Hyperplane Arrangements in the Grassmannian

This paper investigates the Euler characteristic of the Grassmannian with $d$ hyperplane sections removed, providing combinatorial formulas and computational methods for both generic and special arrangements in complex and real settings, with applications to algebraic complexity in likelihood and scattering equations.

The Gist

Imagine you are standing in a vast, multi-dimensional garden. This garden isn't made of flowers, but of shapes and spaces. Specifically, it's a place called the Grassmannian.

To understand this paper, let's break it down into a story about gardening, puzzles, and counting rooms.

1. The Garden (The Grassmannian)

Think of the Grassmannian as a giant, complex playground where every point represents a specific "flat" shape (like a line, a plane, or a sheet) floating inside a bigger space.

• If you have a 3D room, a "line" floating in it is one type of shape. A "flat sheet" is another.
• The Grassmannian is the map of all possible lines or sheets you could draw in that room.

2. The Walls (Hyperplane Arrangements)

Now, imagine someone starts building walls inside this garden.

• In math, these walls are called hyperplanes.
• When you build a wall, you cut the garden in half.
• If you build many walls, you chop the garden into many small, separate rooms (or regions).

The authors of this paper are asking a very specific question: "If we build a certain number of these walls, how many separate rooms do we end up with?"

But there's a twist:

• The Complex Garden (Math World): In the "complex" version of this garden (which is like a magical, multi-layered reality), the number of rooms is actually a measure of how hard it is to solve a specific puzzle called the "Likelihood Equation." This equation is used by statisticians to find the best guess for data and by physicists to calculate how particles crash into each other.
• The Real Garden (Our World): In the "real" version (the one we can actually see), the number of rooms tells us how many distinct "sign patterns" exist. It's like asking: "If I stand in a room, which side of every wall am I on?"

3. The Two Types of Walls

The paper studies two different ways of building these walls:

• Random Walls (Generic Arrangements): Imagine throwing darts at a board and building walls wherever the darts land. These walls are messy and random. The authors found a combinatorial formula (a fancy counting recipe) to predict exactly how many rooms you get, no matter how many walls you throw.
• Special Walls (Schubert Arrangements): These are walls built according to a strict, elegant rule. They are like walls that only appear in specific, symmetrical patterns. These are crucial for particle physics (specifically, calculating how particles scatter). However, these walls are tricky because they sometimes have "kinks" or "singularities" (like a wall that folds in on itself), making them harder to count.

4. The "Euler Characteristic" (The Magic Number)

The paper focuses on a number called the Euler characteristic.

• Analogy: Think of this number as the "net complexity" of the garden.
• If you have a simple sphere, the number is 2. If you have a donut, it's 0.
• In this paper, this number acts as a score.
  • In the Complex Garden, the score tells you the Maximum Likelihood Degree. In plain English: How many different solutions does the "best guess" puzzle have? If the score is 10, there are 10 possible answers to the puzzle.
  • In the Real Garden, the score helps count the number of distinct rooms.

5. How They Solved It

The authors didn't just guess; they built a toolkit:

1. The Recipe Book (Symbolic Math): They used a powerful mathematical tool called the "Chow Ring" (think of it as a library of rules for how shapes intersect) to write down a formula. This formula lets you plug in the number of walls ($d$) and instantly get the number of rooms.
2. The Computer Simulation (Numerical Math): For the tricky "Special Walls" (Schubert), the formulas get messy. So, they wrote computer code (using a language called Julia) to simulate the walls, find the "critical points" (the peaks and valleys of the landscape), and count the rooms directly.
3. The Real-World Test: For the real garden, they used a technique called Morse Theory. Imagine walking up and down a hilly landscape. By tracking where you can go up and where you get stuck, they mapped out every single room and checked if the rooms were "connected" (contractible) or weirdly shaped (like a donut).

6. Why Does This Matter?

• For Statisticians: It tells them how complicated their data models are. If the number of solutions is huge, the model is very complex and hard to solve.
• For Physicists: It helps them calculate the probability of particles interacting. The "scattering equations" they solve are the same as the "likelihood equations" in statistics. Understanding the geometry of these equations helps predict the behavior of the universe at the smallest scales.
• For Mathematicians: It connects two seemingly different worlds: the smooth, perfect world of complex numbers and the jagged, varied world of real numbers. They showed that while the rules are similar, the real world can be surprisingly weird (some rooms aren't even simple shapes!).

