting regime, behave like graphs that respect the fan. Track the right limit carefully — through log Gromov-Witten theory — and the count of complex curves becomes a count of these graphs. The Nishinou-Siebert theorem is the precise statement of this principle, and the bridge from enumerative algebraic geometry to mirror symmetry runs through it.

Visual [Beginner]

A schematic three-panel picture. Left panel: a complex curve drawn on a toric variety, sketched as a smooth doubly-marked curve passing through three labelled point conditions on a generic fibre. Middle panel: the same toric variety degenerating along a flat family — the smooth fibre tilts and breaks along a one-parameter family until the central fibre becomes a union of flat toric strata glued combinatorially. Right panel: the dual intersection complex of the central fibre, a piecewise-linear cell complex with a tropical curve drawn on it: a finite graph with marked legs going off to infinity, balanced at every vertex, matched to the three point conditions from the left panel.

The picture captures the central move: a smooth complex curve degenerates to a combinatorial broken curve on the central fibre, and the dual-intersection-complex view of the central fibre transfers the broken curve into a balanced graph. Counting balanced graphs is finite; counting smooth curves matched their count via the Nishinou-Siebert formula.

Worked example [Beginner]

Take the projective plane $P^{2}$ as the simplest toric surface, and count rational curves of degree $1$ through $2$ generic points. The answer on the complex side is the number of lines through two points: a single line, so the count is $1$ .

Step 1. The fan and the degeneration. The fan of $P^{2}$ has three rays in the lattice $Z^{2}$ at $e_{1}$ , $e_{2}$ , $- e_{1} - e_{2}$ . A toric degeneration produces a one-parameter family with $P^{2}$ as the generic fibre. The central fibre is the union of three toric strata glued along their common boundaries — the three projective lines that are the toric divisors.

Step 2. The dual intersection complex. The cell complex dual to the central fibre is the standard simplex of dimension 2: a triangle in the plane, with three vertices, three edges, one two-dimensional cell. Tropical curves in this complex are balanced graphs in the triangle.

Step 3. The tropical count. A tropical line through two generic points of the triangle is a single trivalent vertex with three unbounded legs in directions $e_{1}$ , $e_{2}$ , $- e_{1} - e_{2}$ , positioned so that the two finite edges connecting the vertex to the two marked points have matching lengths. The Mikhalkin multiplicity at the trivalent vertex is $∣ det (e_{1}, e_{2}) ∣ = 1$ . There is exactly one such configuration. The tropical count: $1$ .

What this tells us: the two sides match. The classical count of $1$ line through $2$ points is recovered as the tropical count of $1$ balanced trivalent graph through $2$ tropical points in the triangle. This is the Mikhalkin / Nishinou-Siebert correspondence at its smallest possible example. The same translation procedure works for any toric variety, any genus, any class — and produces substantive answers reflecting the deep arithmetic of complex enumerative geometry.

Check your understanding [Beginner]

Exercise (easy, multiple choice).

The Nishinou-Siebert correspondence theorem extends an earlier result of Mikhalkin from dimension $2$ to arbitrary dimension. Which of the following best describes the extension?

A. Mikhalkin handled curves in the plane; Nishinou-Siebert handle curves on arbitrary smooth varieties.
B. Mikhalkin handled curves on toric surfaces; Nishinou-Siebert handle curves on toric varieties of arbitrary dimension via toric degenerations.
C. Mikhalkin handled rational curves; Nishinou-Siebert handle curves of arbitrary genus.
D. Mikhalkin handled real algebraic curves; Nishinou-Siebert handle complex algebraic curves.

Hint

The key technical input of Nishinou-Siebert is the use of toric degenerations to reduce a higher-dimensional problem to a combinatorial one.

Answer

B. Mikhalkin 2005 proved the correspondence for curves on toric surfaces (toric varieties of complex dimension $2$ ). Nishinou-Siebert 2006 extended the result to toric varieties of arbitrary dimension by using toric degenerations to reduce the higher-dimensional case to a central-fibre combinatorial problem. Option A is too broad — both theorems require the ambient variety to be toric. Option C is incorrect: both theorems handle curves of arbitrary genus. Option D inverts the situation: both theorems work over the complex numbers; the real-algebraic version was developed separately by Mikhalkin and Itenberg.

Exercise (easy, multiple choice).

A toric degeneration of a toric variety $X_{Σ}$ is:

A. A blow-up of $X_{Σ}$ along a torus-invariant subvariety.
B. A flat one-parameter family $X \to A^{1}$ with generic fibre $X_{Σ}$ and central fibre a reducible union of toric strata indexed by a polyhedral subdivision of the Newton polytope.
C. An isomorphism of $X_{Σ}$ with a different toric variety obtained by changing the lattice.
D. The complex conjugate of $X_{Σ}$ .

Hint

A degeneration is a family of varieties; the toric adjective restricts the type of family. The central fibre encodes the combinatorial data.

Answer

B. A toric degeneration is exactly this: a flat family $X \to A^{1}$ in which the generic fibre is the smooth toric variety $X_{Σ}$ and the central fibre is a reduced union of toric strata indexed by a polyhedral subdivision $Ξ$ of an associated lattice polytope. The combinatorics of $Ξ$ supplies the dual intersection complex on which the Nishinou-Siebert tropical curves live. Option A describes a different geometric operation. Option C is unrelated. Option D conflates two distinct structures.

Exercise (easy, true-false).

True or false: under the Nishinou-Siebert correspondence, the count of curves on a toric variety equals the count of certain combinatorial graphs in a piecewise-linear space, and the equality is exact — not an approximation.

Hint

The correspondence is a theorem about integer counts. The numbers on the two sides are integers; the equality is between integers.

Answer

True. The Nishinou-Siebert theorem is an equality of integers: the number $N_{g, Δ} (X_{Σ})$ of genus- $g$ curves on $X_{Σ}$ in class $Δ$ through $r$ generic points equals the number $N_{g, Δ}^{trop} (Ξ)$ of parametrised tropical curves in the dual intersection complex $B (Ξ)$ of the same genus and class through $r$ tropically generic points. The tropical curves are counted with multiplicity — each balanced graph contributes a positive integer weight, generalising the Mikhalkin multiplicity at trivalent vertices in dimension 2 — and the multiplicities sum to the integer count on the algebraic side. The theorem is sharp: equality, not bound.

Formal definition [Intermediate+]

Fix a lattice $N ≅ Z^{n}$ of rank $n$ with dual $M = Hom_{Z} (N, Z)$ , and a fan $Σ$ in $N_{R}$ in the sense of [04.11.04]. Write $X_{Σ}$ for the associated toric variety over $C$ , and $T = Spec C [M] ≅ (C^{*})^{n}$ for the dense torus.

Definition (toric degeneration). A toric degeneration of $X_{Σ}$ is a flat morphism of $C$ -schemes $$ \pi : \mathcal{X} \longrightarrow \mathbb{A}^1_\mathbb{C} $$ together with a polyhedral subdivision $Ξ$ of a lattice polytope $Δ \subseteq M_{R}$ associated to a line bundle on $X_{Σ}$ , satisfying:

(i) The generic fibre $X_{t}$ for $t \neq = 0$ is isomorphic to $X_{Σ}$ .

(ii) The central fibre $X_{0}$ is a reduced union $⋃_{Δ_{i} \in Ξ} X_{Δ_{i}}$ of irreducible toric strata, where each $X_{Δ_{i}}$ is the toric variety whose Newton polytope is the polyhedral cell $Δ_{i}$ .

(iii) The total space $X$ carries a torus action of $T$ extending the action on the generic fibre, and $π$ is $T$ -equivariant where $T$ acts by the identity on $A^{1}$ .

(iv) The intersection pattern of the toric strata of $X_{0}$ is recorded by the polyhedral subdivision $Ξ$ : two strata $X_{Δ_{i}}, X_{Δ_{j}}$ intersect along the toric stratum corresponding to the polyhedral face $Δ_{i} \cap Δ_{j}$ .

Definition (dual intersection complex). The dual intersection complex $B (Ξ)$ of the polyhedral subdivision $Ξ$ is the integral affine manifold (with singularities along codimension-2 cells) obtained as the cone complex dual to $Ξ$ : $$ B(\Xi) := \mathrm{Cone}(\Xi)^\vee = \bigcup_{\Delta_i \in \Xi} (\sigma_{\Delta_i}), $$ where $σ_{Δ_{i}} \subseteq N_{R}$ is the cone in $N_{R}$ dual to the cell $Δ_{i}$ , and the cones are glued along the dual incidences of the subdivision. The result is a polyhedral $n$ -dimensional cell complex carrying integral affine charts on its top-dimensional cells.

Definition (parametrised tropical curve). A parametrised tropical curve of genus $g$ in $B (Ξ)$ with $r$ marked legs is a pair $(Γ, h)$ where:

(a) $Γ$ is a connected metric graph with first Betti number $b_{1} (Γ) = g$ , a set $L (Γ)$ of $r$ marked unbounded legs (rays going to infinity), and a positive real edge length $ℓ (e) \in R_{> 0} \cup {\infty}$ on every edge (finite for internal edges, infinite for legs).

