Newton Polygons & Tropical Geometry

• Newton polygons are convex hulls of exponent vectors that encode key properties of polynomials and serve as a bridge to tropical geometrical structures.
• Tropical geometry employs piecewise-linear tropicalizations of polynomials, using Newton subdivisions to analyze solution counts, singularities, and intersection theory.
• These concepts have practical applications in enumerative invariants, resolving singularities, and understanding fiber degeneracies in algebraic systems.

A Newton polygon (or Newton polytope in higher dimensions) is the convex hull in the relevant lattice of the exponent vectors of nonzero monomials in a polynomial or Laurent polynomial. Tropical geometry is a field in which polynomials over valued fields (typically non-Archimedean) are studied via their Newton polygons and associated piecewise-linear structures arising from tropicalization. Newton polygons underpin the passage from algebraic to combinatorial-geometry structures, encoding solution counts, fiber degeneracies, resolution data, and moduli in both classical algebraic and tropical settings.

1. Newton Polygons and Their Properties

Given a polynomial in one or several variables over a field (possibly with a non-Archimedean valuation), the Newton polygon (or more generally, Newton polytope) is defined as the convex hull of its exponent support: $$\mathrm{Newt}(f) = \mathrm{Conv}\bigl\{ a \in \mathbb{Z}^n \mid c_a \neq 0 \bigr\} \subset \mathbb{R}^n$$ for $f(x) = \sum_{a \in A} c_a x^a$ with $x^a = x_1^{a_1} \cdots x_n^{a_n}$ (Kaveh et al., 2018, Brugallé et al., 2015, Kazarnovskii et al., 2017).

In the univariate case, if $P(x) = \sum_{i=0}^d a_i x^i$, its Newton polygon in $\mathbb{R}^2$ is the lower convex hull of the points $(i, v(a_i))$, where $v$ is a valuation (possibly $v \equiv 0$ for degree in the absence of valuations) (Gunn, 2019, Garoufalidis, 2011).

Each compact face (or edge, in dimension 2) specifies a reduced polynomial (face polynomial) obtained by restricting to exponents in that face; the combinatorics and geometry of these faces control singularity, reducibility, and degeneration structure (Takahashi, 2016, Manon et al., 2018).

2. Newton Polygon Subdivisions and Duality

Given valuations on polynomial coefficients, one can "lift" each exponent point $(a_1,\dots,a_n)$ to $(a_1,\dots,a_n,v(c_a))$ in $f(x) = \sum_{a \in A} c_a x^a$0. The lower convex hull of these lifted points projects to a polyhedral subdivision of the Newton polytope, called the Newton subdivision (Brugallé et al., 2015, Tewari, 2020).

There exists a combinatorial duality between the Newton subdivision and the tropical hypersurface (the corner locus of the tropicalization of $f(x) = \sum_{a \in A} c_a x^a$1). Specifically:

• $f(x) = \sum_{a \in A} c_a x^a$2-cells in the Newton subdivision correspond dually to $f(x) = \sum_{a \in A} c_a x^a$3-cells in the tropical variety.
• Edges in the subdivision are dual to facets (maximal cells) of the tropical hypersurface.
• The lattice length or volume of a dual cell gives the balancing weights on the tropical complex (Brugallé et al., 2015, Kaveh et al., 2018, Tewari, 2020).

This duality is foundational for expressing algebraic and combinatorial invariants in a unified fashion.

3. Tropicalization and Tropical Hypersurfaces

The tropicalization of a polynomial $f(x) = \sum_{a \in A} c_a x^a$4 over a valued field is a piecewise-linear function: $f(x) = \sum_{a \in A} c_a x^a$5 The tropical (hyper)surface is the non-differentiability locus, i.e., the set where at least two terms attain the maximum (Kaveh et al., 2018, Brugallé et al., 2015): $f(x) = \sum_{a \in A} c_a x^a$6 This is a rational polyhedral complex of pure codimension 1, and its geometry is governed by the normal fan of the Newton polytope.

In the context of systems of equations, the intersection of multiple tropical hypersurfaces—calculated as stable intersection of their respective tropical varieties—encodes the combinatorial data of the intersection points, with multiplicities given by mixed volumes of the dual faces (Rabinoff, 2010, Brugallé et al., 2015).

4. Enumerative and Topological Invariants from Newton Polyhedra

The Newton polytope encodes numerous invariants of the corresponding algebraic or tropical variety.