Summary

This paper is a guidebook for counting rooms in a multi-dimensional garden.

• It gives you a formula to count rooms when the walls are random.
• It provides computer code to count rooms when the walls follow special, physics-inspired rules.
• It reveals that in the "real" version of this garden, the rooms can be surprisingly strange, sometimes looping back on themselves like a Möbius strip, unlike the simple rooms in the "complex" version.

Ultimately, by counting these rooms, the authors are helping scientists solve the hardest puzzles in data science and particle physics.

Technical Summary

1. Problem Statement

The paper investigates the topological Euler characteristic ($\chi$) of a smooth very affine variety obtained by removing an arrangement of $d$ hyperplanes from the Grassmannian $\text{Gr}_K(k,n)$, where $K$ is either the complex field $\mathbb{C}$ or the real field $\mathbb{R}$.

The specific varieties of interest are:
$$X = \text{Gr}_K(k,n) \setminus \mathcal{H}$$
where $\mathcal{H} = \{H_1, \dots, H_d\}$ is a collection of hyperplanes in the ambient projective space $\mathbb{P}^{\binom{n}{k}-1}$ (via the Plücker embedding).

Motivation:

• Algebraic Statistics: The Euler characteristic of such a variety equals the Maximum Likelihood (ML) degree of the corresponding statistical model. This degree represents the algebraic complexity of solving likelihood equations (finding critical points of the log-likelihood function).
• Theoretical Physics: The problem is linked to scattering equations in particle physics, specifically the Cachazo–He–Yuan (CHY) formalism and the $(k,n)$ CEGM formalism. The moduli space of scattering amplitudes is related to the configuration space $X(k,n) = \text{Gr}_\mathbb{C}(k,n)/(\mathbb{C}^*)^n$. Understanding the topology of the Grassmannian itself (before quotienting) provides insights into these amplitudes and positive geometries (e.g., the amplituhedron).

The authors focus on two types of hyperplane arrangements:

1. Generic Arrangements: Hyperplanes in general position.
2. Schubert Arrangements: Hyperplanes defined by the vanishing of specific Plücker coordinates (Schubert divisors), which are physically significant but often singular.

2. Methodology

The authors employ a combination of algebraic geometry, combinatorics, and numerical topology.

A. Combinatorial Framework (Complex Case)

• Intersection Poset: They define the intersection poset $L(\mathcal{H})$ of the arrangement within the Grassmannian, ordered by reverse inclusion ($x \le y \iff y \subseteq x$).
• Möbius Inversion: Using the Möbius function $\mu$ of the poset, they derive a general formula for the Euler characteristic of the complement:
  $$\chi(\text{Gr}_K(k,n) \setminus \mathcal{H}) = \sum_{y \in L(\mathcal{H})} \chi(y) \mu(y)$$
• Genericity: For generic arrangements, $L(\mathcal{H})$ is a truncated boolean algebra. This simplifies the formula to a polynomial in $d$ involving the Euler characteristics of intersections of $i$ generic hyperplanes.

B. Computational Techniques (Complex Case)

• Chern-Schwartz-MacPherson (CSM) Classes: For generic hyperplanes, the Euler characteristic of intersections is computed using CSM classes. The authors utilize the transformation relating the generating function of Euler characteristics of hyperplane sections to the CSM class of the Grassmannian.
• Symbolic Computation: They use Macaulay2 (specifically the CharacteristicClasses package) to compute Chern classes of the Grassmannian and derive the polynomials $P_{k,n}(d)$.
• Numerical Homotopy Continuation: For Schubert arrangements (which may be singular), CSM classes are difficult to compute. Instead, the authors use HomotopyContinuation.jl (Julia) to numerically count the number of non-singular critical points of the scattering potential (likelihood equations), which corresponds to the ML degree.
• Recursive Decomposition: To handle singular Schubert intersections theoretically, they develop recursive relations based on partitioning the Grassmannian relative to subspaces defined by the divisors (e.g., Proposition 4.7).

C. Numerical Topology (Real Case)

• Morse Theory: Unlike the complex case, the real Euler characteristic counts the number of connected regions (which are not necessarily contractible). The authors adapt an algorithm from [10] based on Morse theory.
• Algorithm 1:
  1. Parametrize the complement of one Schubert divisor as $\mathbb{R}^{k(n-k)}$.
  2. Define a rational Morse function $r$ vanishing on the hypersurfaces.
  3. Compute critical points of $-\log|r|$ and their indices (via Hessian eigenvalues).
  4. Track gradient ascent paths to determine connectivity between critical points, thereby identifying distinct regions and their sign patterns.