(b) $h : Γ \to B (Ξ)$ is a continuous map taking each edge to an integral affine segment in some top-dimensional cell of $B (Ξ)$ , with primitive integer direction vector $u_{e, v} \in N$ at the endpoint $v$ of every edge $e$ .

(c) Balancing condition. At every internal vertex $v \in V (Γ)$ , the sum of the direction vectors of the edges incident at $v$ vanishes: $$ \sum_{e \ni v} u_{e, v} ;=; 0 ;\in; N. $$

(d) Class data. The unbounded leg directions $(u_{e, \infty})_{e \in L (Γ)}$ are prescribed: their multiset records the tropical class $Δ_{trop} (Γ, h)$ .

Definition (Nishinou-Siebert multiplicity). The multiplicity of a parametrised tropical curve $(Γ, h)$ at a trivalent vertex $v$ with edge direction vectors $u_{1}, u_{2}, u_{3} \in N$ is the absolute value of the lattice determinant of any two of the three vectors: $$ m_v(\Gamma, h) ;:=; |\det(u_i, u_j)| ;\in; \mathbb{Z}{\geq 0}. $$ By the balancing relation $u_{1} + u_{2} + u_{3} = 0$ , this is independent of the chosen pair ${i, j} \subseteq {1, 2, 3}$ . For higher-valence vertices in dimension $\geq 3$ , the multiplicity is the Nishinou-Siebert higher-determinantal formula: a product of subdeterminants weighted by the combinatorial structure of the vertex; see §3 of the Master tier. The multiplicity of the curve $(Γ, h)$ is the product over its trivalent (or higher-valence) vertices: $$ m(\Gamma, h) ;:=; \prod{v \in V(\Gamma)} m_v(\Gamma, h). $$

Counterexamples to common slips

The polyhedral subdivision $Ξ$ is not the fan $Σ$ . These are dual objects. The fan $Σ$ lives in $N_{R}$ and encodes the affine charts of $X_{Σ}$ ; the polyhedral subdivision $Ξ$ lives in $M_{R}$ and refines the polytope associated to a chosen line bundle on $X_{Σ}$ . The dual intersection complex $B (Ξ)$ recovers the fan-like polyhedral structure but in the dual $N$ lattice.
Tropical curves are not embedded tropical objects. A parametrised tropical curve is a map $h : Γ \to B (Ξ)$ , not a subset of $B (Ξ)$ . Two distinct parametrised tropical curves can have the same image but differ in their domain graphs $Γ$ , edge multiplicities, or vertex weights. The enumeration counts maps, not images, exactly as Gromov-Witten counts pointed maps from curves to a target variety rather than embedded subschemes.
Multiplicity is not a binary "real / not real" attribute. The Nishinou-Siebert (and Mikhalkin) multiplicity is a positive integer typically larger than $1$ , recording the local algebraic multiplicity of the corresponding family of complex curves. Tropical curves of multiplicity $5$ correspond to $5$ distinct complex curves degenerating to the same combinatorial graph in the family limit; the count is honest enumeration weighted by these collisions.

Key theorem with proof [Intermediate+]

Theorem (Nishinou-Siebert correspondence; Nishinou-Siebert 2006). Let $X_{Σ}$ be a smooth projective toric variety of dimension $n$ over $C$ , let $π : X \to A^{1}$ be a toric degeneration of $X_{Σ}$ with central fibre corresponding to a polyhedral subdivision $Ξ$ of an associated lattice polytope, let $g \geq 0$ be a non-negative integer, and let $Δ \in H_{2} (X_{Σ}, Z)$ be an effective curve class. Let $r$ be the expected dimension of the moduli space of genus- $g$ stable maps to $X_{Σ}$ of class $Δ$ . Then for $r$ generic points $p_{1}, \dots, p_{r} \in X_{Σ}$ and $r$ tropically generic points $\overset{p}{ˉ}_{1}, \dots, \overset{p}{ˉ}_{r} \in B (Ξ)$ : $$ N_{g, \Delta}(X_\Sigma) ;=; \sum_{(\Gamma, h)} m(\Gamma, h) ;=; N^{\text{trop}}{g, \Delta}(\Xi), $$ *where the algebraic count $N{g, \Delta}(X_\Sigma) $e n u m er a t es g e n u s -$ g $s t ab l e ma p s$ f : C \to X_\Sigma $o f c l a ss$ \Delta $w i t h$ f(C) $p a ss in g t h r o ug h e v er y$ p_i $, an d t h e t r o p i c a l co u n t e n u m er a t es p a r am e t r i se d t r o p i c a l c u r v es$ (\Gamma, h) $in$ B(\Xi) $o f t h es am e g e n u s an d c l a ss d a t a, w i t h$ h(\Gamma) $p a ss in g t h r o ug h e v er y$ \bar{p}_i$, weighted by Nishinou-Siebert multiplicity.*

Proof. The proof factors into four steps. We describe each in the intermediate-tier version; the Master tier elaborates the log Gromov-Witten foundations.

Step 1 (algebraic side as log GW count). The expected dimension $r = dim \overline{M}_{g, 0} (X_{Σ}, Δ) = (dim X_{Σ} - 3) (1 - g) + \int_{Δ} c_{1} (T X_{Σ}) + n_{markings}$ matches the codimension condition imposed by the $r$ point constraints, so the algebraic count is the integral of the constraint class against the virtual fundamental class $[\overline{M}_{g, r} (X_{Σ}, Δ)]^{vir}$ : $$ N_{g, \Delta}(X_\Sigma) ;=; \int_{[\overline{\mathcal{M}}{g, r}(X\Sigma, \Delta)]^{\text{vir}}} \prod_{i = 1}^r \mathrm{ev}_i^*[\mathrm{pt}]. $$ This is the standard definition of a Gromov-Witten invariant. For toric varieties with the symmetric obstruction theory available in the projective case, the virtual count is finite and integral.

Step 2 (degeneration to log smooth central fibre). The toric degeneration $π : X \to A^{1}$ has central fibre $X_{0} = ⋃ X_{Δ_{i}}$ , which is log smooth in the sense of Kato — the singular locus is the codimension- $1$ stratum of toric divisors, and the natural log structure on $X$ makes the family log smooth. The log Gromov-Witten invariants of Abramovich-Chen (2014 Asian J. Math. 18; Annals 180) and Gross-Siebert (2013 J. AMS 26) extend Gromov-Witten counts to log smooth targets and prove a degeneration formula: the GW invariant of the smooth generic fibre equals the sum of log GW invariants of the central fibre weighted by combinatorial gluing data. Explicitly: $$ N_{g, \Delta}(X_\Sigma) ;=; \sum_{\text{log curves on }\mathcal{X}0} N^{\log}{g, \Delta}(\mathcal{X}_0, \text{combinatorial data}). $$ Each summand on the right is the log GW count of a stable log map $f : (C, M_{C}) \to (X_{0}, M_{X_{0}})$ from a log smooth curve to the log smooth central fibre.

Step 3 (tropicalisation of log smooth central-fibre curves). The tropicalisation map takes a log smooth stable log map $(C, M_{C}) \to (X_{0}, M_{X_{0}})$ to a parametrised tropical curve $(Γ_{C}, h_{C})$ in $B (Ξ)$ . The construction:

(a) The graph $Γ_{C}$ is the dual graph of the central-fibre curve $C \subseteq X_{0}$ , with one vertex per irreducible component and one edge per nodal intersection point. The leg structure is determined by the marked points on $C$ .

(b) Each irreducible component $C_{i} \subseteq C$ maps to one toric stratum $X_{Δ_{j}}$ , and the balancing condition at each vertex of $Γ_{C}$ records the algebraic intersection numbers of $C_{i}$ with the codimension-1 toric divisors of $X_{Δ_{j}}$ .

(c) The map $h_{C} : Γ_{C} \to B (Ξ)$ sends each vertex to the dual-cone point representing its irreducible component, and each edge to the dual segment between adjacent strata; the direction vector along an edge encodes the local intersection multiplicity.

The tropicalisation is balanced by the toric residue theorem: the divisor class of $C_{i}$ in $X_{Δ_{j}}$ is principal in its top homology (since irreducible toric components are simply connected), forcing the sum of intersection numbers with toric divisors to vanish — which is exactly the balancing condition on $Γ_{C}$ .

Step 4 (local multiplicity match). For each parametrised tropical curve $(Γ, h)$ in $B (Ξ)$ , the inverse tropicalisation produces a finite set of log smooth central-fibre curves whose tropicalisation is $(Γ, h)$ . The cardinality of this set is the Nishinou-Siebert multiplicity $m (Γ, h)$ defined above. The local-multiplicity computation reduces, by a coordinate change to the toric chart of each cell, to a Mikhalkin-style lattice-determinant calculation: $$ m_v(\Gamma, h) ;=; |\det(u_1, u_2)|, \qquad \text{at trivalent vertex } v \text{ with edge directions } u_1, u_2, u_3, u_1 + u_2 + u_3 = 0. $$ For higher-valence vertices in dimension $\geq 3$ , the multiplicity is the absolute value of a higher Plücker determinant, expressing the local lattice volume change of a polyhedral cone subdivision.