• Degree: The tropical degree of a hypersurface equals the lattice perimeter (or, in higher dimensions, the sum of appropriate facet measures) of the Newton polytope (Brugallé et al., 2015).
• Genus (in plane curves): The genus bound is given by the count of interior lattice points of the Newton polygon. For nonsingular tropical curves, $f(x) = \sum_{a \in A} c_a x^a$7 (Brugallé et al., 2015, Coles et al., 2020).
• Milnor number: For a convenient, Newton nondegenerate plane curve singularity, Kouchnirenko's formula computes the Milnor number using area and boundary lattice data of the Newton diagram's complement, with a tropical-geometric proof via duality to the corresponding tropical curve (Takahashi, 2016).
• Combinatorics of moduli: The moduli space dimension of tropical curves with fixed Newton polygon matches that of nondegenerate algebraic curves with the same polygon (under genericity and maximality), with explicit combinatorial formulae in terms of the subdivision and boundary point types (Coles et al., 2020).

5. Newton Polygons in Fiber Degeneracy and Non-Properness

Newton polygons also control finer geometric phenomena, such as non-properness loci and fiber degeneracies.

• For a polynomial map $f(x) = \sum_{a \in A} c_a x^a$8 over a Puiseux series field, the set $f(x) = \sum_{a \in A} c_a x^a$9 where $x^a = x_1^{a_1} \cdots x_n^{a_n}$0 fails to be finite ("Jelonek set") is a hypersurface. Its coordinatewise valuation, the "tropical non-properness set," is computable as the locus where virtual fibers of the tropicalization $x^a = x_1^{a_1} \cdots x_n^{a_n}$1 become degenerate (contain dicritical half-lines) (Hilany, 2022).
• The polyhedral method for determining this set relies on Newton polytopes of the coordinates, face-resultant sets supported on dicritical faces, and the combinatorics of Minkowski sums of these polytopes. The normal fan of the Minkowski sum controls the combinatorics of the tropical non-properness set (Hilany, 2022).
• This construction links the algebraic data of multivariate resultants and the combinatorics of polyhedral geometry in the tropical context.

6. Newton Polytopes and Intersection Theory

Newton polytopes underlie tropical and classical intersection theory in several ways:

• The volume and mixed volumes of Newton polytopes control solution counts for systems of polynomial equations via the Bernstein–Kushnirenko–Khovanskii (BKK) theorem: the number of isolated solutions in $x^a = x_1^{a_1} \cdots x_n^{a_n}$2 to a generic system $x^a = x_1^{a_1} \cdots x_n^{a_n}$3 is $x^a = x_1^{a_1} \cdots x_n^{a_n}$4 (Kaveh et al., 2018, Kazarnovskii et al., 2017).
• The tropical solution count arises from stable intersection of tropical hypersurfaces, mirroring the algebraic count and recovering multiplicities from mixed cells of the dual subdivisions (Rabinoff, 2010, Kazarnovskii et al., 2017).
• The ring of conditions for $x^a = x_1^{a_1} \cdots x_n^{a_n}$5 (the "intersection theory") is modeled both by balanced tropical fans (associated to Newton fans) and by the polytope algebra generated under Minkowski sum by Newton polytopes, with perfect duality via volume and mixed-volume pairings (Kazarnovskii et al., 2017).

7. Broader Structures and Applications

Newton polygons and their tropical avatars appear throughout algebraic, combinatorial, and computational geometry, including:

• Resolution of singularities: In the analysis of plane curve singularities, Newton polygons recursively encode the steps in toric (and tropical) resolution, organizing the data in complexes such as the "lotus," which encompasses all possible Newton fans encountered in the process (Barroso et al., 2019).
• Hyperfield frameworks: Newton polygons, together with tropical and sign hyperfields, unify Descartes' rule of signs and the Newton polygon rule for zeros/multiplicities of polynomials over formally-real or valued fields (Gunn, 2019).
• Tropical differential algebraic geometry: The supports and vertex sets of Newton polygons encode solvability and constraint information in differential and partial differential equations, allowing for tropical tests of solution supports via combinatorial properties of the polygons (Falkensteiner et al., 2020).
• Liouvillian exceptional points: The order and anisotropy of non-Hermitian degeneracies (exceptional points) in Lindblad systems are encoded by Newton polygons of the characteristic polynomial in two variables (spectral parameter, perturbation), with edge slopes dictating Puiseux expansion exponents and the tropical corner locus tracking EP scaling orders (P et al., 9 Oct 2025).

Newton polygons thus serve as a central combinatorial and polyhedral tool, mediating between algebraic data (e.g., coefficients, singularities, recurrences) and tropical or piecewise-linear geometries that encode degenerations, counting, intersection, and fiber structure across a spectrum of contexts (Brugallé et al., 2015, Kaveh et al., 2018, Hilany, 2022, Takahashi, 2016, Coles et al., 2020, Kazarnovskii et al., 2017, Tewari, 2020, Barroso et al., 2019, Gunn, 2019, P et al., 9 Oct 2025).