3. Key Contributions and Results

A. General Formulas

The paper establishes a unified formula for the Euler characteristic of the complement of a hyperplane arrangement in the Grassmannian:
$$\chi(\text{Gr}_K(k,n) \setminus \mathcal{H}) = \sum_{y \in L(\mathcal{H})} \chi(y) \mu(y)$$
For generic arrangements over $\mathbb{C}$, this yields a polynomial $P_{k,n}(d)$ of degree $k(n-k)$:
$$P_{k,n}(d) = \sum_{i=0}^{k(n-k)} (-1)^i \chi_{k,n}(i) \binom{d}{i}$$
where $\chi_{k,n}(i)$ is the Euler characteristic of the intersection of $i$ generic hyperplanes.

B. Specific Computations

The authors provide explicit polynomials and numerical values for specific cases:

• $\text{Gr}_\mathbb{C}(2,4)$:
  • Generic case: $P_{2,4}(d) = \frac{1}{12}(72 - 86d + 47d^2 - 10d^3 + d^4)$.
  • Schubert case: $P^S_{2,4}(d) = \frac{1}{12}(72 - 98d + 47d^2 - 10d^3 + d^4)$.
  • The difference highlights the impact of singularity in Schubert arrangements.
• $\text{Gr}_\mathbb{C}(2,5)$: Explicit polynomials are derived for both generic and Schubert cases, showing distinct ML degrees.
• Special Arrangements: They analyze specific non-generic configurations relevant to physics (e.g., lines forming a cycle in $\mathbb{P}^3$ and the six coordinate hyperplanes), computing their specific Euler characteristics.

C. Recursive Relations for Schubert Divisors

The paper provides a method to compute $\chi^S_{k,n}(d)$ (Euler characteristic of $d$ generic Schubert divisors) without full CSM computation:

• Base Cases: $\chi^S_{k,n}(0) = \binom{n}{k}$ and $\chi^S_{k,n}(1) = \binom{n}{k} - 1$.
• Recursion: For $n > dk$, $\chi^S_{k,n}(d) = \binom{n-1}{k-1} + \chi^S_{k,n-1}(d)$.
• Stabilization: For large $d$ ($d \ge \binom{n}{k} - k(n-k)$), the intersection becomes smooth, and $\chi^S_{k,n}(d) = \chi_{k,n}(d)$.

D. Real Case Findings

The study of $\text{Gr}_\mathbb{R}(k,n) \setminus \mathcal{H}$ reveals significant differences from hyperplane arrangements in $\mathbb{R}^N$:

1. Non-Contractibility: Regions are not always contractible. In Example 5.1, regions have the homology type of a circle ($\chi=0$) or even $\chi=-2$.
2. Sign Patterns vs. Regions: The number of regions does not necessarily equal the number of realizable sign patterns; a single sign pattern can correspond to multiple disconnected regions.
3. Algorithmic Success: The Morse-theoretic algorithm successfully computes the number of regions and their topological invariants for $\text{Gr}_\mathbb{R}(2,4)$ with random Schubert arrangements.

4. Significance

• Bridging Fields: The work solidifies the connection between algebraic statistics (ML degrees), algebraic geometry (Grassmannian topology), and theoretical physics (scattering amplitudes).
• Computational Tools: It provides practical, implementable methods (Julia/Macaulay2 code) to compute ML degrees for complex Grassmannian models, which were previously difficult to calculate due to singularities in Schubert arrangements.
• Topological Insight: The real case results challenge standard intuitions about hyperplane arrangements, showing that the topology of the Grassmannian induces complex behaviors (non-contractible regions) not seen in affine space.
• Physics Applications: The results directly apply to the $(k,n)$ CEGM formalism and the study of the amplituhedron, offering a way to compute scattering amplitudes for generalized bi-adjoint $\phi^3$ theories and potentially non-perturbative geometries for $\mathcal{N}=4$ Super Yang-Mills.

5. Conclusion

The paper successfully generalizes Zaslavsky's theory of hyperplane arrangements to the Grassmannian setting. It provides a combinatorial formula for the Euler characteristic, develops recursive and numerical tools to handle singular Schubert arrangements, and uncovers novel topological phenomena in the real case. The availability of code and explicit polynomials makes these results immediately applicable to researchers in algebraic statistics and high-energy physics.