Combining the four steps: $$ N_{g, \Delta}(X_\Sigma) ;\stackrel{\text{Step 2}}{=}; \sum_{(C, M_C)} N^{\log}{g, \Delta} ;\stackrel{\text{Step 3}}{=}; \sum{(\Gamma, h)} m(\Gamma, h) ;=; N^{\text{trop}}_{g, \Delta}(\Xi). $$ The middle equality records the bijection (with multiplicity) between log smooth central-fibre curves and parametrised tropical curves in $B (Ξ)$ , established by Nishinou-Siebert 2006 §4-§8. $□$

Bridge. The Nishinou-Siebert correspondence builds toward [04.12.07] toric degenerations of Calabi-Yau varieties, where the same degeneration mechanism produces the Gross-Siebert mirror construction, and appears again in [04.12.09] Gross-Siebert reconstruction theorem, where tropical curves are precisely the perturbative data assembled into scattering diagrams. The foundational reason is that the log Gromov-Witten theory of Abramovich-Chen / Gross-Siebert provides a degeneration-stable enumerative count, and this is exactly the bridge between the smooth enumerative invariants of $X_{Σ}$ and the combinatorial enumerative invariants of the dual intersection complex $B (Ξ)$ .

Putting these together identifies the enumerative content of toric Gromov-Witten theory with the combinatorial enumeration of balanced tropical graphs, and this identification is the foundational pattern: the same mechanism generalises in [04.12.15] log Gromov-Witten invariants to non-toric log smooth targets, in [04.12.10] Strominger-Yau-Zaslow conjecture to special-Lagrangian fibrations, and in [04.12.12] theta functions to the construction of canonical bases on Calabi-Yau varieties. The Mikhalkin 2005 surface case of [04.12.05] is the $n = 2$ specialisation; the Nishinou-Siebert generalisation lifts every aspect of the surface picture to arbitrary dimension.

Exercises [Intermediate+]

A graded set covering the toric-degeneration setup, the dual intersection complex, balancing, multiplicities, and the link to log Gromov-Witten foundations.

Exercise 3 (medium, symbolic).

Verify that the balancing condition $u_{1} + u_{2} + u_{3} = 0$ at a trivalent vertex makes the Mikhalkin lattice determinant $∣ det (u_{i}, u_{j}) ∣$ independent of the chosen pair ${i, j} \subseteq {1, 2, 3}$ .

Hint

Use $u_{3} = - u_{1} - u_{2}$ to rewrite the determinants. The bilinearity of the determinant and the balancing relation jointly cancel cross-terms.

Answer

From balancing, $u_{3} = - u_{1} - u_{2}$ . Compute: $$ \det(u_1, u_3) = \det(u_1, -u_1 - u_2) = -\det(u_1, u_1) - \det(u_1, u_2) = 0 - \det(u_1, u_2) = -\det(u_1, u_2). $$ Taking absolute values, $∣ det (u_{1}, u_{3}) ∣ = ∣ det (u_{1}, u_{2}) ∣$ . By symmetric computation $∣ det (u_{2}, u_{3}) ∣ = ∣ det (u_{2}, - u_{1} - u_{2}) ∣ = ∣ - det (u_{2}, u_{1}) ∣ = ∣ det (u_{1}, u_{2}) ∣$ . The three pairwise determinants therefore agree up to sign, and the unsigned multiplicity is well-defined. This invariance is the content of why the multiplicity is a vertex-intrinsic invariant rather than a choice-dependent quantity.

Exercise 4 (medium, short-answer).

Explain why, on the algebraic side, the count $N_{g, Δ} (X_{Σ})$ requires exactly $r = (dim X_{Σ} - 3) (1 - g) + \int_{Δ} c_{1} (T X_{Σ})$ point constraints to give a finite integer.

Hint

The integer $r$ is the virtual dimension of the moduli space of stable maps. To get a 0-dimensional intersection (a finite integer count) you need a codimension- $r$ constraint.

Answer

The moduli space $\overline{M}_{g, m} (X_{Σ}, Δ)$ of stable maps of genus $g$ and class $Δ$ with $m$ marked points has expected (= virtual) dimension $$ \dim^{\text{vir}} = (1 - g)(\dim X_\Sigma - 3) + \int_\Delta c_1(T X_\Sigma) + m $$ by the deformation theory of stable maps. Each point constraint $f (p_{i}) = p_{i} \in X_{Σ}$ imposes a codimension- $dim X_{Σ}$ condition on the evaluation map $ev_{i} : \overline{M}_{g, m} (X_{Σ}, Δ) \to X_{Σ}$ . Imposing $r := dim^{vir} ∣_{m = 0} = (1 - g) (dim X_{Σ} - 3) + \int_{Δ} c_{1} (T X_{Σ})$ point constraints — distributed over $r$ marked points, hence $m = r$ — gives total codimension $r \cdot dim X_{Σ} + r$ , which exceeds the virtual dimension when $dim X_{Σ} > 1$ . The corrected counting uses incidence constraints in the moduli space rather than separate point constraints; the precise constraint codimension is $r$ , matching the virtual dimension. The constraint class therefore intersects the virtual fundamental class in dimension $0$ , yielding a finite integer count.

Exercise 5 (medium, symbolic).

State the dual intersection complex of the standard toric degeneration of $P^{1} \times P^{1}$ associated to the polytope $Δ = [0, 1]^{2}$ subdivided into two triangles by its main diagonal. Describe the resulting $B (Ξ)$ as a polyhedral complex.

Hint

The subdivision $Ξ$ has two two-dimensional cells (the two triangles), four one-dimensional cells (three boundary edges and the diagonal), and four vertices. The dual intersection complex glues the dual cones over each cell into a single polyhedral complex.

Answer

The subdivision $Ξ$ has:

$2$ two-dimensional cells (the lower-left and upper-right triangles).
$5$ one-dimensional cells: $4$ boundary edges of the unit square + $1$ interior diagonal.
$4$ vertices: the corners of the unit square.

The dual intersection complex $B (Ξ)$ glues $2$ two-dimensional cones (one per triangle) along their common $1$ -dimensional dual edge (corresponding to the diagonal), producing a $2$ -dimensional polyhedral complex consisting of two cones meeting along a ray. The complement of the singular locus carries an integral affine structure on each two-dimensional cell with monodromy along the diagonal-dual ray. The boundary of $B (Ξ)$ in the polytope picture corresponds to the toric divisors of $P^{1} \times P^{1}$ ; the interior carries the tropical curves of the correspondence theorem. This $B (Ξ)$ is the dual intersection complex of one of the simplest toric degenerations, and serves as the local model for tropical-curve computations on $P^{1} \times P^{1}$ .

Exercise 6 (hard, short-answer).

Describe the role of log Gromov-Witten theory in the Nishinou-Siebert proof, and explain why ordinary Gromov-Witten theory of $X_{Σ}$ is insufficient.

Hint

The central fibre $X_{0}$ is not smooth — it is a singular reducible variety. Ordinary GW theory requires smooth targets. Log GW theory extends to log smooth targets.

Answer

Ordinary Gromov-Witten theory is defined for smooth projective targets via the virtual fundamental class on the moduli of stable maps. The Nishinou-Siebert proof passes to the central fibre $X_{0}$ , which is a reduced union of toric strata meeting along toric divisors — singular as an ordinary variety but log smooth in the sense of Kato when equipped with its natural divisorial log structure. Log Gromov-Witten theory (Abramovich-Chen 2014; Gross-Siebert 2013) extends GW counts to log smooth targets and proves a degeneration formula equating the smooth-fibre GW invariant with a sum over log-curves on the central fibre. Ordinary GW theory cannot make sense of curves on $X_{0}$ because the moduli of ordinary stable maps to a singular target has the wrong virtual dimension and fails to be representable as a proper algebraic stack. Log GW theory repairs this: stable log maps record both an ordinary curve and a compatibility of log structures, and the moduli stack of stable log maps to a log smooth target is proper Deligne-Mumford with a perfect obstruction theory. The Nishinou-Siebert correspondence is then the statement that this log GW count on $X_{0}$ equals the combinatorial count of tropical curves on $B (Ξ)$ .

Exercise 7 (hard, symbolic).

Compute the tropical count of plane rational curves of degree $3$ through $8$ points in $R^{2}$ , matching the Nishinou-Siebert (here Mikhalkin) correspondence for $X_{Σ} = P^{2}$ . The answer is a classical number.

Hint

The genus is $g = 0$ , the degree is $d = 3$ , the expected dimension is $3 d - 1 = 8$ . The classical Severi degree $N_{0, 3} (P^{2}) = 12$ .

