Lorentzian polynomials and the incidence geometry of tropical linear spaces

This paper introduces the concept of Lorentzian proper position to characterize elementary quotients of M-convex functions and analyze the incidence geometry of tropical linear spaces, revealing that while certain classical incidence properties fail for general tropical spaces, they hold for those possessing adjoints.

The Gist

Imagine you are trying to understand the shape of a complex, multi-dimensional object. In the world of classical mathematics, we have a very strict set of rules (geometry) that tell us how lines intersect, how planes meet, and how shapes fit together. For centuries, mathematicians have trusted these rules to be universal.

But then, a new world appeared: Tropical Geometry.

Think of Tropical Geometry not as a new planet, but as a "pixelated" or "simplified" version of our familiar world. In this world, addition becomes taking the minimum, and multiplication becomes addition. It's like looking at a smooth curve through a very low-resolution screen; the curves become sharp, angular lines, and the smooth surfaces become jagged polyhedra.

This paper, by Jidong Wang, is a detective story. The author asks: "Do the old rules of geometry still work in this pixelated, tropical world?"

Here is the breakdown of the investigation using simple analogies:

1. The New Tool: "Lorentzian Proper Position"

To solve the mystery, the author invents a new measuring tool called Lorentzian Proper Position.

• The Analogy: Imagine you have two stacks of blocks. In the old world, you could easily tell if one stack was "stable" enough to support the other. In this new tropical world, the rules are different. The author creates a special "compatibility test" (Lorentzian Proper Position) to see if two tropical shapes can coexist without collapsing.
• The Discovery: He finds that this test is incredibly powerful. It acts like a bridge connecting two different fields of math: the study of polynomials (equations with variables) and matroids (abstract structures that generalize the idea of "independence," like which vectors in space don't overlap).

2. The First Big Question: Do Lines Always Cross? (The "Intersection" Problem)

In normal geometry, if you draw two lines on a flat piece of paper, they almost always cross at a single point. If you have a 3D space, two planes usually intersect in a line.

• The Tropical Twist: The author asks: "If I have a tropical plane and I draw two tropical lines on it, do they always cross?"
• The Result:
  • Yes, for simple cases: If the space is 2-dimensional (like a flat sheet), the answer is YES. Any two lines on a tropical plane will eventually meet. This is a comforting result that feels like the old rules.
  • No, for complex cases: But as soon as the space gets bigger (4 dimensions or more), the rules break! The author constructs specific examples where two lines on a tropical plane never touch, even though they are "supposed" to.
  • The Metaphor: Imagine two cars driving on a 2D road map; they will eventually cross. But in a 4D hyper-road, they might drive past each other in a dimension you can't see, never colliding. The "logic" of intersection fails in higher dimensions.

3. The Second Big Question: Can We Connect Any Dots? (The "Interpolation" Problem)

In normal geometry, if you have a few points, you can usually draw a line or a plane that passes through all of them.

• The Tropical Twist: "If I pick $d-1$ random points on a tropical shape, can I always find a tropical line that passes through all of them?"
• The Result: It depends on the shape!
  • If the shape has a special "adjoint" property (a fancy way of saying it has a hidden symmetry or a "mirror image" structure), then YES, you can connect the dots.
  • If the shape is "weird" (lacking this symmetry), then NO. You might pick three points that simply cannot be connected by a single tropical line.
  • The Metaphor: Think of it like a game of Connect-the-Dots. On a normal piece of paper, you can always connect the dots. But on a crumpled, pixelated piece of paper (the tropical world), sometimes the dots are arranged in a way that no single straight line can touch them all.

4. The Third Big Question: Can We Build a Ladder? (The "Flag" Problem)

Imagine you have a large room (a high-dimensional space) and a tiny point inside it. Can you build a staircase of rooms, getting smaller and smaller, until you reach that point?

• The Tropical Twist: "Can we always find a chain of nested tropical shapes, shrinking down from a big space to a single point?"
• The Result: Yes! The author proves that you can always build this "staircase" (or flag) in the tropical world. This is a rare victory for the old rules; even in this pixelated world, you can always step down from a big space to a small one without getting stuck.

5. The "Why" and the "So What?"

Why does this matter?

• The "Why": The author uses the new "Lorentzian" tool to prove these geometric facts. It's like using a high-tech X-ray to see the skeleton of the tropical world. He shows that the geometry of these shapes is deeply linked to the stability of certain mathematical equations (polynomials).
• The "So What":
  1. It breaks our intuition: We thought geometry was universal. This paper shows that in the "tropical" limit (which is actually very important for computer science, optimization, and economics), the rules change. Things that must happen in the real world (like lines crossing) don't have to happen here.
  2. It creates new tools: By understanding these broken rules, mathematicians can build better algorithms for solving complex problems. The "Lorentzian" tool is a new way to check if a solution is stable.
  3. It connects the dots: It links the abstract world of "matroids" (which are like the DNA of shapes) to the concrete world of "polynomials" (equations).

Summary

This paper is a journey into a "pixelated" version of geometry. The author discovers that while some familiar rules (like building a staircase) still hold, others (like lines always crossing or dots always connecting) break down in higher dimensions. He uses a new mathematical "ruler" (Lorentzian polynomials) to measure these shapes, proving that the tropical world is a fascinating, slightly chaotic place where the laws of geometry are rewritten, but still follow a deep, hidden logic.

Technical Summary

Here is a detailed technical summary of the paper "Lorentzian polynomials and the incidence geometry of tropical linear spaces" by Jidong Wang.

1. Problem Statement

The paper addresses the structural properties of tropical linear spaces (geometric objects arising from valuated matroids) and their relationship to Lorentzian polynomials (a class of polynomials generalizing stable polynomials and associated with M-convex functions).