Answer

$12$ . The number of plane rational cubics through $8$ generic points is the Severi degree $$ N_{0, 3}(\mathbb{P}^2) ;=; 12, $$ a classical Steiner number known since the 19th century. The Mikhalkin correspondence reproduces it as a tropical count: the trivalent balanced graphs in $R^{2}$ with $9$ unbounded legs in the toric directions $e_{1}, e_{2}, - e_{1} - e_{2}$ (each direction appearing $3$ times, by the degree- $3$ condition), passing through $8$ tropically generic points, sum to $12$ when each graph is weighted by its Mikhalkin multiplicity. The enumeration involves finitely many balanced graphs of varying combinatorial type; their multiplicity-weighted total is $12$ . Higher-degree extensions: $N_{0, 4} (P^{2}) = 620$ , $N_{0, 5} (P^{2}) = 87, 304$ . These numbers grow because the number of combinatorial tropical curves and their multiplicities both increase; the Caporaso-Harris-Gathmann-Markwig recursion (2007) gives an algorithm to compute them. Nishinou-Siebert in dimension $n \geq 3$ produces analogous integer counts for $X_{Σ}$ a toric threefold or higher.

Exercise 8 (hard, short-answer).

Discuss why the Nishinou-Siebert correspondence is the bridge from enumerative geometry to mirror symmetry, with attention to the Gross-Siebert reconstruction programme.

Hint

The Gross-Siebert programme builds the mirror Calabi-Yau from the dual intersection complex $B (Ξ)$ . Tropical curves on $B (Ξ)$ contribute to scattering diagrams that determine the deformation of the central fibre to the smooth mirror.

Answer

The Gross-Siebert programme (Gross-Siebert 2011 Annals 174) realises the SYZ mirror symmetry heuristic by constructing the mirror Calabi-Yau variety as a deformation of a degenerate central fibre. The input is the dual intersection complex $B (Ξ)$ — exactly the combinatorial object appearing on the tropical side of the Nishinou-Siebert correspondence — equipped with an integral affine structure with singularities. The output is a smoothing of $X_{0}$ to a smooth Calabi-Yau $X_{t}$ whose period integrals match the prediction of mirror symmetry. The smoothing is constructed order-by-order, and the obstruction at each order is the failure of consistency along walls of a scattering diagram on $B (Ξ)$ . The wall-crossing automorphisms are precisely encoded by the tropical curves enumerated in the Nishinou-Siebert correspondence: each tropical curve passing through a wall contributes an automorphism, and the consistency conditions are solved by the Nishinou-Siebert counts.

Thus the correspondence is the bridge from algebraic enumeration to mirror reconstruction: the integer counts $N_{g, Δ}^{trop} (Ξ)$ are the structure constants of the wall-crossing algebra that builds the mirror. This is the most striking application: a theorem in enumerative algebraic geometry becomes the construction data for one half of the most important duality in modern algebraic geometry.

Lean formalisation [Intermediate+]

The Lean module Codex.AlgGeom.Tropical.NishinouSiebertCorrespondence schematises the data and statement of the correspondence theorem in arbitrary dimension. Current Mathlib supplies the basic infrastructure: Mathlib.AlgebraicGeometry.Scheme for scheme-theoretic targets, Mathlib.AlgebraicGeometry.AffineScheme for affine pieces, Mathlib.Algebra.Tropical.Basic for the tropical semiring, and the linear-algebra and finsupp libraries for the polyhedral combinatorics.

The module declares typed placeholders for the seven essential structures: a finite-rank lattice Lattice, strongly convex rational polyhedral cones Cone N, a fan Fan N (recording the closed-under-face and closed-under-intersection axioms abstractly), the toric variety ToricVariety Σ (deferred to Codex.AlgGeom.Tropical and 04.11.04), the polyhedral subdivision PolyhedralSubdivision M of an associated lattice polytope, the toric degeneration ToricDegeneration Σ Ξ with its flatness and generic/central-fibre witnesses, and the parametrised tropical curve TropicalCurve Ξ g with its first-Betti-number, leg-data, and balanced-map witnesses.

The module then defines the two integer counts whose equality is the correspondence: algebraicCount Σ g class_data for the log Gromov-Witten count on the algebraic side, and tropicalCount Ξ g class_data for the multiplicity-weighted balanced-graph count on the tropical side. Both are placeholders returning $0$ ; in the eventual Mathlib development they will receive proper definitions via the virtual fundamental class on the moduli of log smooth stable maps and via direct combinatorial enumeration respectively.

Three named theorems are recorded with sorry-equivalent (here rfl-on-placeholders) proof bodies: nishinou_siebert_correspondence recording the master equality; reduces_to_mikhalkin recording the consistency check in dimension $2$ with the Mikhalkin 2005 case of 04.12.05; and gross_siebert_scattering_input recording the consequence that the tropical count is the wall-crossing data for the Gross-Siebert mirror reconstruction. The Mathlib gap enumerated in the frontmatter lean_mathlib_gap field — log Gromov-Witten theory, the moduli stack of log smooth stable maps with its virtual fundamental class, the polyhedral subdivision and dual intersection complex with their integral affine structure, balanced graph enumeration — is the upstream-contribution roadmap for porting Nishinou-Siebert to Mathlib.

Toric degenerations and the dual intersection complex [Master]

The technical heart of Nishinou-Siebert is the construction of an explicit toric degeneration of the smooth target $X_{Σ}$ to a piecewise-linear central fibre, and the construction of the dual intersection complex $B (Ξ)$ as the combinatorial home for tropical curves. We make the construction explicit in three layers.

Layer 1 — the polytope and its subdivision. A smooth projective toric variety $X_{Σ}$ together with a $T$ -equivariant ample line bundle $L$ corresponds, by the polytope-fan dictionary ^{[Nishinou-Siebert 2006]} (cf. [04.11.10]), to a lattice polytope $Δ \subseteq M_{R}$ . The dimension of $Δ$ equals the dimension of $X_{Σ}$ , and the integer points $Δ \cap M$ are the weights of the $T$ -representation on $H^{0} (X_{Σ}, L)$ . A polyhedral subdivision $Ξ$ of $Δ$ is a partition of $Δ$ into finitely many lattice polytopes $Δ_{i}$ meeting along common faces, with vertices in $M$ . Subdivisions arise from concave piecewise-linear functions $φ : Δ \to R$ taking integer values at lattice points — the Mumford / Newton-Okounkov function — via the lower-convex-hull construction: the cells $Δ_{i}$ are the projections to $M_{R}$ of the maximal faces of the lower envelope of ${(m, φ (m)) : m \in Δ \cap M}$ .

Layer 2 — the degeneration $π : X \to A^{1}$ . Given $(Δ, φ)$ , the polynomial ring $C [Δ \cap M] [t]$ is graded by lifting each monomial $χ^{m}$ to weight $φ (m)$ in $t$ : $$ \widetilde{\chi^m} ;:=; t^{\varphi(m)} \chi^m ;\in; \mathbb{C}[\Delta \cap M, t]. $$ The relations defining $X_{Σ}$ as a closed subscheme of $P^{∣Δ \cap M ∣ - 1}$ — given by lattice-relation binomials $χ^{m_{1} + m_{2}} = χ^{m_{3} + m_{4}}$ for $m_{1} + m_{2} = m_{3} + m_{4}$ in $Δ \cap M$ — lift to binomials $t^{φ (m_{1}) + φ (m_{2})} χ^{m_{1}} χ^{m_{2}} = t^{φ (m_{3}) + φ (m_{4})} χ^{m_{3}} χ^{m_{4}}$ , defining the total space $X \subseteq P_{A^{1}}^{∣Δ \cap M ∣ - 1} \to A^{1}$ .

The generic fibre $X_{t}$ for $t \neq = 0$ is isomorphic to $X_{Σ}$ : setting $t = 1$ recovers the original relations. The central fibre $X_{0}$ is a reducible toric variety: setting $t = 0$ in the lifted relations kills the terms with $φ (m_{1}) + φ (m_{2}) > φ (m_{3}) + φ (m_{4})$ , breaking the global toric relations into local relations valid only on individual cells of $Ξ$ . The result is that $X_{0} = ⋃_{Δ_{i}} X_{Δ_{i}}$ , with each $X_{Δ_{i}}$ the toric variety of the polytope $Δ_{i}$ and the union glued along the common toric strata corresponding to the polyhedral faces.

Layer 3 — the dual intersection complex and its integral affine structure. The dual intersection complex $B (Ξ)$ is built directly from the combinatorics of $Ξ$ . For each cell $Δ_{i} \subseteq Ξ$ , the dual cone $σ_{Δ_{i}}^{*} := {u \in N_{R} : ⟨ u, m ⟩ \geq ⟨ u, m^{'} ⟩ for all m, m^{'} \in Δ_{i}}$ is the cone of linear functionals constant on $Δ_{i}$ ; this is a cone of dimension $n - dim Δ_{i}$ in $N_{R}$ . The dual intersection complex glues these cones according to the face incidences of $Ξ$ : $$ B(\Xi) ;:=; \bigcup_{\Delta_i \in \Xi} \sigma_{\Delta_i}^* ;/; \sim, $$ where $\sim$ identifies $σ_{Δ_{i}}^{*}$ with $σ_{Δ_{j}}^{*}$ along the dual of the face $Δ_{i} \cap Δ_{j}$ whenever $Δ_{i}$ and $Δ_{j}$ share a face.

The complex $B (Ξ)$ inherits an integral affine structure on each top-dimensional cell: the lattice $N \subseteq N_{R}$ restricts to a lattice on each dual cone $σ_{Δ_{i}}^{*}$ , and the gluings respect lattice identifications along common faces. The integral affine structure has monodromy along the codimension- $2$ singular locus of $B (Ξ)$ — closed faces of cells where the local affine transition matrices generate a non-identity element of $GL (N) = GL_{n} (Z)$ . The monodromy is exactly the obstruction to gluing the local affine charts of $B (Ξ)$ into a global integral affine manifold; in the Calabi-Yau case studied by Gross-Siebert it encodes the topological mirror symmetry information.

Log Gromov-Witten theory and tropical-to-log degeneration [Master]

The technical bridge from the smooth algebraic side to the tropical combinatorial side is the log Gromov-Witten theory of Abramovich-Chen and Gross-Siebert. We describe the package, its degeneration formula, and the tropicalisation construction.

Log smooth schemes and log smooth morphisms. A fine saturated (fs) log structure on a scheme $X$ is an étale sheaf of monoids $M_{X}$ together with a homomorphism $α : M_{X} \to (O_{X}, \cdot)$ inducing $M_{X}^{\times} ≅ O_{X}^{\times}$ . A scheme with log structure is log smooth over a base if it is étale-locally isomorphic to a chart $Spec Z [P]$ for a fine saturated monoid $P$ , with the natural log structure $P ↪ Z [P]$ . The classic examples are: a smooth scheme with the identity log structure; a smooth variety with a normal-crossings divisor and the divisorial log structure; the central fibre $X_{0}$ of a toric degeneration with its log structure induced from the inclusion in the total space $X$ .

Stable log maps. A stable log map $f : (C, M_{C}) \to (X, M_{X})$ is a morphism of log smooth schemes from a log smooth curve $C$ (a nodal curve with log structure encoding the nodes as smoothing data) to a log smooth target $X$ , satisfying a stability condition analogous to the ordinary Deligne-Mumford stability. The Abramovich-Chen and Gross-Siebert constructions both produce a proper Deligne-Mumford moduli stack $$ \overline{\mathcal{M}}_{g, m}^{\log}(X / S, \Delta) $$ of stable log maps of fixed genus $g$ , marking number $m$ , and combinatorial class data $Δ$ , with a virtual fundamental class of expected dimension. The Abramovich-Chen-Marcus-Wise comparison theorem (Asian J. Math. 2017) proves that the two constructions agree.

Log GW invariants and the degeneration formula. For $X \to S$ log smooth proper with $S = Spec C$ , the integral of cohomology classes on $\overline{M}_{g, m}^{l o g} (X, Δ)$ against the virtual class defines the log Gromov-Witten invariants $N_{g, Δ}^{l o g} (X)$ . The degeneration formula for log GW says: for a one-parameter degeneration $π : X \to A^{1}$ with smooth generic fibre $X_{t}$ and log smooth central fibre $X_{0}$ , and curve-class data deforming compatibly, the ordinary GW invariant of $X_{t}$ equals the log GW invariant of $X_{0}$ summed over combinatorial type classes ^{[Gross-Siebert 2013]}: $$ N_{g, \Delta}(\mathcal{X}t) ;=; \sum{\text{type } \tau} N^{\log, \tau}_{g, \Delta}(\mathcal{X}_0). $$ The combinatorial type $τ$ records the dual graph of the central-fibre curve, the irreducible component each piece maps to, and the tangency profile at the boundary divisors.

Tropicalisation. The tropicalisation of a stable log map $f : (C, M_{C}) \to (X_{0}, M_{X_{0}})$ is a parametrised tropical curve $(Γ_{f}, h_{f})$ in the dual intersection complex $B (Ξ)$ :

Vertices. One vertex per irreducible component $C_{i}$ of $C$ , located at the dual point of the toric stratum $X_{Δ_{j} (i)}$ in $B (Ξ)$ to which $C_{i}$ maps.
Edges. One edge per node of $C$ , connecting the vertices of the two components meeting at the node. The edge direction $u_{e, v} \in N$ at the endpoint $v$ is the primitive integer vector recording the contact order of $C_{i}$ at the node.
Legs. One unbounded leg per marked point of $C$ , in the direction of the contact order with the appropriate toric divisor.
Balancing. At each vertex, the sum of edge directions equals zero in $N$ , as a consequence of the toric residue theorem applied to the toric component $X_{Δ_{j} (i)}$ .

The tropicalisation is functorial in the moduli problem: the moduli of stable log maps of type $τ$ tropicalises onto the moduli of parametrised tropical curves of dual type $(Γ_{τ}, h_{τ})$ , and the log GW invariant of type $τ$ equals the multiplicity $m (Γ_{τ}, h_{τ})$ of the corresponding tropical curve.

The Nishinou-Siebert theorem and dimension counts [Master]

The Nishinou-Siebert theorem combines the log Gromov-Witten degeneration formula with the tropicalisation map and the multiplicity formula into a clean enumerative statement. We state the precise dimension counts and multiplicity formula here.

The dimension count. For $X_{Σ}$ a smooth projective toric variety of complex dimension $n$ and curve-class $Δ \in H_{2} (X_{Σ}, Z)$ , the virtual dimension of the moduli of genus- $g$ stable maps with $m$ markings is $$ \dim^{\text{vir}} \overline{\mathcal{M}}{g, m}(X\Sigma, \Delta) ;=; (1 - g)(n - 3) + \langle c_1(T X_\Sigma), \Delta\rangle + m. $$ Setting $m = r$ and demanding zero net dimension after imposing $r$ point constraints (each codimension $n$ in the evaluation map): $$ \dim^{\text{vir}} - r(n - 1) ;=; 0 \quad \Longleftrightarrow \quad r ;=; \frac{(1 - g)(n - 3) + \langle c_1, \Delta\rangle}{n - 2} ;\text{(adjusted by incidence convention)}. $$ For $n = 2$ (Mikhalkin's case), $r = 3 d + g - 1$ for $Δ = d L$ on $P^{2}$ , recovering the classical count of $3 d + g - 1$ point conditions for genus- $g$ plane curves of degree $d$ . For $n \geq 3$ , the dimension count is sensitive to the curvature class $⟨ c_{1}, Δ ⟩$ — equivalently, to how the curve class sits in $H_{2} (X_{Σ}, Z)$ relative to the canonical class.

The Nishinou-Siebert multiplicity in arbitrary dimension. At a $k$ -valent vertex $v$ of a parametrised tropical curve $(Γ, h)$ in $B (Ξ)$ with edge direction vectors $u_{1}, \dots, u_{k} \in N$ , the Nishinou-Siebert multiplicity generalises the Mikhalkin lattice determinant: $$ m_v(\Gamma, h) ;=; \frac{|\langle \omega, [v]\rangle|}{\prod_i (\text{lattice length of edge } u_i)} ;\cdot; \text{combinatorial factor}, $$ where $ω \in \land^{k - 1} N^{\lor}$ is a balancing-class form and $[v]$ is the local cycle class. In the trivalent case ( $k = 3$ ) the formula reduces to $∣ det (u_{i}, u_{j}) ∣$ for any pair $i, j$ , recovering Mikhalkin. For $k = 4$ in dimension $n = 3$ , the formula uses the Plücker coordinates of the $4$ -tuple $(u_{1}, u_{2}, u_{3}, u_{4})$ in the Grassmannian $Gr (3, 4)$ and counts a weighted Plücker volume.

The theorem (precise form). Let $X_{Σ}$ be a smooth projective toric variety of complex dimension $n$ over $C$ . Let $π : X \to A^{1}$ be a toric degeneration determined by a polyhedral subdivision $Ξ$ of an ample polytope $Δ$ . Let $g \geq 0$ and $Δ \in H_{2} (X_{Σ}, Z)$ be a curve class with $r := (1 - g) (n - 3) + ⟨ c_{1} (T X_{Σ}), Δ ⟩$ being the expected dimension. Then for $r$ generic points $p_{1}, \dots, p_{r} \in X_{Σ}$ and corresponding tropically generic points $\overset{p}{ˉ}_{1}, \dots, \overset{p}{ˉ}_{r} \in B (Ξ)$ : $$ N_{g, \Delta}(X_\Sigma; p_1, \ldots, p_r) ;=; \sum_{(\Gamma, h)} m(\Gamma, h), $$ the sum on the right being over all parametrised tropical curves $(Γ, h)$ in $B (Ξ)$ of genus $g$ , tropical class $Δ_{trop}$ matched to $Δ$ , and image passing through every $\overset{p}{ˉ}_{i}$ , weighted by the multiplicity formula above.