Specifically, the author investigates the incidence geometry of tropical linear spaces, asking whether classical geometric properties regarding the intersection of subspaces, the interpolation of points, and the completion of flags hold in the tropical setting. The paper seeks to:

1. Characterize elementary quotients of M-convex functions and valuated matroids using Lorentzian polynomials.
2. Determine the structure of the moduli space of codimension-1 tropical linear subspaces (the relative Dressian, $Dr_1(\mu)$).
3. Test classical incidence axioms (intersection of subspaces, point interpolation, flag completion) in the tropical context and identify where they fail or require new conditions (like the existence of adjoints).

2. Methodology

The author employs a dual approach, leveraging the correspondence between discrete combinatorial structures and continuous analytic objects:

• Lorentzian Proper Position: The paper introduces a new notion called Lorentzian proper position ($f \ll_L g$), analogous to the "proper position" of stable polynomials. It is defined such that $g + w_{n+1}f$ is Lorentzian. This concept is used to characterize elementary quotients of M-convex functions.
• Tropicalization: The author utilizes the tropicalization map, which transforms Lorentzian polynomials over the field of Puiseux series into M-convex functions (and valuated matroids). This allows the transfer of convexity properties from the space of polynomials to the space of tropical linear spaces.
• Combinatorial Geometry: The paper analyzes the lattice of elementary quotients ($\widehat{Qt}_1(M)$) and linear subclasses of matroids. It establishes that the relative Dressian $Dr_1(M)$ is isomorphic to the order complex of this lattice.
• Geometric Proofs: For incidence problems, the author uses techniques involving stable intersections, truncations, and contractions of valuated matroids. The proof of intersection properties for tropical lines relies on sweeping hyperplanes and analyzing the dynamics of initial matroids.

3. Key Contributions and Results

A. Lorentzian Characterization of Quotients

• Theorem A: Establishes a bijection between the relation of elementary quotients of valuated matroids ($\mu \twoheadrightarrow \theta$) and the Lorentzian proper position of their associated basis generating polynomials ($f^\theta_q \ll_L f^\mu_q$).
• Convexity of Moduli Spaces: Using the convexity of the set of Lorentzian polynomials in proper position, the paper proves that the set of codimension-1 tropical linear subspaces of a given tropical linear space, $Dr_1(\mu)$, is tropically convex (Theorem A and Corollary 1.4). This recovers and strengthens previous results by Fink and Moci.
• Asymmetry of Convexity: The paper proves that while the set $\{g \mid f \ll_L g\}$ is a closed convex cone, the set $\{h \mid h \ll_L f\}$ is not necessarily convex (Theorem B and Example 3.21). This asymmetry implies that the "inversion" operation (analogous to matroid duality) does not preserve the Lorentzian property.

B. Structural Results on Relative Dressians

• Defining Equations: Theorem 4.3 shows that $Dr_1(\mu)$ is cut out by three-term incidence relations (degenerate, parallel, and concurrent hyperplanes).
• Order Complex Structure: Theorem 4.7 proves that for a matroid $M$, the space $Dr_1(M)$ is the order complex of the elementary quotient lattice $\widehat{Qt}_1(M)$.
• Tropical Polytopes: Theorem 4.10 demonstrates that $Dr_1(M)$ is a tropical polytope generated by the join-irreducible elements of the quotient lattice.

C. Adjoints and Incidence Geometry

The paper introduces adjoints for valuated matroids, generalizing the classical notion for matroids.

• Theorem D: If a valuated matroid $\mu$ has an adjoint, then any set of $d-1$ points in $\text{Trop}(\mu)$ is contained in a codimension-1 tropical linear subspace (a tropical analog of the classical property that $d-1$ points lie on a hyperplane).
• Failure of Incidence Axioms:
  • Intersection (P1): For tropical planes ($d=3$), any two tropical lines intersect (Theorem C). However, for $d \ge 4$, this fails. The poset of matroids on $[n]$ ordered by quotient is not upper semimodular for $n \ge 8$ (Theorem 6.16, Example 6.14).
  • Flag Completion (P3): While partial flags can often be completed, the paper shows (Theorem G) that there exist flags of tropical linear spaces that cannot be completed with a prescribed underlying matroid flag. Specifically, a generic tropical line with a uniform matroid structure cannot be embedded in a tropical plane with a specific projective plane matroid structure.

D. Applications to Lorentzian Polynomials

• Theorem B: The convex cone spanned by the basis generating polynomials of elementary quotients of a matroid is contained within the set of polynomials in Lorentzian proper position with respect to the matroid's polynomial. This connects the combinatorial structure of matroid quotients to the analytic geometry of Lorentzian polynomials.

4. Significance

• Unification of Fields: The paper successfully bridges the theory of Lorentzian polynomials (a hot topic in combinatorial optimization and algebraic geometry) with tropical geometry and matroid theory. It provides a new algebraic tool (Lorentzian proper position) to study tropical incidence geometry.
• Counterexamples to Classical Intuition: It rigorously demonstrates that tropical geometry does not simply inherit all classical linear incidence properties. The failure of upper semimodularity in the poset of matroid quotients for $n \ge 8$ is a significant structural discovery.
• New Definitions: The introduction of adjoints for valuated matroids and the relative Dressian provides a framework for studying subspaces within tropical linear spaces, analogous to how the Dressian parametrizes all tropical linear spaces.
• Methodological Innovation: The use of the "Lorentzian proper position" to prove tropical convexity results offers a powerful new technique for analyzing moduli spaces in tropical geometry.

In summary, Wang's work provides a deep structural analysis of the space of tropical linear subspaces, revealing both the power of Lorentzian polynomials in characterizing these spaces and the fundamental limitations of tropical incidence geometry compared to its classical counterpart.