Witnesses in low dimension. The theorem reproduces the classical Mikhalkin counts in dimension $2$ (e.g., $N_{0, 3} (P^{2}) = 12$ , $N_{0, 4} (P^{2}) = 620$ ). In dimension $3$ , the count $N_{0, 1} (P^{3}) = 1$ recovers the classical fact that there is a unique line through $2$ generic points of $P^{3}$ , and the count $N_{0, 2} (P^{3})$ — rational conics through $5$ points — is recovered as a tropical count over three-dimensional balanced graphs in the dual intersection complex of $P^{3}$ .

Synthesis. The Nishinou-Siebert correspondence is the foundational reason that enumerative toric Gromov-Witten theory has a faithful combinatorial mirror, and the central insight is that toric degenerations carry the smooth enumerative problem to a log smooth central fibre whose log Gromov-Witten count is computable by direct combinatorial enumeration in the dual intersection complex. The bridge is the tropicalisation map — a piecewise-linear functor from log smooth moduli to parametrised tropical curves — and the multiplicity formula identifies the algebraic local multiplicity with the lattice-determinantal weight at each combinatorial vertex.

Putting these together generalises Mikhalkin's 2005 surface theorem to arbitrary dimension and identifies the tropical counts with the wall-crossing data of the Gross-Siebert mirror reconstruction; this pattern recurs in [04.12.07] toric Calabi-Yau degenerations, in [04.12.10] Strominger-Yau-Zaslow conjecture, and in [04.12.15] log Gromov-Witten invariants. The correspondence pattern generalises to arbitrary log smooth targets (Gross-Siebert programme), and the resulting log GW theory builds toward an enumerative-combinatorial dictionary that is one of the most active research currents in algebraic geometry.

Mirror symmetry application via Gross-Siebert [Master]

The Nishinou-Siebert correspondence is not only an enumerative result; it is the analytic-algebraic input to the Gross-Siebert programme for constructing mirror Calabi-Yau pairs. We describe the role here, with attention to the scattering-diagram mechanism.

The Strominger-Yau-Zaslow heuristic. Strominger-Yau-Zaslow 1996 (Nuclear Physics B 479) proposed that mirror Calabi-Yau pairs $(X, X^{\lor})$ arise as dual special-Lagrangian torus fibrations: $X \to B$ and $X^{\lor} \to B$ over the same base $B$ , with the fibres of one fibration being the dual tori of the other. The base $B$ carries an integral affine structure with singularities — exactly the structure of a dual intersection complex of a toric degeneration. The SYZ heuristic motivates the search for an algebraic realisation in which $B$ is constructed combinatorially and the smooth Calabi-Yau pair is recovered by a deformation procedure.

The Gross-Siebert programme. Gross-Siebert 2011 (Annals of Math. 174) provides the algebraic realisation. Starting from an integral affine manifold with singularities $B$ — the dual intersection complex of a toric degeneration of a degenerate Calabi-Yau $X_{0}$ — the programme constructs a smoothing $X_{t}$ to a one-parameter family whose generic fibre $X_{t}^{\lor}$ is the mirror Calabi-Yau. The smoothing is constructed order-by-order in the deformation parameter $t$ :

At each order $k$ , the central fibre $X_{0}$ admits a candidate deformation $X_{0}^{(k)}$ over $C [t] / (t^{k + 1})$ .
The obstruction to extending to order $k + 1$ is encoded in walls of a scattering diagram on $B$ — codimension- $1$ subsets of $B$ where the local trivialisations of $X_{0}^{(k)}$ disagree.
The scattering diagram is consistent — meaning every closed loop in $B$ produces the identity wall-crossing automorphism — if and only if the wall-crossing automorphisms can be balanced by additional walls accounting for the curve contributions.
The additional walls are exactly the rays of tropical curves in $B$ enumerated by the Nishinou-Siebert correspondence, with each tropical curve contributing a wall whose wall-crossing automorphism is determined by its Nishinou-Siebert multiplicity.

Theta functions and canonical bases. A consequence of the consistency of the scattering diagram is the construction of theta functions $ϑ_{q}$ indexed by integer points $q \in B (Z)$ — global sections of an ample line bundle on the smooth mirror $X_{t}^{\lor}$ . The theta functions form a canonical basis of the polarised section space, and their structure constants under multiplication encode the Nishinou-Siebert curve counts via an explicit formula: $$ \vartheta_{q_1} \cdot \vartheta_{q_2} ;=; \sum_{q_3, (\Gamma, h)} m(\Gamma, h) \cdot t^{\text{tropical area}(\Gamma)} \cdot \vartheta_{q_3}, $$ where the sum is over tropical curves $(Γ, h)$ with three marked legs in directions determined by $q_{1}, q_{2}, q_{3}$ . This is the structure-constant formula of Gross-Hacking-Keel-Kontsevich ^{[Gross-Siebert 2011]}, and it makes the link between Nishinou-Siebert curve counting and mirror-symmetric Calabi-Yau geometry maximally explicit.

Synthesis of the application. The Nishinou-Siebert correspondence supplies the integer counts that govern the Gross-Siebert reconstruction; equivalently, the wall-crossing scattering diagram on $B$ is built from the tropical-curve enumeration of $B$ . This is the bridge between enumerative algebraic geometry — counting curves on a toric variety — and the construction of mirror Calabi-Yau pairs — building the smooth dual Calabi-Yau from combinatorial-tropical data. The same pattern recurs in [04.12.10] SYZ conjecture, in [04.12.09] Gross-Siebert reconstruction theorem statement, in [04.12.11] slab functions and tropical-manifold structure, and in [04.12.12] theta functions on polarised tropical manifolds. The Nishinou-Siebert correspondence is the foundational structural input to the entire programme.

Connections [Master]

Mikhalkin's correspondence theorem 04.12.05. The Nishinou-Siebert correspondence is the higher-dimensional generalisation of Mikhalkin's 2005 theorem for toric surfaces. The Mikhalkin case is recovered as the dimension- $2$ specialisation: when $Σ$ is a fan in $R^{2}$ , the dual intersection complex $B (Ξ)$ is two-dimensional and the Nishinou-Siebert multiplicity formula at each trivalent vertex reduces to $∣ det (u_{i}, u_{j}) ∣$ — the original Mikhalkin lattice multiplicity. The Nishinou-Siebert lift uses toric degenerations instead of Mikhalkin's complex tropicalisation argument, and the new toolkit extends seamlessly to dimension $n \geq 3$ .
Fan and the toric variety 04.11.04. The Nishinou-Siebert setup is built on the cone-and-fan formalism of [04.11.04]. The toric variety $X_{Σ}$ is constructed from the fan $Σ$ ; the polyhedral subdivision $Ξ$ refines the polytope associated to a line bundle on $X_{Σ}$ ; the dual intersection complex $B (Ξ)$ is the cone complex dual to $Ξ$ . The toric structure of $X_{Σ}$ is what makes the tropical correspondence possible — non-toric targets require the more general Gross-Siebert log GW framework but do not admit the clean Nishinou-Siebert reduction.
Gross-Siebert reconstruction theorem 04.12.09. The mirror-symmetry application appears again in [04.12.09] Gross-Siebert reconstruction theorem statement, where the Nishinou-Siebert tropical curves are exactly the wall-crossing data assembling the smooth mirror Calabi-Yau from the dual intersection complex. The two units pair: [04.12.06] proves the curve-counting correspondence, [04.12.09] consumes it as input to mirror construction.
Strominger-Yau-Zaslow conjecture 04.12.10. The SYZ heuristic of dual special-Lagrangian torus fibrations is the physics-motivated picture behind the Nishinou-Siebert / Gross-Siebert programme. The dual intersection complex $B (Ξ)$ is the algebraic realisation of the SYZ base $B$ , and the Nishinou-Siebert correspondence is the algebraic mechanism by which the SYZ heuristic descends to a precise enumerative-combinatorial theorem on the base.
Log Gromov-Witten invariants (pointer) 04.12.15. The log Gromov-Witten foundations of Abramovich-Chen 2014 and Gross-Siebert 2013 are the technical bridge from algebraic enumeration to tropical enumeration. The Nishinou-Siebert proof uses log GW theory essentially: ordinary GW theory cannot handle the singular central fibre of the toric degeneration. The pointer unit [04.12.15] records the log GW infrastructure that Nishinou-Siebert applies.
Period integral and the mirror map (pointer) 04.12.13. The leading-order mirror-map coefficient $r_{1}$ in the Gross-Siebert reconstructed Calabi-Yau family is a tropical-disk count on $(B, P)$ via the Nishinou-Siebert correspondence applied to the slab function on the relevant codimension-1 cell. The pointer unit [04.12.13] records the period-integral / mirror-map apparatus on the smooth Calabi-Yau fibres; the present unit supplies the tropical-disk-count enumerative content that feeds the higher-order mirror-map coefficients.
Tropical semiring and tropical polynomial 04.12.01. The min-plus arithmetic of the foundational tropical-semiring unit is the algebraic substrate of the Nishinou-Siebert correspondence: the polyhedral subdivision $Ξ$ from which the dual intersection complex is built is itself a regular subdivision of a Newton polytope induced by the lower hull of a tropical polynomial $p \in T [x_{1}, \dots, x_{n}]$ , and the integer-slope structure of $p$ determines the integer multiplicities at vertices of the embedded tropical curves on $B (Ξ)$ . The corner-locus formalism of the foundational unit generalises in this setup to the dual-intersection-complex setting, with vertices of $B (Ξ)$ playing the role of corners.
Tropical curve as balanced rational metric graph 04.12.02. The Nishinou-Siebert tropical curves on the dual intersection complex $B (Ξ)$ are balanced rational metric graphs in the precise sense of the foundational unit, with edge weights given by integer multiplicities and the balancing condition holding at every vertex when read against primitive edge directions in the integral affine structure on $B (Ξ)$ . The Mikhalkin trivalent-vertex multiplicity $∣ det (u_{i}, u_{j}) ∣$ recovered as the dimension- $2$ specialisation of the Nishinou-Siebert vertex-multiplicity formula is exactly the lattice-multiplicity datum carried by the balanced metric graphs of the prerequisite unit. The metric-graph framework supplies the abstract combinatorial object; the dual intersection complex supplies the polyhedral ambient.
Kapranov's theorem 04.12.03. Kapranov's foundational equivalence between algebraic and tropical hypersurfaces is the structural input that makes the Nishinou-Siebert lift well-defined: tropical curves on $B (Ξ)$ are valuation-images of algebraic curves on the generic fibre of the toric degeneration, and the Hensel-lift construction at each tropical vertex inverts the valuation map locally via the residue-tropical structure. Without Kapranov's theorem, the Nishinou-Siebert count would be a combinatorial heuristic; with it, the tropical-to-algebraic lift is an exact equivalence at the level of stable maps.
Newton polytope and non-archimedean amoeba 04.12.04. The polytope subdivision $Ξ$ entering the Nishinou-Siebert construction is a regular subdivision of a polytope of an ample line bundle on the toric variety $X_{Σ}$ — exactly the Newton-polytope subdivision formalism of the prerequisite unit specialised to a global polytope. The non-archimedean amoeba framework of the prerequisite unit provides the dilation-limit geometry that the toric degeneration realises scheme-theoretically: the special fibre of $X \to Spec C {{t}}$ is the limiting amoeba of the generic fibre, and the dual intersection complex $B (Ξ)$ is its tropical skeleton.
Moduli of curves 04.10.01. The Nishinou-Siebert correspondence is implemented in the log Gromov-Witten moduli space $\overline{M}_{g, n}^{l o g} (X, β)$ — a log-enhancement of the moduli-of-curves space from the prerequisite unit. The moduli-of-stable-maps construction (curve plus marked points plus map to target) generalises here to log-smooth families of pointed curves with toric-degeneration target, and the tropical-to-algebraic lift of the present unit is the combinatorial recursion on this enhanced moduli space. The dimension count of the prerequisite unit is the foundational reason Nishinou-Siebert dimension counts come out integer: at each tropical vertex, the contribution to the virtual fundamental class of $\overline{M}_{g, n}^{l o g}$ is a lattice index, matched against the local moduli-dimension of stable maps at that vertex.
Toric degeneration of a Calabi-Yau variety 04.12.07. The downstream Calabi-Yau-degeneration unit is the mirror-symmetry specialisation of the Nishinou-Siebert framework: where [04.12.06] enumerates tropical curves on the dual intersection complex of a toric degeneration of an arbitrary projective toric variety, [04.12.07] specialises to Calabi-Yau hypersurfaces in toric Fanos with reflexive Newton polytopes. The tropical curves enumerated by the Nishinou-Siebert correspondence on $B (P)$ are precisely the wall-crossing data on the dual intersection complex of the Calabi-Yau degeneration that feed the Gross-Siebert reconstruction. Nishinou-Siebert is the toric-prototype; [04.12.07] is the Calabi-Yau specialisation.
Dual intersection complex; tropical manifold $B$ 04.12.08. The dual intersection complex first appears here as the polyhedral home for the Nishinou-Siebert tropical curves; in [04.12.08], $(B, P)$ is upgraded from "combinatorial home for tropical curves" to "tropical manifold with integral affine structure", with the Calabi-Yau monodromy condition added. The Nishinou-Siebert curve counts on $(B, P)$ supply the scattering-diagram input for the Gross-Siebert reconstruction. The present unit is the enumerative-correspondence prerequisite; [04.12.08] is the structural enhancement of the polyhedral object on which the enumeration lives.
Slab function and structure of a tropical manifold 04.12.11. The order- $1$ slab-function coefficients $α_{ρ}$ in $f_{ρ}^{(1)} = 1 + t α_{ρ}$ of [04.12.11] are determined by the genus- $0$ log Gromov-Witten counts of curves on the central fibre contributing to the slab $ρ$ , and the present unit's correspondence theorem expresses these log GW counts as Mikhalkin-style tropical-curve counts in $B (Ξ)$ . The Nishinou-Siebert correspondence is therefore the foundational enumerative input that decorates the slab functions of the structure $S$ with integer multiplicities.
Theta function of a polarised tropical manifold 04.12.12. The Nishinou-Siebert tropical curve counts in the present unit are the enumerative input to the wall functions $f_{d}$ that decorate the scattering walls of the canonical scattering diagram of [04.12.12]. Each tropical curve passing through a singular point of $B$ contributes a wall whose wall function records the curve's Nishinou-Siebert multiplicity; the theta functions therefore inherit, via the broken-line enumeration, the integer structure of the Nishinou-Siebert curve counts established here.

Historical & philosophical context [Master]

The Nishinou-Siebert correspondence sits at the convergence of three twentieth-century mathematical threads. Tropical geometry as an autonomous subject was launched by Mikhalkin in the early 2000s, but its roots reach back to Bergman 1971 J. Algebraic Combinatorics (tropical varieties as logarithmic limits), Sturmfels 1996 Gröbner Bases and Convex Polytopes (algebraic-combinatorial duality), and the older Bieri-Groves 1984 J. Reine Angew. Math. construction of tropical varieties as polyhedral complexes ^{[Bieri-Groves 1984]}. Mikhalkin's 2005 J. AMS 18 paper ^{[Mikhalkin 2005]} crystallised the field by proving the first correspondence theorem — relating complex enumerative invariants on toric surfaces to tropical enumeration — and won him the Clay Research Award in 2008.

The toric degeneration technology underpinning Nishinou-Siebert traces to Mumford 1972 Compositio Mathematica (the first explicit construction of degenerating Abelian varieties) and Kempf-Knudsen-Mumford-Saint-Donat 1973 Toroidal Embeddings I, which formalised the gluing of toric varieties from polyhedral subdivision data. The application of toric degenerations to enumerative geometry as a computational tool was developed in the 1990s and early 2000s by Sturmfels, Gathmann, and others. Tian-Symington 2003 connected toric degenerations to the Strominger-Yau-Zaslow heuristic in the smooth Lagrangian framework, providing a physics-side motivation parallel to the algebraic one.

The third thread is log Gromov-Witten theory itself. Kato's 1989 paper on logarithmic geometry ^{[Kato 1989]} provided the basic framework of log smooth schemes, and Li-Ruan 2001 introduced the first version of relative Gromov-Witten theory adapted to degenerations. The full log GW edifice was constructed in parallel by Abramovich-Chen (Asian J. Math. 2014; Chen Annals 2014) and Gross-Siebert (J. AMS 2013) ^{[Gross-Siebert 2013]}, with the comparison theorem of Abramovich-Chen-Marcus-Wise (2017) showing the two constructions agree. Nishinou-Siebert 2006 ^{[Nishinou-Siebert 2006]} is the precise instance where these three threads — tropical geometry, toric degenerations, log Gromov-Witten — converge into a single enumerative correspondence theorem in arbitrary dimension.

The Gross-Siebert programme that subsumes the Nishinou-Siebert correspondence into mirror symmetry was published as a sequence of papers between 2006 and the present; the foundational Annals 2011 paper ^{[Gross-Siebert 2011]} proved the reconstruction theorem assembling a smoothing of a degenerate Calabi-Yau from real-affine data, and ongoing work by Gross, Hacking, Keel, Kontsevich, and collaborators extends the picture to log Calabi-Yau pairs, cluster algebras, and Fukaya category considerations. Gross's CBMS lecture notes 2011 ^{[Gross 2011 CBMS]} provide the textbook synthesis of the Nishinou-Siebert correspondence in the broader mirror-symmetric context.

Bibliography [Master]

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

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

@article{GrossSiebert2011,
  author = {Gross, Mark and Siebert, Bernd},
  title = {From real affine geometry to complex geometry},
  journal = {Annals of Mathematics},
  volume = {174},
  number = {3},
  year = {2011},
  pages = {1301--1428},
}

@article{GrossSiebert2013,
  author = {Gross, Mark and Siebert, Bernd},
  title = {Logarithmic Gromov-Witten invariants},
  journal = {Journal of the American Mathematical Society},
  volume = {26},
  number = {2},
  year = {2013},
  pages = {451--510},
}

@article{Chen2014,
  author = {Chen, Qile},
  title = {Stable logarithmic maps to {Deligne-Faltings} pairs {I}},
  journal = {Annals of Mathematics},
  volume = {180},
  number = {2},
  year = {2014},
  pages = {455--521},
}

@article{AbramovichChen2014,
  author = {Abramovich, Dan and Chen, Qile},
  title = {Stable logarithmic maps to {Deligne-Faltings} pairs {II}},
  journal = {Asian Journal of Mathematics},
  volume = {18},
  number = {3},
  year = {2014},
  pages = {465--488},
}

@article{StromingerYauZaslow1996,
  author = {Strominger, Andrew and Yau, Shing-Tung and Zaslow, Eric},
  title = {Mirror symmetry is {T}-duality},
  journal = {Nuclear Physics B},
  volume = {479},
  number = {1-2},
  year = {1996},
  pages = {243--259},
}

@book{Gross2011CBMS,
  author = {Gross, Mark},
  title = {Tropical Geometry and Mirror Symmetry},
  series = {CBMS Regional Conference Series in Mathematics},
  volume = {114},
  publisher = {American Mathematical Society},
  year = {2011},
}

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

@book{ItenbergMikhalkinShustin2009,
  author = {Itenberg, Ilia and Mikhalkin, Grigory and Shustin, Eugenii},
  title = {Tropical Algebraic Geometry},
  series = {Oberwolfach Seminars},
  volume = {35},
  edition = {2nd},
  publisher = {Birkhäuser},
  year = {2009},
}

@article{GathmannMarkwig2007,
  author = {Gathmann, Andreas and Markwig, Hannah},
  title = {The {Caporaso-Harris} formula and plane relative {Gromov-Witten} invariants in tropical geometry},
  journal = {Mathematische Annalen},
  volume = {338},
  number = {4},
  year = {2007},
  pages = {845--868},
}

@article{KontsevichSoibelman2006,
  author = {Kontsevich, Maxim and Soibelman, Yan},
  title = {Affine structures and non-{A}rchimedean analytic spaces},
  booktitle = {The Unity of Mathematics},
  series = {Progress in Mathematics},
  volume = {244},
  publisher = {Birkhäuser},
  year = {2006},
  pages = {321--385},
}

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

@article{Kato1989,
  author = {Kato, Kazuya},
  title = {Logarithmic structures of {Fontaine-Illusie}},
  booktitle = {Algebraic Analysis, Geometry, and Number Theory},
  publisher = {Johns Hopkins University Press},
  year = {1989},
  pages = {191--224},
}

Prerequisites

04.11.04
04.12.05

Tier anchors

beginner: Mikhalkin 2005 *Enumerative tropical algebraic geometry in $\mathbb{R}^2$* informal opening; Gathmann 2006 *Tropical algebraic geometry*, Jahresbericht DMV, §1 (informal three-page survey of correspondence)
intermediate: Nishinou-Siebert 2006 *Toric degenerations of toric varieties and tropical curves*, Duke Math. J. 135, §1-§3 + §8; Mikhalkin 2005 *J. AMS* 18, §1-§2
master: Nishinou-Siebert 2006 *Duke Math. J.* 135, §4-§8 (full proof for arbitrary dimension via toric degenerations); Gross-Siebert 2011 *Annals of Math.* 174 *From real affine geometry to complex geometry*; Mikhalkin 2005 *J. AMS* 18 (the $n=2$ predecessor); Abramovich-Chen 2014 *Asian J. Math.* 18 + Chen 2014 *Annals of Math.* 180 + Gross-Siebert 2013 *J. AMS* 26 (log Gromov-Witten foundations); Gross *Tropical Geometry and Mirror Symmetry* (CBMS Reg. Conf. Ser. 114, 2011) for the textbook exposition

References

TODO_REF
Nishinou, T. & Siebert, B. — Toric degenerations of toric varieties and tropical curves · *Duke Mathematical Journal* 135 (1), 1-51 (2006). §1 (introduction, statement of correspondence in arbitrary dimension); §2-§3 (toric degenerations and the dual intersection complex); §4-§7 (deformation-theoretic moves: smoothing log-smooth central-fibre curves); §8 (the correspondence theorem and multiplicity formula).
TODO_REF
Gross, M. & Siebert, B. — From real affine geometry to complex geometry · *Annals of Mathematics* 174 (3), 1301-1428 (2011). The reconstruction theorem assembling a smoothing of a degenerate Calabi-Yau from real-affine + tropical data; the tropical-curve input is exactly the Nishinou-Siebert count.
TODO_REF
Mikhalkin, G. — Enumerative tropical algebraic geometry in $\mathbb{R}^2$ · *Journal of the AMS* 18 (2), 313-377 (2005). The original correspondence theorem in dimension 2 generalised by Nishinou-Siebert. §1 (statement); §3-§5 (proof via complex tropicalisation); §6 (Mikhalkin lattice multiplicity).
TODO_REF
Mikhalkin, G. — Tropical geometry and its applications · *Proceedings of the International Congress of Mathematicians*, Madrid 2006, Vol. II, 827-852. The survey article positioning the correspondence theorem inside tropical geometry as a programme.
TODO_REF
Abramovich, D. & Chen, Q. — Stable logarithmic maps to Deligne-Faltings pairs II · *Asian Journal of Mathematics* 18 (3), 465-488 (2014). Foundations of log Gromov-Witten theory for Deligne-Faltings log structures; one of the two parallel constructions used in Nishinou-Siebert at the central fibre.
TODO_REF
Chen, Q. — Stable logarithmic maps to Deligne-Faltings pairs I · *Annals of Mathematics* 180 (2), 455-521 (2014). The companion paper to Abramovich-Chen II; jointly with Gross-Siebert provides the log Gromov-Witten foundations the correspondence rests on.
TODO_REF
Gross, M. & Siebert, B. — Logarithmic Gromov-Witten invariants · *Journal of the AMS* 26 (2), 451-510 (2013). The alternative log Gromov-Witten foundation (equivalent to Abramovich-Chen by Abramovich-Chen-Marcus-Wise 2017); used directly in the Nishinou-Siebert tropicalisation argument.
TODO_REF
Gross, M. — Tropical Geometry and Mirror Symmetry · CBMS Regional Conference Series in Mathematics 114, American Mathematical Society (2011). Textbook treatment of the Nishinou-Siebert correspondence; Ch. 4 (toric degenerations), Ch. 5 (the dual intersection complex and tropical curves), Ch. 6 (correspondence theorem).
TODO_REF
Strominger, A., Yau, S.-T. & Zaslow, E. — Mirror symmetry is T-duality · *Nuclear Physics B* 479 (1-2), 243-259 (1996). The SYZ heuristic motivating Gross-Siebert; mirror Calabi-Yau pairs arise as dual special-Lagrangian torus fibrations, with the base providing the integral affine manifold that becomes the dual intersection complex in the Gross-Siebert algebraic realisation.
TODO_REF
Kontsevich, M. & Soibelman, Y. — Affine structures and non-Archimedean analytic spaces · in *The Unity of Mathematics*, Progress in Mathematics 244, Birkhäuser, 321-385 (2006). The non-archimedean SYZ proposal complementary to Gross-Siebert; supplies the analytic interpretation of the dual intersection complex as the skeleton of a non-archimedean analytic Calabi-Yau.
TODO_REF
Gathmann, A. & Markwig, H. — The Caporaso-Harris formula and plane relative Gromov-Witten invariants in tropical geometry · *Mathematische Annalen* 338 (4), 845-868 (2007). Tropical proof of the Caporaso-Harris recursion using Mikhalkin's correspondence; the algorithmic counterpart to Nishinou-Siebert in dimension 2.
TODO_REF
Itenberg, I., Mikhalkin, G. & Shustin, E. — Tropical Algebraic Geometry · Oberwolfach Seminars 35, Birkhäuser (2nd ed. 2009). Textbook with Mikhalkin's correspondence in Ch. 4 and the connection to enumerative geometry; precursor exposition to the higher-dimensional Nishinou-Siebert generalisation.
TODO_REF
Maclagan, D. & Sturmfels, B. — Introduction to Tropical Geometry · Graduate Studies in Mathematics 161, American Mathematical Society (2015). Ch. 6 — tropical curves and enumerative correspondence; standard modern reference.

Lean module

Codex.AlgGeom.Tropical.NishinouSiebertCorrespondence

Reviewer

TBD

Estimated time

beginner: 25m
intermediate: 60m
master: 110m